DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 09/29/2023 is being considered by the examiner.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 6-12 and 19 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claim 6 recites “evaluating whether the reference classified vibration level is comparable to the classified vibration level.” Claims 7-10 depend, directly or indirectly, from claim 6 and therefore inherit this limitation. Claim 8 further recites identifying a track anomaly “in response to the reference classified vibration level being comparable to the classified vibration level.” Claim 11 recites evaluating whether a reference classified vibration level from a second vibration sensor is “comparable to” the classified vibration level. Claim 12 recites determining that the anomaly is a malfunctioning suspension in response to determining that the reference classified vibration level is “not comparable” to the classified vibration level. Claim 19 recites corresponding method limitations.
Applicant argues that “comparable” is definite because the specification states that the reference classified vibration level is compared with the classified vibration level for “common or comparable labels/features” and because comparable levels indicate a common cause while non-comparable levels indicate a localized cause.
Applicant’s argument is not persuasive.
The specification language cited by Applicant does not define an objective boundary for “comparable.” The phrase “common or comparable labels/features” is circular because it uses “comparable” to explain “comparable.” It does not identify how much overlap is required, which feature or label must match, whether the comparison is based on amplitude, frequency, time sequence, label identity, number of labels, confidence score, correlation value, pattern similarity, threshold distance, same-location occurrence, or another criterion.
The claims also do not specify the required test for comparability. For example, the claims do not state whether two classified vibration levels are “comparable” when they have one common label, all common labels, matching label order, similar amplitudes, similar dominant frequencies, matching temporal patterns, matching severity levels, or similarity above a predetermined numerical threshold. Different comparison rules could produce different results for the same vibration data.
This uncertainty is material because the claims use “comparable” and “not comparable” as decision points for identifying different anomaly types. Under claim 8, “comparable” levels cause the processor to identify a track anomaly. Under claim 12, “not comparable” levels cause the processor to identify a malfunctioning suspension. Because the claims do not define the boundary between those two outcomes, the metes and bounds of the claimed subject matter are unclear.
Applicant’s statement that one of ordinary skill would understand “comparable” to mean “sufficiently similar in relevant characteristics to indicate a common cause” does not resolve the issue. The claims do not recite “sufficiently similar,” do not define which characteristics are relevant, and do not state what degree of similarity is sufficient. The functional consequence of the decision does not define the objective scope of the decision criterion itself.
Accordingly, the rejection of claims 6-12 and 19 under 35 U.S.C. §112(b) is maintained.
Applicant may overcome this rejection by amending the claims to recite an objective comparison criterion. Examples include reciting that the classified vibration level and reference classified vibration level have matching classification labels, matching severity labels, matching predetermined pattern labels, a similarity score exceeding a predetermined similarity threshold, a correlation value exceeding a predetermined correlation threshold, a difference between amplitudes or frequencies below a predetermined tolerance, or another definite comparison standard supported by the originally filed disclosure.
REFERENCES USED
Reference 1: US Patent No. 9,365,223 B2, “System and Method for Monitoring Railcar Performance.”
Reference 2: US Patent No. 9,395,276 B2, “Method and System for Detection and Analysis of Railway Bogie Operational Problems.”
Reference 3: EP 2 940 440 A1, “Identification of the Presence of a Potentially Damaging Resonant Vibration State of a Mechanical Device.”
Reference 4: US Patent Application Publication No. 2021/0287459 A1, “Digital Twin Systems and Methods for Transportation Systems.”
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-20 are rejected under 35 U.S.C. §103 as being unpatentable over Reference 1 in view of Reference 2 and Reference 3, and further in view of Reference 4.
───────
A system for predicting anomalies in a wheel system of a railway vehicle and a track upon which the railway vehicle is run, the system comprising: a vibration sensor positioned to sense vibrations of the wheel system; and an electronic processor communicatively coupled to the vibration sensor, the electronic processor configured to: determine a speed of the railway vehicle; determine whether the speed exceeds a predetermined speed threshold; obtain, in response to the speed exceeding the predetermined speed threshold, a vibration measurement; determine, from the vibration measurement, a classified vibration level; determine whether the classified vibration level is indicative of an anomaly; derive, in response to the classified vibration level being indicative of the anomaly, a dominant vibration frequency of the vibration measurement; perform a comparison between the dominant vibration frequency and a predetermined frequency threshold; identify, based on the comparison and a reference classified vibration level, the anomaly existing within either or both of the railway vehicle wheel system or the track; and perform a mitigation action in response to identifying the anomaly.
Claim 1 is rejected under 35 U.S.C. §103 as being unpatentable over Reference 1 in view of Reference 2 and Reference 3.
Analysis
Reference 1 discloses a system for monitoring a railcar using railcar-mounted sensing devices. Reference 1 teaches a mote 10 connected to a sensor 20 and in communication with a central monitoring unit 32, mobile base station 42, and land-based station 44. Reference 1 further teaches that sensor 20 may be an accelerometer or vibration-type sensor used for dynamic railcar measurements, including adapter acceleration, axle acceleration, bearing fault detection, track damage detection, and truck hunting detection. The monitored railcar runs on railroad track, and Reference 1 expressly evaluates railcar wheel, bearing, truck, and track-related operating conditions. Thus, Reference 1 teaches a system for predicting or detecting anomalies in a railway vehicle wheel/truck system and in the track upon which the railway vehicle runs.
Reference 1 teaches the claimed vibration sensor positioned to sense vibrations of the wheel system. In particular, Reference 1 discloses an adapter-mounted accelerometer corresponding to sensor 20. The adapter-mounted accelerometer samples dynamic bearing data and dynamic data from the adapter or other truck component. Because the adapter and truck components are associated with the railcar wheelset and bearing assembly, sensor 20 is positioned to sense vibrations of the wheel system.
Reference 1 teaches an electronic processor communicatively coupled to the vibration sensor. The mote 10 receives data from sensor 20, and the data is communicated to central monitoring unit 32, mobile base station 42, and/or land-based station 44 for processing, heuristic analysis, alarms, and further action. The processing components of Reference 1 therefore correspond to the claimed electronic processor communicatively coupled to the vibration sensor.
Reference 1 teaches determining a speed of the railway vehicle. Reference 1 discloses an axle-mounted accelerometer used to measure radial acceleration, which is converted to vehicle speed using the wheel and axle diameters. Reference 1 also discloses an axle RPM sensor for generating a signal corresponding to vehicle speed. These disclosures teach determining the speed of the railway vehicle.
Reference 1 does not expressly state, in the same words as claim 1, that the vibration measurement is obtained only “in response to” the speed exceeding a predetermined speed threshold. However, Reference 1 teaches that vehicle speed is determined from railcar sensor data and that dynamic vibration-related data is analyzed for bearing faults, track damage, and truck hunting. Reference 2 further teaches that railway acceleration/vibration signatures are evaluated using predefined levels, predefined functions, and speed-related operating information, including GPS time-velocity-position information. Reference 2 also teaches that hunting oscillation occurs above a critical speed and that acceleration patterns are evaluated relative to thresholds. It would have been obvious to configure the processor of Reference 1 to determine whether the railcar speed exceeds a predetermined speed threshold before obtaining or processing a vibration measurement so that vibration classification is performed only under operating conditions where the vibration signature is meaningful, repeatable, and capable of supporting reliable anomaly detection.
Reference 1 teaches obtaining a vibration measurement. Reference 1 discloses sampling dynamic bearing data with an adapter-mounted accelerometer, sampling dynamic data for track damage detection, and sampling dynamic data for truck hunting detection. Reference 2 likewise teaches accelerometers that monitor dynamic acceleration forces and vibrational movement of railway wagon structures. The sampled dynamic acceleration or vibration signal corresponds to the claimed vibration measurement.
