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 .
Response to Arguments
Applicant’s arguments, see applicant’s remarks, filed 5/12/26 with regard to the
35 USC § 101 rejection, with respect to claims 15, 17, 19 and 29 have been fully considered and are persuasive. The 35 USC § 101 rejection of the claims has been withdrawn.
Applicant's arguments filed 5/12/26, with regard to the 35 USC § 103 have been
fully considered but they are not persuasive.
The applicant argues that neither Atlas nor Wegerich teaches the limitation “monitor vibrations of a target asset at the resonance frequencies with a machine learning model trained to generate estimated values at the resonance frequencies that are consistent with the reference asset”
The examiner respectfully disagrees with the applicant’s argument because the rejection does not rely upon Wegerich as teaching resonance-frequency identification. Rather, Atlas teaches identifying resonance frequencies of a reference asset by analyzing vibration responses to swept excitation using cross power spectral densities (see ¶¶ [0058] and [0066]-[0067]). The rejection relies on Wegerich for its teaching of a multivariate state-estimation model trained using data from normal or acceptable operation to generate estimated values representative of expected system behavior and to compare monitored values against such estimated values to detect abnormal conditions (see ¶¶ [0018] and [0039]-[0040]).
When the teachings of Wegerich are applied to the resonance-frequency vibration measurements identified by Atlas, the state-estimation model of Wegerich necessarily monitors vibration values at the identified resonance frequencies and generates corresponding estimated values representative of expected operation. The claim does not require the model itself to identify the resonance frequencies; rather, it requires monitoring vibrations at the resonance frequencies with a model trained to generate estimated values at those frequencies that are consistent with the reference asset. The combination of Atlas and Wegerich teaches this limitation because Atlas supplies the resonance-frequency features to be monitored, while Wegerich supplies the trained multivariate estimation model that generates expected values for the monitored features based on learned relationships from reference or normal operation.
Accordingly, the rejection does articulate and provide evidence that the combination of Atlas and Wegerich teaches or at least renders obvious “monitor vibrations of a target asset at the resonance frequencies with a machine learning model trained to generate estimated values at the resonance frequencies that are consistent with the reference asset”.
The applicant argues that identifying resonance frequencies as monitorable parameters does not teach or suggest monitoring vibrations at the resonance frequencies with a machine learning model trained to generate estimated values at the resonance frequencies, as specifically recited in the claims. The cited rationale also does not explain why one of ordinary skill in the art would have modified Atlas to generate predicted resonance-frequency vibration values using machine learning, rather than using the direct spectral and modal analysis techniques expressly disclosed by Atlas.
The examiner clarifies that the rejection does not equate these concepts. Instead, the resonance frequencies identified by Atlas constitute the vibration features or parameters that are subsequently monitored, and Wegerich teaches applying a trained state-estimation model to monitored parameters to generate corresponding estimated values. A person of ordinary skill in the art would have understood that once the resonance frequencies of interest are identified by the methodology of Atlas, those frequencies become the monitored variables supplied to the state-estimation model of Wegerich. The model of Wegerich would therefore generate estimated values for vibration measurements at the identified resonance frequencies based on learned relationships derived from reference or normal operating data.
Accordingly, the rejection does not rely on mere identification of resonance frequencies to satisfy the claimed machine-learning limitation. Rather, the combination teaches (1) identifying resonance frequencies to be monitored (Atlas) and (2) monitoring those identified parameters with trained multivariate estimation model that generates expected values for the monitored parameters (Wegerich). Therefore, the combined teaching of Atlas and Wegerich teach or at least render obvious “monitor vibrations of a target asset at the resonance frequencies with a machine learning model trained to generate estimated values at the resonance frequencies that are consistent with the reference asset” as recited in the claim.
Under 35 USC § 101
Although claims 1, 8 and 15 recite mathematical operations and data analysis,
claims 1, 8 and 15 also recite additional elements such as “record vibrations of a reference asset while the reference asset is operated based on a test pattern that sweeps over a range of workload for the reference asset”. The inclusion of these additional elements integrates the identified judicial exception into a practical application that effects a meaningful way beyond generally linking the use of the judicial exception to a particular technological environment, such that the claim as a whole is more than a drafting effort designed to monopolize the exception. Therefore, claims 1-20 are considered to be eligible under 35 USC 101. Accordingly, claims 1-20 are eligible under 35 U.S.C. § 101.
Examiner reminds to the Applicant that during patent examination, the pending
claims must be given the broadest reasonable interpretation consistent with the specification. Under a broadest reasonable interpretation (BRI), words of the claim must be given their plain meaning, unless such meaning is inconsistent with the specification. The plain meaning of a term means the ordinary and customary meaning given to the term by those of ordinary skill in the art at the relevant time. See MPEP 2111.01. Moreover, although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Claim Rejections - 35 USC § 103
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 1-10, 12-18 and 20 are rejected under 35 U.S.C. 103 as being
unpatentable over Atlas et al. (Pub. No. US 2010/0161254) (hereinafter Atlas) in view of Wegerich et al. (Pub. No. US 2006/0036403) (hereinafter Wegerich).