Reference 1 in view of Reference 2 teaches determining, from the vibration measurement, a classified vibration level. Reference 1 teaches computing features from vibration or dynamic data, including FFT-based dominant modes, rolling-frequency amplification, probability density function values, kurtosis, impulsivity, and time-series events. Reference 2 teaches processing acceleration measurements using thresholding, averaging, frequency analysis, predefined acceleration levels, predefined values, functions, and action-related characteristics such as imminent danger, maintenance before service, and future maintenance. A processed vibration measurement assigned to a threshold state, acceleration level, action level, defect state, or warning class corresponds to a classified vibration level.
Reference 1 in view of Reference 2 teaches determining whether the classified vibration level is indicative of an anomaly. Reference 1 teaches that amplification at rolling frequency indicates a likely bearing fault; high kurtosis or impulsivity indicates track defects; and FFT isolation of known hunting frequencies detects truck hunting. Reference 2 teaches that acceleration above predefined levels identifies derailment, wheel flat, rail defect, sun kink, hunting, and severe suspension defect conditions. These teachings correspond to determining whether the classified vibration level is indicative of an anomaly.
Reference 1 and Reference 3 teach deriving, in response to the classified vibration level being indicative of the anomaly, a dominant vibration frequency of the vibration measurement. Reference 1 teaches using FFT on sampled dynamic bearing data to isolate dominant modes and any shifting or relative amplification, with amplification at rolling frequency indicating a likely fault. Reference 3 teaches a capturing device 110 capturing vibration signal VS, initial filter device 111.1 filtering the vibration signal, and QPLL device 111.2 identifying the frequency PVCF of a predominant vibration component PVC in the vibration signal. The predominant vibration component PVC and its frequency PVCF correspond to the claimed dominant vibration frequency.
Reference 3 teaches performing a comparison between the dominant vibration frequency and a predetermined frequency threshold. Reference 3 discloses comparison device 111.3 comparing the frequency PVCF of predominant vibration component PVC against stored potentially damaging resonant frequencies PDRF, with assignment governed by a predeterminable frequency difference threshold FDIFT. Reference 3 also teaches threshold amplitude information TAI selected as a function of frequency PVCF and stored in comparison device 111.3. This teaches comparing the dominant vibration frequency to a predetermined frequency threshold or stored frequency criterion.
Reference 1 in view of Reference 2 and Reference 3 teaches identifying, based on the comparison and a reference classified vibration level, the anomaly existing within either or both of the railway vehicle wheel system or the track. Reference 1 teaches identifying bearing faults, track defects, and truck hunting based on dynamic vibration data from sensor 20 and processing by mote 10, central monitoring unit 32, mobile base station 42, and/or land-based station 44. Reference 2 teaches comparing acceleration levels and stored data, including previous wagon history and comparable acceleration levels from other units or trains at a same geo-positional point, to distinguish wheel flats from rail defects. Reference 3 teaches using frequency PVCF, stored resonant frequencies PDRF, frequency difference threshold FDIFT, amplitude information AI, and threshold amplitude information TAI to identify damaging vibration states in a wheelset 106, wheels 106.1, and wheelset shaft 106.2. The stored, previous, or other-unit acceleration/vibration level of Reference 2 corresponds to the claimed reference classified vibration level, and the combined comparison logic identifies whether the anomaly is in the wheel system, bearing/wheelset, truck/bogie, or track.
Reference 1, Reference 2, and Reference 3 teach performing a mitigation action in response to identifying the anomaly. Reference 1 teaches sending alarms and alerts to a customer or remote monitoring station so that maintenance can be scheduled. Reference 2 teaches warning the train driver or dispatcher, sending imminent danger messages, and identifying maintenance action levels. Reference 3 teaches comparison device 111.3 issuing motor control signal MCS to motor controller 112 of control device 109, causing traction converter device 108 to initiate a reaction to counteract the damaging resonant vibration state, including reducing traction or braking effort. These actions correspond to the claimed mitigation action.
Motivation
It would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to modify the railcar monitoring system of Reference 1 with the threshold-based acceleration pattern analysis and reference-data comparison of Reference 2 because both references address detection of railway vehicle and track anomalies using railcar-mounted sensors and processed vibration/acceleration data. Reference 1 already teaches railcar speed, vibration/dynamic measurements, FFT, bearing faults, track defects, and alerts, while Reference 2 provides a railway-specific technique for classifying acceleration/vibration levels using predefined thresholds and comparing current levels with stored or other-unit reference levels to distinguish wheel and track defects. A person of ordinary skill in the art would have been motivated to combine these teachings to improve anomaly classification accuracy and reduce false positives between wheel/bearing defects and fixed track defects.
It also would have been obvious to incorporate the predominant-frequency analysis and threshold comparison of Reference 3 into the combined system because Reference 1 already uses FFT to isolate dominant modes and rolling-frequency amplification, and Reference 3 provides a rail-vehicle vibration technique for identifying a predominant vibration component PVC, frequency PVCF, stored damaging resonant frequencies PDRF, and threshold information TAI. The combination would predictably improve the ability of the system to identify wheelset-related anomalies based on vibration frequency and to initiate mitigation when a damaging vibration state is detected.
UPDATED 103 REJECTION LANGUAGE
Claims 1-20 are rejected under 35 U.S.C. §103 as being unpatentable over Reference 1 in view of Reference 2 and Reference 3, and further in view of Reference 4.
Reference 1 is the primary reference because Reference 1 is directed to monitoring railcar performance using railcar-mounted sensors and applying processing to detect or predict railcar operational failures. Reference 1 teaches mote 10, sensor 20, CMU 32, railcar 38, wheel bearing 39, wheel 40, vehicle-speed determination, accelerometer/vibration sensing, dynamic bearing data, FFT dominant modes, bearing fault indication, track damage detection, and truck hunting detection.
Reference 2 teaches railway-specific acceleration/vibration classification using sensor/control unit package 5, bogie 1, wheel pairs 2a and 2b, axles 3a and 3b, three-axis acceleration monitoring, predefined thresholds, predefined levels, predetermined acceleration patterns, trigger-event processing, comparable levels from other units at the same geo-positional point, rail defects, wheel flats, sun kinks, suspension defects, GPS time-velocity-position information, and warning actions.
Reference 3 teaches rail-vehicle vibration analysis using control arrangement 102, capturing device 110, analyzing device 111, initial filter device 111.1, QPLL device 111.2, comparison device 111.3, vibration signal VS, predominant vibration component PVC, predominant vibration frequency PVCF, stored potentially damaging resonant frequencies PDRF, frequency-difference threshold FDIFT, threshold amplitude information TAI, wheelset 106, wheels 106.1, wheelset shaft 106.2, motor control signal MCS, motor controller 112, and traction converter device 108.
Reference 4 teaches digital twin system 200, digital twin simulation system 206, digital twin dynamic model system 208, dynamic model datastore 228, digital twin datastore 269, vibration sensors 235, vibration measurements used as model inputs, vibration severity values, probability-of-failure outputs, and bearing vibration fault level states including normal, suboptimal, critical, and alarm. Reference 4 is relied upon to expressly teach the amended “labels” limitation of independent claims 1 and 14 and the digital-twin simulation limitations of claims 3 and 16.
It would have been obvious to modify Reference 1’s railcar monitoring system with Reference 2’s threshold/pattern-based railway acceleration classification and reference-level comparison because both references address detection of railcar, wheel, bogie, and track defects using onboard railway sensors. Reference 1 already teaches obtaining and processing railcar vibration/acceleration data to detect bearing faults, track damage, and truck hunting. Reference 2 provides railway-specific classification using predefined acceleration/vibration levels, patterns, comparable reference levels, GPS/velocity-position information, and warning actions. The modification would have predictably improved anomaly classification and reduced false positives by comparing a current classified vibration/acceleration level with reference levels from other sensors, vehicles, or locations.
It would have been obvious to further incorporate Reference 3’s predominant-frequency analysis into the combined system because Reference 1 already teaches FFT-based dominant-mode analysis and bearing/wheel/truck vibration detection, while Reference 3 provides a detailed rail-vehicle technique for identifying a predominant vibration frequency and comparing it to stored damaging resonant-frequency criteria. The modification would have predictably improved identification of wheelset, wheel, and bearing-related vibration anomalies based on dominant vibration frequency.
It would have been obvious to further incorporate Reference 4’s labeled vibration fault level states and vibration-measurement-based dynamic modeling because the combined system already collects vibration measurements and classifies railway anomalies. Reference 4 provides known transportation-system digital-twin techniques for using vibration sensor data to generate fault/severity states, including normal, suboptimal, critical, and alarm labels. Using such labels in Reference 1’s railcar anomaly system would have predictably improved organization, communication, and downstream use of vibration classifications.
Accordingly, amended claims 1-20 remain unpatentable under 35 U.S.C. §103 over Reference 1 in view of Reference 2 and Reference 3, and further in view of Reference 4.
───────
The system of claim 1, wherein the comparison includes determining whether the classified vibration level exceeds a predetermined vibration level threshold.
Analysis
Reference 1, Reference 2, and Reference 3 teach the system of claim 1 as set forth above.
Reference 2 teaches determining whether an acceleration or vibration-related level exceeds a predefined level or threshold. The sensor/control unit package 5 monitors acceleration in three dimensions, and the microprocessor compares processed acceleration signals with predefined values or functions to determine wheel flat development and other bogie defects. Reference 2 further teaches that acceleration levels exceeding a given threshold trigger warnings and that increased acceleration forces exceeding a predefined level identify events such as derailment and wheel flat. Thus, Reference 2 teaches determining whether a classified vibration or acceleration level exceeds a predetermined vibration level threshold.
Reference 3 also teaches threshold comparison. Comparison device 111.3 compares amplitude information AI, derived from the vibration signal VS and predominant vibration component PVC, with threshold amplitude information TAI selected as a function of frequency PVCF. The threshold amplitude information TAI represents the level at which vibration becomes damaging to wheelset 106, wheels 106.1, wheelset shaft 106.2, or the press fit. This further teaches comparing a classified vibration level to a predetermined vibration level threshold.
Motivation
It would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to include a vibration-level threshold in the combined system because Reference 1 already processes vibration data to detect bearing faults, track damage, and hunting, while Reference 2 and Reference 3 teach that railway vibration/acceleration events are reliably classified by comparing measured or processed levels to predetermined thresholds. The use of a threshold would have predictably provided a simple and repeatable criterion for determining whether the detected vibration condition is sufficiently severe to be treated as an anomaly.
───────
The system of claim 1, wherein determining the classified vibration level includes performing a simulation with a digital twin using the vibration measurement.
Analysis
Reference 1, Reference 2, and Reference 3 teach the system of claim 1 as set forth above.
Reference 4 teaches determining transportation-related vibration severity and failure information using a digital twin. Reference 4 discloses digital twin system 200, digital twin management system 202, digital twin I/O system 204, digital twin simulation system 206, digital twin dynamic model system 208, dynamic model datastore 228, vibration sensors 235, output module 273, and digital twin datastore 269. Reference 4 teaches that dynamic model system 208 retrieves digital twins from digital twin datastore 269, retrieves dynamic models from dynamic model datastore 228, selects vibration measurements from vibration sensors 235 as input data, runs dynamic models using those vibration measurements and historical failure data, calculates output values such as probability of failure, downtime, or shutdown, and updates the digital twin based on the output values.
Reference 4 therefore teaches determining a vibration-related classification or severity state by performing simulation or dynamic-model processing with a digital twin using vibration measurements. Applying Reference 4 to the vibration measurement of Reference 1, as classified using the threshold and frequency analysis of References 2 and 3, yields the claimed feature of determining the classified vibration level by performing a simulation with a digital twin using the vibration measurement.
Motivation
It would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to incorporate the digital twin simulation of Reference 4 into the railcar anomaly monitoring system of Reference 1, as modified by References 2 and 3, because the combined system already uses vibration measurements, thresholds, frequency analysis, historical/reference data, and anomaly classifications. Reference 4 provides a known digital-twin framework for using vibration measurements from vibration sensors 235 as inputs to dynamic models to update transportation-system or machine-component digital twins with failure-related outputs. A person of ordinary skill would have been motivated to use such a digital twin to improve prediction accuracy, support simulation-based classification, and correlate real-time railcar vibration measurements with modeled component-health states.
───────
The system of claim 1, wherein, in response to the dominant vibration frequency exceeding the predetermined frequency threshold, the electronic processor identifies that the anomaly is a wheel anomaly, a bearing anomaly, or both.
Analysis
Reference 1, Reference 2, and Reference 3 teach the system of claim 1 as set forth above.
Reference 1 teaches that an adapter-mounted accelerometer corresponding to sensor 20 samples dynamic bearing data, and FFT is used on sampled data to isolate dominant modes, shifting, and relative amplification. Reference 1 teaches that amplification at rolling frequency indicates a likely bearing fault. Thus, when the dominant vibration feature occurs at the relevant rolling or bearing frequency, Reference 1 identifies a bearing anomaly.
Reference 3 teaches determining frequency PVCF of predominant vibration component PVC and comparing that frequency to stored potentially damaging resonant frequencies PDRF using comparison device 111.3 and frequency difference threshold FDIFT. Reference 3 further teaches that the threshold amplitude information TAI is associated with wheelset 106, wheels 106.1, wheelset shaft 106.2, and press-fit damage. Thus, Reference 3 teaches identifying a wheelset or wheel-related anomaly based on a dominant-frequency comparison.
Reference 2 teaches that wheel flat conditions and other wheel/bogie events are identified from acceleration patterns and predefined acceleration levels. Accordingly, the combined references teach identifying that the anomaly is a wheel anomaly, a bearing anomaly, or both in response to the dominant vibration frequency exceeding or satisfying the predetermined frequency criterion.
Motivation
It would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to use dominant-frequency thresholding to identify wheel and bearing anomalies because Reference 1 already teaches that rolling-frequency amplification indicates a likely bearing fault, while Reference 3 teaches assigning a predominant vibration frequency to damaging wheelset vibration scenarios using stored frequency thresholds. Combining these teachings would have predictably improved discrimination of bearing, wheel, and wheelset anomalies from general track-induced vibration events.
───────
The system of claim 4, wherein the electronic processor determines that the anomaly the anomaly includes at least one selected from the group consisting of a deformation in a wheel of the wheel system, a loose wheel of the wheel system, a worn bearing, a loose bearing, and a broken bearing.
Analysis
Reference 1, Reference 2, and Reference 3 teach the system of claim 4 as set forth above.
Reference 1 teaches determining bearing faults from dynamic bearing data sampled by an adapter-mounted accelerometer corresponding to sensor 20. Reference 1 further teaches that amplification at rolling frequency indicates a likely bearing fault, and that kurtosis or acoustic features may be used as indicators of bearing damage. A worn, loose, or broken bearing is a type of bearing fault or bearing damage, and therefore falls within the bearing anomalies taught by Reference 1.
Reference 2 teaches wheel flat detection and wheel flat size estimation using acceleration measurements from the sensor/control unit package 5 mounted near bogie 1 and associated with wheel pairs 2a and 2b and axles 3a and 3b. A wheel flat is a deformation or defect of a wheel of the wheel system. Reference 2 therefore teaches identifying deformation in a wheel of the wheel system.
Reference 3 teaches wheelset damage associated with wheels 106.1, wheelset shaft 106.2, and the press fit between wheels 106.1 and wheelset shaft 106.2. A damaging vibration state involving the press fit between the wheels and shaft corresponds to a loose wheel or wheel attachment anomaly. Thus, the combined references teach determining that the anomaly includes a wheel deformation, loose wheel, worn bearing, loose bearing, or broken bearing.
Motivation
It would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to classify the detected wheel/bearing anomaly into specific wheel and bearing defect types because Reference 1 already identifies bearing faults from dynamic vibration data, Reference 2 identifies wheel flats and wheel flat size from acceleration data, and Reference 3 identifies damaging wheelset vibration involving wheels 106.1, shaft 106.2, and press-fit damage. Specific defect classification would have predictably improved maintenance triage by indicating whether the detected anomaly requires wheel repair, wheel replacement, bearing inspection, or bearing replacement.