As per claims 1, 8 and 15, Atlas teaches record vibrations of a reference asset while the reference asset is operated based on a test pattern that sweeps over a range of workload for the reference asset (see ¶¶ [0066]-[0067] and Fig. 7A-7B, i.e., exciting a structure using a swept excitation (chirp vibration) and recording vibration responses with a plurality accelerometers during sweep); determine cross power spectral densities between the recorded vibrations and the test pattern at intervals to identify resonance frequencies of the reference asset (see ¶¶ [0067], [0079] and Fig. 11-13, i.e., determining resonance frequencies from spectral/cross-spectral processing, including cross power spectral density plots showing resonances).
While Atlas discloses recording vibration responses of a reference asset while the asset undergoes swept excitation/chirp excitation across a frequency range and computing cross power spectral densities (CPSD) between excitation and measured vibration signals to identify resonance frequencies (see ¶¶ [0058], [0066]-[0067] and Figs. 11-13), Atlas does not disclose monitoring vibrations of a target asset at the resonance frequencies with a machine learning model trained to generate estimated values at the resonance frequencies that are consistent with the reference asset, wherein the machine learning model generates the estimated values for one of the resonance frequencies based on observed vibration values for one or more others of the resonance frequencies; nor does Atlas disclose detecting resonant vibration amplification based on a dissimilarity between vibration values for the target asset at the resonance frequencies and the estimated values; and generate an electronic alert that the target asset is undergoing the resonant vibration amplification.
Wegerich, however, discloses model-based monitoring in which a model generates estimates and residuals (representing dissimilarity between measured sensor values) and alerts are produced when residuals indicate abnormality, stating that “estimates of sensor values” are generated and “subtracted from the actual sensor values to provide residual signals” and that “for any senor where a decision is made that the residual is non-zero, an alert is generated” (see ¶ [0013]); Wegerich further explains that an empirical model “requires historic data from which to “learn” normal states of operation, in order to generate sensor estimates (see ¶¶ [0018], [0039]-[0040] and [0071]-[0074]). Wegerich further teaches a multivariate state-estimation model trained on normal/reference operation and generating estimated values for monitored variables based on relationships among other observed variables of the system (see ¶¶ [0018] and [0039]-[0040]). Under BRI, the claimed “machine learning model” reads on the multivariate state-estimation of Wegerich multivariate state-estimation model is a machine learning model. Likewise, the claimed “resonance frequencies” correspond to the vibration-frequency features identified by Atlas and multivariate state-estimation model necessary generates an estimated value for one or monitored variable using observed correlated values. It would have been obvious to one having ordinary skill in the art before the effective filling date of the claimed invention to apply the learned/model-based residual alerting of Wegerich to the resonance-frequency vibration features identified by Atlas, because resonance frequencies identified from swept excitation represent natural monitored parameters for condition monitoring, thereby enabling automated detection and alerting of abnormal resonant vibration amplification based on dissimilarity between measured vibration values and model-generated estimated values.
The examiner notes, for claim 8, that the chirp vibration/excitation is commonly applied over defined time intervals, which inherently establishes a duty-cycle for the excitation. Accordingly, sweeping a duty-cycle constitutes a predictable implementation choice when applying chirp-type swept excitation under broadest reasonable interpretation.
As per claims 2, 9 and 16, the combination of Atlas and Wegerich teach the system as stated above. Atlas further teaches controlling the operation of the reference asset to cause the reference asset to operate according to the test pattern, wherein the test pattern is a sine sweep of load that covers a range of operation for the reference asset, and wherein the sine sweep changes in periodicity linearly over the course of the sweep (see ¶ [0009], i.e., “a vibration source is employed for exciting the container to vibrate”, further Atlas explains that “instantaneous frequency and phase of the vibration source are tracked independently…the periodicity of the current waveform allows for a much more accurate estimate of the vibration source frequency and its phase” (see ¶ [0058]). Under the broadest reasonable interpretation, exciting an asset using a vibration source whose frequency and phase are tracked and managed necessarily implies controlling operation of the asset or its excitation system according to a defined test pattern.
Atlas further discloses swept excitation over a range of operation. Specifically, ¶ [0066] discloses that “each vibration scan included two stepped vibration chirps…one with a frequency increasing from about 20 Hz to about 90 Hz, and one with frequency increasing from about 85 Hz to about 135 Hz”, this disclosure directly corresponds to a test pattern that sweeps over a range of operation, where swept excitation is technically equivalent to a chirp or swept sine input.