───────
The system of claim 1, wherein identifying the anomaly includes obtaining the reference classified vibration level from a second vibration sensor of the railway vehicle for a predetermined vibration level pattern; and wherein identifying the anomaly includes evaluating whether the reference classified vibration level is comparable to the classified vibration level.
Analysis
Reference 1, Reference 2, and Reference 3 teach the system of claim 1 as set forth above.
Reference 2 teaches that a train may be equipped with several systems, including systems mounted close to bogies and multiple systems on a wagon. Reference 2 discloses sensor/control unit package 5 mounted on bogie 1 or sprung bogie structure 1b, and teaches that each wagon may include two systems mounted close to each bogie. Each package 5 includes an acceleration sensor, control unit, storage, GPS, and transceiver, and monitors acceleration/vibration signatures.
Reference 2 teaches obtaining reference levels from other sensor units and evaluating whether levels are comparable. For rail defects, Reference 2 teaches that if a sensor unit measures vertical acceleration levels above a certain damaging level and other units, including units mounted on other trains, sense comparable levels at the same geo-positional point, this is a strong indication of a rail defect. Reference 2 also teaches front and back sensors detecting acceleration events at different times as a train passes a rail defect. These other-unit or second-sensor measurements correspond to a reference classified vibration level obtained from a second vibration sensor.
Reference 2 further teaches predetermined patterns and thresholds. The microprocessor compares processed acceleration signals with predefined values or functions, and the signal processing may include correlator circuits to isolate forces that are significant when detected from more than one source. Thus, Reference 2 teaches obtaining a reference classified vibration level from a second sensor for a predetermined vibration level pattern and evaluating whether that reference level is comparable to the classified vibration level.
Motivation
It would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to add second-sensor reference comparison to the railcar monitoring system of Reference 1 because Reference 2 teaches that comparing acceleration/vibration levels from multiple railway sensor units improves discrimination between a wheel defect, which repeats with the same vehicle, and a rail or track defect, which appears at the same location for different sensors or trains. A person of ordinary skill would have been motivated to compare the first classified vibration level with a reference classified vibration level from a second sensor to reduce false positives and identify whether the anomaly is localized to the vehicle or to the track.
───────
The system of claim 6, wherein the second vibration sensor is positioned on a same side of the railway vehicle as the vibration sensor.
Analysis
Reference 1, Reference 2, and Reference 3 teach the system of claim 6 as set forth above.
Reference 2 teaches using multiple sensor/control unit packages 5 mounted close to bogies of the same railway wagon and using front/back sensor comparisons to distinguish rail defects from wheel defects. The bogie 1 includes wheel pairs 2a and 2b mounted on axles 3a and 3b, and the monitored railway wagon travels on left and right rails of the track. Reference 2 teaches using sensor placement to determine whether a rail defect occurs as the vehicle passes the track location.
Although Reference 2 does not expressly use the phrase “same side,” it would have been obvious to place the second vibration sensor on the same side of the railway vehicle as the first vibration sensor when the purpose is to compare vibration levels caused by the same rail, same wheel path, and same side of the bogie. Placing the second sensor on the same side allows both sensors to evaluate the same rail-side vibration event rather than mixing left-rail and right-rail events.
Motivation
It would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to position the second sensor on the same side of the vehicle as the first sensor because Reference 2 teaches using multiple railway sensor units to determine whether acceleration/vibration events are associated with rail defects. A person of ordinary skill would have understood that same-side placement provides a more direct comparison for defects on the same rail and improves the reliability of determining whether a vibration pattern is caused by the track rather than by a particular wheel or bogie component.
───────
The system of claim 7, wherein the electronic processor is further configured to identify, in response to the reference classified vibration level being comparable to the classified vibration level, an anomaly with the track the wheel system is on.
Analysis
Reference 1, Reference 2, and Reference 3 teach the system of claim 7 as set forth above.
Reference 2 expressly teaches that comparable acceleration levels detected by other units at the same geo-positional point indicate a rail defect. Reference 2 also teaches that rail defects may be detected by acceleration sensors in front and back of the train as the vehicle passes the defect. This teaches identifying a track anomaly when a reference classified vibration level from another sensor is comparable to the classified vibration level.
Reference 1 also teaches track damage detection using an adapter-mounted accelerometer corresponding to sensor 20. Reference 1 computes probability density function and kurtosis from dynamic data, and high kurtosis or impulsivity indicates track defects. Thus, Reference 1 provides the primary railcar monitoring system and Reference 2 supplies the comparative multi-sensor track-defect logic.
Motivation
It would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to identify a track anomaly when a second sensor produces a comparable classified vibration level because Reference 2 teaches that repeated comparable levels at the same location indicate a rail defect, while Reference 1 already detects track damage using railcar dynamic vibration data. The combination would have predictably allowed the processor to distinguish a track defect from a vehicle-specific wheel or bearing defect.
───────
The system of claim 8, wherein the vibration sensor is positioned vertical with respect to a wheel axle of the wheel system and the second vibration sensor is positioned horizontal with respect to a wheel axle of the wheel system.
Analysis
Reference 1, Reference 2, and Reference 3 teach the system of claim 8 as set forth above.
Reference 2 teaches acceleration monitoring along three mutually orthogonal axes, including a vertical axis relative to the railroad track and other axes corresponding to longitudinal and lateral directions. Reference 2 further teaches that the sensor/control unit package 5 includes a 3D accelerometer with lateral sensitivity, longitudinal sensitivity, and vertical sensitivity. The vertical measurement axis corresponds to a vibration sensor positioned to measure vertical vibration with respect to a wheel axle, and the lateral or longitudinal measurement axis corresponds to a horizontal measurement direction with respect to the wheel axle.
Reference 1 similarly teaches accelerometer-based measurements at railcar components, including adapter acceleration, axle acceleration, and truck-component dynamic data. It would have been obvious to implement the orthogonal sensing of Reference 2 as separate vertical and horizontal vibration sensors rather than a single 3D accelerometer, because separate sensors for different axes were a known equivalent implementation for measuring directional vibration components.
Motivation
It would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to position one sensor to measure vertical vibration and another sensor to measure horizontal vibration because Reference 2 teaches that railway anomalies produce different signatures in vertical, lateral, and longitudinal acceleration. Separating vertical and horizontal measurements would predictably improve classification of vertical track defects, lateral track defects, hunting, wheel flats, and bogie events by preserving direction-specific vibration information.
───────
The system of claim 9, wherein the electronic processor is further configured to identify, in response to the predetermined vibration pattern being present within the reference classified vibration level, that the anomaly includes a horizontal geometric defect with the track the wheel system is on.
Analysis
Reference 1, Reference 2, and Reference 3 teach the system of claim 9 as set forth above.
Reference 2 teaches detecting rail defects and sun kinks using lateral acceleration. A sun kink is a buckling in the rail track, and Reference 2 teaches detecting such a condition by lateral accelerations above a specific threshold in several railroad wagons in a train. Lateral acceleration corresponds to a horizontal vibration pattern with respect to the track and wheel axle. Thus, Reference 2 teaches identifying a horizontal geometric defect in the track when a predetermined lateral or horizontal vibration pattern is present.
Reference 1 teaches track damage detection using dynamic data from an adapter-mounted accelerometer corresponding to sensor 20 and determining track defects from high kurtosis or impulsivity. Combined with the lateral-axis and threshold teachings of Reference 2, the system identifies a horizontal geometric track defect from a predetermined vibration pattern present in the reference classified vibration level.
Motivation
It would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to identify a horizontal geometric track defect from a predetermined horizontal vibration pattern because Reference 2 teaches that lateral acceleration above a threshold in multiple wagons indicates sun kinks or lateral track buckling, and Reference 1 teaches detecting track damage from railcar dynamic sensor data. The combination would predictably improve track-defect classification by distinguishing horizontal/lateral geometry defects from vertical rail defects and wheel-related events.
───────
The system of claim 1, wherein identifying the anomaly includes evaluating whether a reference classified vibration level from a second vibration sensor is comparable to the classified vibration level, wherein the second vibration sensor is positioned on a second wheel system of the railway vehicle.
Analysis
Reference 1, Reference 2, and Reference 3 teach the system of claim 1 as set forth above.
Reference 2 teaches multiple sensor systems on a railway wagon, including systems mounted close to bogies. The bogie 1 includes wheel pairs 2a and 2b mounted on axles 3a and 3b, and each wagon may include multiple systems close to different bogies or wheel assemblies. Thus, a second sensor/control unit package 5 may be positioned on or near a second wheel system of the railway vehicle.
Reference 2 teaches comparing acceleration/vibration levels from different sensor units and using comparable levels to identify rail defects or other events. Reference 2 also teaches correlator circuits that isolate forces significant only when detected from more than one source. Accordingly, Reference 2 teaches evaluating whether a reference classified vibration level from a second vibration sensor positioned on a second wheel system is comparable to the classified vibration level from the first wheel system.
Motivation
It would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to compare vibration levels from first and second wheel systems because Reference 2 teaches that multiple sensor systems on a railway wagon or train can distinguish vehicle-specific defects from track defects by comparing acceleration levels across sensor locations. A person of ordinary skill would have been motivated to place the second sensor on a second wheel system to determine whether the anomaly repeats across wheel systems or is localized to a particular wheel, bogie, or suspension assembly.
───────
The system of claim 11, wherein the electronic processor determines that the anomaly is a malfunctioning suspension of the wheel system in response to determining that the reference classified vibration level is not comparable to the classified vibration level.
Analysis
Reference 1, Reference 2, and Reference 3 teach the system of claim 11 as set forth above.
Reference 2 teaches that the system detects severe defects in the suspension system of an individual railway wagon. Reference 2 also teaches comparing acceleration levels and using sensor data from different systems to determine whether an event is a rail defect, wheel flat, derailment, hunting, or suspension-related defect. Where the classified vibration level from one wheel system is not comparable to the reference classified vibration level from a second wheel system, the anomaly is localized to the first wheel system rather than being a common track event.
In view of Reference 2’s teaching of severe suspension defects and multi-sensor comparison, it would have been obvious for the processor to determine that a non-comparable vibration condition localized to one wheel system corresponds to a malfunctioning suspension associated with that wheel system.
Motivation
It would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to classify a non-comparable localized vibration level as a suspension malfunction because Reference 2 expressly identifies severe suspension defects as detectable bogie operational defects and teaches comparing levels from multiple sensor units. If a track defect produces comparable events across sensors but a suspension defect produces a localized response, then non-comparability between the first and second wheel-system vibration levels provides a predictable indication of a local suspension anomaly.
───────
The system of claim 1, wherein the electronic processor is further configured to: determine a velocity profile for the railway vehicle; and identify the anomaly by identifying the anomaly based on the comparison, the reference classified vibration level, and the velocity profile.
Analysis
Reference 1, Reference 2, and Reference 3 teach the system of claim 1 as set forth above.
Reference 1 teaches determining vehicle speed from radial axle acceleration or axle RPM. Reference 2 teaches using GPS time-velocity-position information and acceleration data to evaluate railway events. Reference 2 further teaches identifying wheel flats after braking and subsequent acceleration, detecting events caused by locked brakes, evaluating braking-related acceleration patterns, and identifying hunting oscillation above a critical speed. These teachings correspond to determining a velocity profile for the railway vehicle.
Reference 2 teaches that anomaly identification may depend on whether the railway vehicle is braking, accelerating, moving, or traveling above a critical speed. Reference 3 teaches identifying damaging vibration states based on frequency comparison, and Reference 2 teaches using reference/comparable levels to discriminate wheel and rail defects. Therefore, the combined references teach identifying the anomaly based on the dominant-frequency comparison, the reference classified vibration level, and the velocity profile.
Motivation
It would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to include a velocity profile in the anomaly-identification logic because railway vibration signatures depend on speed, acceleration, braking, and critical-speed conditions. Reference 1 already determines railcar speed, Reference 2 uses velocity-position and operating-state information to classify wheel flats, braking events, hunting, and rail defects, and Reference 3 uses frequency information for damaging vibration states. Combining these inputs would predictably improve anomaly classification by accounting for the vehicle operating condition at the time the vibration event occurs.
───────
A method for predicting anomalies in a wheel system of a railway vehicle and a track upon which the railway vehicle is run, the method comprising: determining, with an electronic processor, a speed of the railway vehicle; determining whether the speed exceeds a predetermined speed threshold; obtaining, in response to the speed exceeding the predetermined speed threshold, a vibration measurement from a vibration sensor positioned to sense vibrations of the wheel system; determining from the vibration measurement, a classified vibration level; determining whether the classified vibration level is indicative of an anomaly; deriving, in response to the classified vibration level being indicative of the anomaly, a dominant vibration frequency of the vibration measurement; performing a comparison between the dominant vibration frequency and a predetermined frequency threshold; identifying, based on the comparison and a reference classified vibration level, the anomaly existing within either or both of the railway vehicle wheel system and the track; and performing a mitigation action in response to identifying the anomaly.
Analysis
Reference 1, Reference 2, and Reference 3 teach the method of claim 14 for substantially the same reasons set forth above for claim 1.
Reference 1 teaches determining, with an electronic processor, a speed of the railway vehicle by using an axle-mounted accelerometer or axle RPM sensor to determine vehicle speed, with processing performed by mote 10, central monitoring unit 32, mobile base station 42, or land-based station 44.
Reference 1 in view of Reference 2 teaches determining whether the speed exceeds a predetermined speed threshold. Reference 1 determines speed and processes vibration data, while Reference 2 teaches speed-related railway anomaly conditions, including hunting above critical speed, and uses predefined levels and thresholds. It would have been obvious to condition vibration acquisition or processing on exceeding a predetermined speed threshold so that vibration signatures are evaluated only when the railway vehicle is moving fast enough to generate reliable anomaly-indicative measurements.
Reference 1 teaches obtaining a vibration measurement from a vibration sensor positioned to sense vibrations of the wheel system. Sensor 20, implemented as an adapter-mounted accelerometer, samples dynamic bearing data and dynamic track or truck-component data associated with the wheel system. Reference 2 further teaches accelerometers monitoring dynamic acceleration forces and vibrational movement.
Reference 1 in view of Reference 2 teaches determining from the vibration measurement a classified vibration level and determining whether that level is indicative of an anomaly. Reference 1 computes FFT dominant modes, rolling-frequency amplification, kurtosis, and impulsivity for bearing faults, track defects, and truck hunting. Reference 2 classifies acceleration measurements using predefined levels, values, functions, thresholds, and action levels to identify wheel flats, rail defects, sun kinks, derailment, hunting, and suspension defects.
Reference 1 and Reference 3 teach deriving, in response to the classified level being indicative of an anomaly, a dominant vibration frequency of the vibration measurement. Reference 1 uses FFT to isolate dominant modes and rolling-frequency amplification, and Reference 3 uses QPLL device 111.2 to identify frequency PVCF of predominant vibration component PVC from vibration signal VS.
Reference 3 teaches performing a comparison between the dominant vibration frequency and a predetermined frequency threshold. Comparison device 111.3 compares frequency PVCF against stored potentially damaging resonant frequencies PDRF using frequency difference threshold FDIFT, and also compares amplitude information AI with threshold amplitude information TAI selected as a function of frequency.
Reference 1 in view of Reference 2 and Reference 3 teaches identifying, based on the comparison and a reference classified vibration level, whether the anomaly exists in the wheel system, track, or both. Reference 1 identifies bearing faults, track defects, and hunting; Reference 2 uses comparable levels from other units and stored history to distinguish rail defects from wheel defects; and Reference 3 identifies wheelset damaging vibration states associated with wheelset 106, wheels 106.1, and shaft 106.2.