While Atlas does not explicitly recite the limitation “linearly changing periodicity”, Atlas discloses the technical equivalent (see ¶ [0066], the frequency increases over time and explains in ¶ [0058] that the instantaneous frequency and phase are coherently tracked. (the examiner notes that a vibration chirp with increasing frequency over time, is by definition, a sine sweep whose periodicity changes continuously over the course of the sweep, where increasing frequency necessarily implies decreasing period). Under BRI, such continuous chirp excitation inherently includes linear or piecewise-linear changes in periodicity, which the conventional implementation of swept sine testing.
As per claims 3, 10 and 17, the combination of Atlas and Wegerich teach the system as stated above. Atlas further teaches subdivide the frequency spectrum of the cross power spectral densities into a plurality of bins (see ¶ [0067], “FIG. 11 illustrates offset plots 140 of the cross power spectral density (PSD) between all seven accelerometers and shows the location dependent resonances”, (the examiner notes that CPSD is by definition a frequency-domain representation sampled at discrete frequency points and under BRI, subdividing the frequency spectrum into bins is an inherent and standard representation of CPSD data (e.g., FFT frequency bins and a person of ordinary skill in the art understand CPSD data as already organized into frequency bins or resolvable frequency intervals); average a plurality of correlated frequencies in each bin over the intervals to produce averaged bins (see ¶ [0066], “each vibration scan included two stepped vibration chirps” and ¶ [0067] describes offset plots of cross PSDs showing resonances across different measurements, ¶ [0073] explains that the evaluation is performed in multiple stages and that mapping data derived from repeated measurements is used and also in ¶ [0067] the discussion of Fig. 11-13 demonstrates that multiple CPSDs are collected over repeated events, positions, or intervals, and resonances are identified based on consistent spectral behavior across those measurements. (the examiner notes that averaging correlated frequencies across intervals (e.g., sweeps or impacts) is a routine signal-processing technique used to improve signal-to-noise ratio and suppress noise when identifying resonances. Under BRI, correlating CPSDs across measurements reasonably encompasses averaging frequency components within bins over intervals); sample a time series from each averaged bin and determine amplitude for the bins from the sampled time series (see ¶ [0058], i.e., “coherent detection” is used, wherein the instantaneous frequency and phase of the vibration source are tracked to demodulate the received vibration signal, and “extremely accurate and low variability estimates of frequency peaks are needed”; ¶ [0072] describes that “sustained resonances continued for more than five seconds”, evidencing that vibration behavior at identified resonance frequencies is tracked over time, sampling (i.e., demodulating) at time series associated with each frequency bin and determining its amplitude (see ¶ [0058]) “The phase from this waveform can then be used to demodulate a received acoustic signal. Vibration source frequency is approximately related to the voltage applied to the source, yet the corresponding periodicity of the current waveform allows for a much more accurate estimate of the vibration source frequency and its phase”); and select the bins which are in a top range of amplitude to be the resonance frequencies (see ¶ [0067], Fig. 11-13, show resonances that are visually and analytically distinguished from surrounding frequencies (the examiner notes that selecting frequency bins with the largest amplitudes or peaks is the standard technique of identifying resonant frequencies). Under BRI, “selecting bins which are in a top range of amplitude” directly maps identifying dominant resonance peaks in PSD or CPSD plots, as explicitly shown and described in Atlas).
As per claim 4, the combination of Atlas and Wegerich teach the system as stated above. Wegerich further teaches (i) trigger, in response to the electronic alert, an adjustment to workload on the target asset to reduce stimulus to the resonance frequencies; or (ii) generate, in response to the electronic alert, instructions for service to remediate the resonant vibration amplification (see ¶¶ [0012]-[0015] and [0038]), i.e., generating, in response to an alert, suggested investigative and repair steps for remediating a detected abnormal condition, which satisfies alternative (ii) under the broadest reasonable interpretation.
As per claim 5, the combination of Atlas and Wegerich teach the system as stated above. While Wegerich teaches detecting abnormal operation based on residuals between measured sensor values and model-estimated values and determining whether those residuals satisfy predefined conditions such as thresholds or statistical tests (see ¶¶ [0013] and [0071]-[0074]) and further teaches evaluating multivariate residual vectors for deviation from expected behavior (see ¶¶ [0085]-[0087]). Wegerich does not explicitly describe a kiviat plot and an annular geometry. Representing such vectors using radar/kiviat plot and assessing deviation via radial or annular measure constitutes a known and conventional visualization and mathematical interpretation of the disclosed residual magnitude. It would have been within the level of ordinary skill in the art to apply appropriate mathematical representations and charting techniques to organize multivariate residual data and facilitate more structured and accurate anomaly detection decisions, particularly in condition monitoring diagnostics systems.