Reference 1, Reference 2, and Reference 3 teach performing a mitigation action in response to identifying the anomaly. Reference 1 teaches alarms and maintenance scheduling, Reference 2 teaches warning the train driver or dispatcher and sending maintenance-action alerts, and Reference 3 teaches motor control signal MCS to motor controller 112 and traction converter 108 to counteract a damaging vibration state.
Motivation
It would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to perform the method of claim 14 by combining Reference 1 with Reference 2 and Reference 3 for the same reasons set forth for claim 1. Reference 1 provides the railcar monitoring platform, Reference 2 provides railway-specific threshold classification and reference-level comparison for wheel, track, and suspension events, and Reference 3 provides dominant-frequency comparison and mitigation for rail-vehicle wheelset vibration. The combination would have predictably improved reliability and specificity of anomaly prediction for railway wheel systems and track.
───────
The method of claim 14, wherein the comparison includes determining whether the classified vibration level exceeds a predetermined vibration level threshold.
Analysis
Reference 1, Reference 2, and Reference 3 teach the method of claim 14 as set forth above.
Reference 2 teaches comparing monitored acceleration signals with predefined acceleration levels, predefined values, and predefined functions. Reference 2 further teaches triggering warnings when acceleration levels exceed a threshold and identifying derailment or wheel flat events when increased acceleration forces exceed a predefined level. Reference 3 teaches comparing amplitude information AI, derived from vibration signal VS and predominant vibration component PVC, with threshold amplitude information TAI in comparison device 111.3. These teachings correspond to determining whether a classified vibration level exceeds a predetermined vibration level threshold.
Motivation
It would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to include the vibration-level threshold determination in the method because Reference 2 and Reference 3 teach that railway vibration/acceleration anomalies are effectively identified by comparing measured or processed levels with predetermined thresholds. The modification would have yielded the predictable benefit of objective, repeatable anomaly classification.
───────
The method of claim 14, wherein determining the classified vibration level includes performing a simulation with a digital twin using the vibration measurement.
Analysis
Reference 1, Reference 2, and Reference 3 teach the method of claim 14 as set forth above.
Reference 4 teaches digital twin system 200 with digital twin simulation system 206 and digital twin dynamic model system 208. Reference 4 teaches retrieving one or more digital twins from digital twin datastore 269, retrieving dynamic models from dynamic model datastore 228, selecting data from vibration sensors 235, running dynamic models using vibration measurements and historical failure data, and calculating output values representing probability of failure, downtime, shutdown, or vibration severity. This corresponds to determining a classified vibration level by performing simulation with a digital twin using the vibration measurement.
Motivation
It would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to use the digital twin simulation of Reference 4 in the method of claim 14 because the method already uses vibration measurements, thresholds, frequency comparison, and reference vibration levels to classify railway anomalies. Reference 4 provides a known transportation-system digital twin architecture for converting vibration sensor inputs into modeled failure or severity outputs. The combination would have predictably improved predictive maintenance and anomaly classification by comparing measured railcar vibration behavior with simulated component-health behavior.
───────
The method of claim 14, further comprising: in response to the dominant vibration frequency exceeding the predetermined frequency threshold, identifying that the anomaly is a wheel anomaly, a bearing anomaly, or both.
Analysis
Reference 1, Reference 2, and Reference 3 teach the method of claim 14 as set forth above.
Reference 1 teaches using FFT on dynamic bearing data to isolate dominant modes and identify bearing faults when amplification occurs at rolling frequency. Reference 3 teaches comparing frequency PVCF of predominant vibration component PVC to stored potentially damaging resonant frequencies PDRF using comparison device 111.3 and frequency difference threshold FDIFT. Reference 3 further associates the damaging vibration state with wheelset 106, wheels 106.1, and wheelset shaft 106.2. Reference 2 teaches wheel flat detection from acceleration patterns and predefined levels. These teachings collectively disclose identifying the anomaly as a wheel anomaly, bearing anomaly, or both in response to the dominant-frequency comparison.
Motivation
It would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to identify the anomaly as a wheel or bearing anomaly based on dominant vibration frequency because Reference 1 teaches rolling-frequency bearing fault identification and Reference 3 teaches damaging wheelset vibration identification using predominant frequency. The combination would have predictably provided a frequency-based method for separating wheel/bearing events from other railcar or track events.
───────
The method of claim 17, further comprising: determines that the anomaly the anomaly includes at least one selected from the group consisting of a deformation in a wheel of the wheel system, a loose wheel of the wheel system, a worn bearing, a loose bearing, and a broken bearing.
Analysis
Reference 1, Reference 2, and Reference 3 teach the method of claim 17 as set forth above.
Reference 2 teaches identifying wheel flats and estimating wheel flat size using acceleration data from sensor/control unit package 5 associated with bogie 1, wheel pairs 2a and 2b, and axles 3a and 3b. A wheel flat corresponds to a deformation in a wheel of the wheel system.
Reference 3 teaches wheelset damaging vibration associated with wheels 106.1, wheelset shaft 106.2, and the press fit between wheels 106.1 and shaft 106.2. A damaging condition involving the press fit corresponds to a loose wheel condition.
Reference 1 teaches bearing fault detection using adapter-mounted accelerometer data, FFT dominant modes, rolling-frequency amplification, kurtosis, acoustic emission, and bearing damage indicators. Worn, loose, and broken bearings are common bearing damage or bearing fault conditions within the scope of Reference 1’s bearing-fault detection. Thus, the combined references teach determining that the anomaly includes a wheel deformation, loose wheel, worn bearing, loose bearing, or broken bearing.
Motivation
It would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to determine the specific wheel or bearing defect type because Reference 1, Reference 2, and Reference 3 each teach detecting particular railcar wheelset or bearing faults from vibration/acceleration signatures. Classifying the anomaly into deformation, loose wheel, worn bearing, loose bearing, or broken bearing would have predictably supported appropriate maintenance action and reduced ambiguity in repair decisions.
───────
The method of claim 14, wherein identifying the anomaly includes obtaining the reference classified vibration level from a second vibration sensor of the railway vehicle for a predetermined vibration level pattern; and wherein identifying the anomaly includes evaluating whether the reference classified vibration level is comparable to the classified vibration level; wherein the second vibration sensor is positioned on a same side of the railway vehicle as the vibration sensor; and wherein the method further comprises identifying, in response to the reference classified vibration level being comparable to the classified vibration level, an anomaly with the track the wheel system is on.
Analysis
Reference 1, Reference 2, and Reference 3 teach the method of claim 14 as set forth above.
Reference 2 teaches using multiple sensor/control unit packages 5 on a railway wagon or train. Each package 5 may include an acceleration sensor, control unit, storage, GPS, and transceiver. Reference 2 teaches that other sensor units may sense comparable acceleration levels at the same geo-positional point and that such comparable levels strongly indicate a rail defect. Reference 2 also teaches front and back sensors detecting rail defects as a train passes the defective location. Thus, Reference 2 teaches obtaining a reference classified vibration level from a second vibration sensor and evaluating whether it is comparable to the classified vibration level.
Reference 2 teaches predetermined patterns and thresholds by using predefined acceleration levels, predefined values, predefined functions, and acceleration patterns to detect wheel flats, rail defects, sun kinks, hunting, and suspension defects. Reference 2’s correlator circuits and multi-source comparison teach evaluating predetermined vibration or acceleration patterns across sensors.
Reference 2 does not expressly require the second sensor to be on the same side of the railway vehicle. However, as explained for claim 7, it would have been obvious to position the second sensor on the same side as the first sensor when comparing vibration levels associated with the same rail or same wheel path. Same-side placement predictably improves track-defect confirmation because both sensors observe the same rail-side vibration event.
Reference 1 teaches track damage detection using dynamic data from sensor 20. Reference 2 teaches that comparable levels from multiple sensors at the same geo-positional point indicate a rail defect. Therefore, the combined references teach identifying, in response to comparable classified and reference classified vibration levels, an anomaly with the track the wheel system is on.
Motivation
It would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to obtain a reference classified vibration level from a same-side second sensor and identify a track anomaly when that level is comparable to the first classified vibration level because Reference 2 teaches that comparable acceleration levels from multiple sensor units at the same track location indicate a rail defect. Same-side placement would have predictably improved the comparison by ensuring that both sensors evaluate the same rail path, while Reference 1’s track damage detection provides the railcar monitoring context for acting on that comparison.
───────
The method of claim 14, further comprising: determining a velocity profile for the railway vehicle; wherein identifying the anomaly includes identifying the anomaly based on the comparison, the reference classified vibration level, and the velocity profile.
Analysis
Reference 1, Reference 2, and Reference 3 teach the method of claim 14 as set forth above.
Reference 1 teaches determining vehicle speed from radial axle acceleration or axle RPM. Reference 2 teaches using GPS time-velocity-position information, acceleration measurements, and vehicle operating-state information. Reference 2 further teaches that wheel flat events may be associated with braking and subsequent acceleration, that locked brakes can produce longitudinal acceleration patterns, and that hunting may occur above a critical speed. These teachings correspond to determining a velocity profile for the railway vehicle.
Reference 3 teaches using frequency comparison to identify damaging vibration states, and Reference 2 teaches using comparable reference levels to distinguish rail defects from wheel defects. Combining these teachings with the velocity profile provides identification based on the frequency comparison, reference classified vibration level, and velocity profile.
Motivation
It would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to identify the anomaly using the velocity profile along with the frequency comparison and reference classified vibration level because railway vibration signatures vary with speed, acceleration, braking, and critical-speed conditions. Reference 1 already determines speed, Reference 2 teaches velocity-position and operating-state relevance to wheel, brake, rail, and hunting anomalies, and Reference 3 teaches dominant-frequency comparison. The combined use of these inputs would have predictably improved anomaly identification accuracy.
Response to Arguments
Applicant's arguments filed 06/12/2026 have been fully considered but they are not persuasive.
Applicant argues that the applied references fail to teach “determine, from the vibration measurement, a classified vibration level, wherein the classified vibration level includes one or more labels indicating one or more characteristic features of the vibration measurement.”
This argument is not persuasive.
The amended limitation is broad. The claims do not require a particular machine-learning architecture, a particular classifier model, a particular label taxonomy, a natural-language label, a confidence score, or any specific data structure for the claimed “labels.” The claims require only that the classified vibration level include one or more labels indicating one or more characteristic features of the vibration measurement.
Reference 1 teaches processing vibration/acceleration data from sensor 20 through mote 10, CMU 32, and/or other processing components to identify railcar operating conditions, including bearing fault indications, track damage detection, and truck hunting detection. Reference 1 teaches that FFT may be performed on sampled dynamic bearing data to isolate dominant modes, shifting, and relative amplification, and that amplification at rolling frequency indicates a likely bearing fault. Reference 1 also teaches that high kurtosis or impulsivity indicates track defects and that FFT may isolate known hunting frequencies. These processed outputs are not raw vibration measurements; they are classified condition indicators derived from characteristic vibration features.
Reference 2 further teaches detecting railway acceleration/vibration patterns and comparing sensed acceleration data to predefined thresholds, predefined criteria, predefined levels, and predefined intervals. Reference 2 teaches identifying wheel flats, rail defects, sun kinks, severe suspension defects, hunting, and other bogie operational problems from such processed acceleration/vibration information. These defect determinations are classified states indicating characteristic features of the measured acceleration/vibration data.
Reference 4 expressly teaches the amended “labels” limitation. Reference 4 teaches digital twin system 200, digital twin simulation system 206, digital twin dynamic model system 208, dynamic model datastore 228, vibration sensors 235, digital twin datastore 269, and output module 273. Reference 4 teaches using vibration measurements from vibration sensors 235 as inputs to dynamic models and generating vibration severity or fault-level outputs. Reference 4 further teaches bearing vibration fault level states, including normal, suboptimal, critical, and alarm. These states are labels indicating characteristic features or severity conditions derived from vibration measurements.
Accordingly, even if Applicant contends that Reference 1 and Reference 2 do not expressly use the term “label,” Reference 4 teaches labeled vibration fault states derived from vibration measurements. The claims do not exclude condition labels, severity labels, threshold-state labels, or fault-state labels. Under the broadest reasonable interpretation consistent with the specification, the claimed “labels” encompass the normal/suboptimal/critical/alarm states of Reference 4, the wheel-flat/rail-defect/sun-kink classifications of Reference 2, and the bearing-fault/track-damage/truck-hunting indicators of Reference 1.
Applicant further argues that the references fail to teach the claimed conditional processing sequence, including obtaining a vibration measurement in response to speed exceeding a threshold, deriving a dominant vibration frequency in response to the classified vibration level being indicative of an anomaly, and identifying the anomaly based on both the frequency comparison and a reference classified vibration level.
This argument is not persuasive.
The rejection is not based on Reference 1 alone. Reference 1 teaches the railcar monitoring platform, including sensor 20, mote 10, CMU 32, railcar 38, wheel bearing 39, wheel 40, railcar speed determination, vibration/acceleration sensing, FFT analysis, bearing fault detection, track damage detection, and truck hunting detection.
Reference 2 teaches railway acceleration/vibration classification using sensor/control unit package 5, bogie 1, wheel pairs 2a and 2b, axles 3a and 3b, GPS/velocity-position information, predefined acceleration levels, predefined thresholds, predetermined acceleration patterns, and trigger-event processing. Reference 2 also teaches speed-dependent railway phenomena, including hunting above a critical speed, and teaches that certain acceleration/vibration patterns trigger further monitoring or warning action.
Reference 3 teaches detailed dominant-frequency analysis of a rail-vehicle vibration signal. Reference 3 teaches capturing device 110, analyzing device 111, initial filter device 111.1, QPLL device 111.2, comparison device 111.3, vibration signal VS, predominant vibration component PVC, predominant vibration frequency PVCF, stored potentially damaging resonant frequencies PDRF, frequency-difference threshold FDIFT, amplitude information AI, and threshold amplitude information TAI.
It would have been obvious to use speed as an initial gating condition in the system of Reference 1 because Reference 1 already determines railcar speed and analyzes vibration/acceleration data, while Reference 2 teaches that railway vibration/acceleration events depend on speed and predefined operating thresholds. A railcar traveling below a predetermined speed may not generate a vibration signature suitable for reliable classification of wheel, bearing, hunting, or track anomalies. Using a predetermined speed threshold before acquiring or processing vibration data would have predictably reduced false positives and avoided unnecessary processing outside the relevant speed range.
It also would have been obvious to derive a dominant vibration frequency only after an initial classified vibration level indicates a possible anomaly. Reference 2 teaches trigger-event processing and threshold/pattern-based classification. Reference 3 teaches more detailed predominant-frequency analysis using QPLL device 111.2 and comparison device 111.3. A person of ordinary skill in the art would have had reason to use a first-stage vibration classification as a screen and then apply the more detailed dominant-frequency analysis of Reference 3 only to vibration measurements that indicate a potential anomaly. This staged diagnostic approach is a predictable implementation that reduces unnecessary processing and focuses frequency analysis on vibration events likely to be relevant.
Applicant’s argument that no single reference discloses the exact claimed sequence is not persuasive because the rejection is based on the combined teachings of the references. The combined system uses Reference 1’s railcar sensor/processor platform, Reference 2’s speed-dependent threshold and pattern classification, Reference 3’s dominant-frequency comparison, and Reference 4’s labeled vibration fault states.
Applicant further argues that the references fail to teach identifying the anomaly based on both the dominant-frequency comparison and the reference classified vibration level.
This argument is not persuasive.
The claims do not require a specific mathematical fusion algorithm, weighting function, simultaneous calculation, single neural network, or particular form of data integration. The claims broadly recite identifying the anomaly “based on the comparison and a reference classified vibration level.” This language encompasses a diagnostic process in which a dominant-frequency comparison is used to identify or confirm wheel/wheelset/bearing-related vibration behavior, and a reference classified vibration level is used to distinguish a vehicle-side anomaly from a track-side anomaly.
Reference 3 teaches the frequency-comparison aspect. Reference 3 identifies a potentially damaging wheelset vibration state by analyzing vibration signal VS, identifying predominant vibration component PVC and frequency PVCF, and comparing the frequency PVCF against stored potentially damaging resonant frequencies PDRF using comparison device 111.3 and frequency-difference threshold FDIFT. Reference 3 also compares amplitude information AI with threshold amplitude information TAI. Reference 3 associates the damaging vibration condition with wheelset 106, wheels 106.1, wheelset shaft 106.2, and the press fit between the wheels and shaft.
Reference 2 teaches the reference classified vibration level aspect. Reference 2 teaches comparing acceleration/vibration levels from different sensor/control unit packages 5 and from other units or trains. Reference 2 teaches that if other units sense comparable acceleration levels at the same geo-positional point, this is a strong indication of a rail defect. Reference 2 also teaches using stored wagon history and comparable acceleration levels to distinguish wheel flats from rail defects.
Reference 1 provides the railcar monitoring architecture and teaches both vehicle-side conditions, including bearing faults and truck hunting, and track-side conditions, including track damage. Accordingly, the combined system identifies the anomaly based on both inputs: Reference 3’s dominant-frequency comparison indicates whether the vibration event corresponds to a wheelset/wheel/bearing vibration state, while Reference 2’s reference classified vibration level indicates whether comparable vibration behavior appears at the same location or across sensors, supporting identification of a track anomaly. The identification decision is therefore based on both the comparison and the reference classified vibration level, as claimed.
Applicant’s assertion that the references merely disclose separate analysis techniques is not persuasive. Reference 1, Reference 2, and Reference 3 all address railway or railcar vibration/acceleration anomaly detection. Combining a frequency-based wheelset diagnostic input with a reference-level track-discrimination input would have predictably improved classification accuracy and reduced false positives between wheel/bearing defects and fixed track defects.
Applicant further argues that the amended claims are patentable because the prior art does not teach identifying whether the anomaly exists within either or both of the railway vehicle wheel system and the track.
This argument is not persuasive.
Reference 1 teaches detection of both railcar component anomalies and track-related anomalies. Reference 1 teaches bearing fault detection from dynamic bearing data obtained by sensor 20 and also teaches track damage detection using dynamic data, probability density function, kurtosis, and impulsivity. Reference 1 also teaches truck hunting detection using FFT-based processing of dynamic data.
Reference 2 teaches distinguishing between wheel-related and track-related defects. Reference 2 teaches detecting wheel flats using acceleration data from sensor/control unit package 5 associated with bogie 1, wheel pairs 2a and 2b, and axles 3a and 3b. Reference 2 further teaches detecting rail defects and sun kinks, including by using comparable acceleration levels from other units at the same geo-positional point. Reference 2 also teaches detecting severe suspension defects.
Reference 3 teaches wheelset-related damaging vibration involving wheelset 106, wheels 106.1, and wheelset shaft 106.2. Reference 3 therefore supports the wheel-system side of the claimed anomaly identification.
Thus, the applied combination teaches identifying whether an anomaly is associated with the wheel system, the track, or both. Reference 1 supplies the railcar and track anomaly monitoring platform, Reference 2 supplies comparative railway defect classification and rail/wheel discrimination, and Reference 3 supplies dominant-frequency wheelset analysis.
Applicant further argues that the amended claims are patentable because the cited references do not teach mitigation action in response to the claimed anomaly identification.
This argument is not persuasive.
Reference 1 teaches alarms, alerts, and reporting to a customer or remote monitoring station so that maintenance may be scheduled or performed. Reference 2 teaches warning actions, including warning the train driver or dispatcher, identifying dangerous conditions, and providing action levels associated with maintenance or imminent danger. Reference 3 teaches active control mitigation in response to a damaging vibration state, including motor control signal MCS sent to motor controller 112 and traction converter device 108 to counteract the damaging vibration condition. These teachings correspond to performing a mitigation action in response to identifying the anomaly.
The claims do not require a particular mitigation action. The claims broadly recite “perform a mitigation action.” The alarms, warnings, maintenance indications, and control actions of References 1-3 fall within this broad limitation.
RESPONSE SPECIFIC TO AMENDED CLAIMS
Claims 1 and 14 were amended to add that the classified vibration level includes one or more labels indicating one or more characteristic features of the vibration measurement. This amendment does not overcome the rejection. Reference 4 teaches vibration-measurement-based digital twin processing and labeled bearing vibration fault level states, including normal, suboptimal, critical, and alarm. References 1 and 2 also teach classified railway vibration/acceleration condition states, including bearing fault, track damage, truck hunting, wheel flat, rail defect, sun kink, and suspension defect conditions.
Claims 2 and 15 were amended to clarify that determining whether the classified vibration level exceeds a predetermined vibration level threshold is a separate determination rather than part of the dominant-frequency comparison. This amendment does not overcome the rejection. Reference 2 teaches comparing measured acceleration/vibration levels with predefined levels, thresholds, criteria, and intervals. Reference 3 teaches comparing amplitude information AI with threshold amplitude information TAI. The amended limitation remains taught or rendered obvious by the applied references.
Claims 5 and 18 were amended to correct typographical and grammatical issues. These amendments do not substantively distinguish the claims from the prior art. Reference 1 teaches bearing fault detection and bearing damage indicators. Reference 2 teaches wheel flat detection. Reference 3 teaches damaging wheelset vibration involving wheels 106.1, wheelset shaft 106.2, and the press fit between the wheels and shaft. The claimed wheel deformation, loose wheel, worn bearing, loose bearing, and broken bearing remain taught or rendered obvious.
Claims 6 and 19 were amended to recite determining whether the classified vibration level and the reference classified vibration level match a predetermined vibration level pattern. This amendment does not overcome the rejection. Reference 2 teaches predefined acceleration patterns, predefined thresholds, predefined levels, and multi-source comparison of comparable acceleration/vibration levels. Reference 2 also teaches using comparable levels from other sensor units at the same geo-positional point to indicate a rail defect. Thus, Reference 2 teaches matching or comparing vibration/acceleration data and reference data to a predetermined pattern or criterion.
Claim 9 was amended to recite that the vibration sensor measures vibrations in a vertical direction and the second vibration sensor measures vibrations in a horizontal direction with respect to the wheel axle. This amendment does not overcome the rejection. Reference 2 teaches three-axis acceleration sensing, including vertical, longitudinal, and lateral acceleration measurements. Implementing these directional measurements using separate vertical and horizontal vibration sensors would have been an obvious design choice for obtaining direction-specific vibration information.
Claim 10 was amended to identify a horizontal geometric defect in response to the classified vibration level and reference classified vibration level matching a predetermined vibration level pattern. This amendment does not overcome the rejection. Reference 2 teaches detecting sun kinks or rail buckling from lateral acceleration above a threshold in several railroad wagons. Lateral acceleration corresponds to horizontal vibration, and a sun kink or rail buckling corresponds to a horizontal geometric track defect.
Claim 13 was amended to remove redundant wording. This amendment does not substantively distinguish the claim from the prior art. Reference 1 teaches vehicle-speed determination. Reference 2 teaches GPS time-velocity-position information and speed-dependent railway events. References 2 and 3 teach identifying anomalies based on operating condition, reference/comparable vibration levels, and frequency-related vibration information.
Claim 20 remains substantively unpatentable for the same reasons as claim 13 and independent claim 14.
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JASON C SMITH whose telephone number is (703)756-4641. The examiner can normally be reached Monday - Friday 8:30 AM - 5:00 PM.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Joseph Morano can be reached at (571) 272-6684. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/Jason C Smith/ Primary Examiner, Art Unit 3615
Prosecution Insights