As per claim 6, the combination of Atlas and Wegerich teach the system as stated above. Wegerich further teaches that detecting resonant vibration amplification is based on detecting an anomaly in a series of residuals between the vibration values and the estimated values for one of the resonance frequencies (see ¶¶ [0013] and [0071]-[0074], i.e., detecting resonant vibration amplification by evaluating a time series of residuals between measured vibration values and model-estimated values at an identified resonance frequency behavior).
As per claim 7, the combination of Atlas and Wegerich teach the system as stated above. Wegerich further teaches that the machine learning model implements a multivariate state estimation technique (see ¶¶ [0006], [0013], [0017], and [0046]-[0047], i.e., machine-learning-based multivariate state estimation technique, wherein a model jointly processes multiple sensor variables representing a system state and generates estimated values mased on learned relationships among the variables; “Further, each snapshot can be thought of as a "state" of the underlying system. Thus, collections of such snapshots preferably represent a plurality of states of the system. As described above, any previously collected sensor data can be filtered to produce a smaller "training" subset (the reference set D) that characterizes all states that the system takes on while operating "normally" or "acceptably" or "preferably").
As per claims 12 and 18, the combination of Atlas and Wegerich teach the system as stated above. Wegerich further teaches triggering, in response to the electronic alert, an adjustment to the workload on the target asset to reduce stimulus to the resonance frequencies (see ¶ [0015], i.e., providing alert outputs to control system for automatic machine control).
As per claims 14 and 20, the combination of Atlas and Wegerich teach the system as stated above. Wegerich further teaches that detecting resonant vibration amplification further comprises: generating a series of residuals between monitored values of the monitored vibrations and the estimated values for one or more of the resonance frequencies; and detecting an anomaly in the series of residuals with a sequential probability ratio test (see ¶¶ [0008], [0013] and [0071]-[0079]).
Claim 11 is rejected under 35 U.S.C. 103 as being unpatentable over Atlas
in view of Wegerich and further in view of Wang (NPL: Dissertation “Geometric Fault Detection Using 3D Kiviat Plots and their Applications” (Year 2017).
As per claim 11, the combination of Atlas and Wegerich teach the system as stated above except for generating a kiviat tube of observations of the monitored vibrations; and generating a graphical user interface that displays the kiviat tube, wherein the electronic alert causes an indication that resonant vibration amplification is detected to be displayed.
However, Wang discloses representing multivariate, time-evolving sensor observations using high-dimensional, time-explicit 3D Kiviat plots (i.e., Kiviat tubes) and detecting faults based on geometric deviation of the plotted data over time from a normal operating region (see pages 21-22 and Fig. 2.9 a-d). It would have been obvious to one having ordinary skill in the art before the effective filling date of the claimed invention to incorporate Wang’s teaching into the combination of Atlas and Wegerich’s teaching because Wang teaches that faults are detected based on geometric deviation of multivariate data from a normal operating region in time-explicit Kiviat plots, thereby providing an interpretable and early indication of abnormal system behavior to support fault detection and alerting.
Claim 19 is rejected under 35 U.S.C. 103 as being unpatentable over Atlas
in view of Wegerich and further in view of Wetherbee et al. (Pub. No. US 2021/0270884) (hereinafter Wetherbee).
As per claim 19, the combination of Atlas and Wegerich teach the system as
stated above except for plotting the monitored values and the estimated values into a kiviat surface; normalizing the kiviat surface to a unit circle of the estimated values; generating an annular residual between the normalized monitored values and normalized estimated values; and comparing the annular residual to a threshold for detecting the resonant vibration amplification.
Wetherbee, however, teaches accessing monitored values and corresponding estimated values for an observation, plotting the values into Kiviat (radar) representation, normalizing the representation to a unit circle based on reference or estimated values, generating an annular residual between normalized monitored and estimated values, comparing the annular residual to a threshold to detect anomalous behavior, and displaying the result via a graphical user interface (see ¶¶ [0058]-[0069] and [0086]-[0090]). It would have been obvious to one having ordinary skill in the art before the effective filling date of the claimed invention to apply the Kiviat-based residual computation and the visualization of Wetherbee to the combination of Atlas and Wegerich teaching because Wetherbee provides a known and effective geometric technique for organizing, normalizing, and evaluating multivariate residuals over time, thereby improving interpretability, sensitivity, and operator awareness of resonant variation amplification, resulting in more accurate and reliable analysis of the system behavior.
Conclusion
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.
Contact information
Any inquiry concerning this communication or earlier communications from the
examiner should be directed to MOHAMED CHARIOUI whose telephone number is (571)272-2213. The examiner can normally be reached Monday through Friday, from 9 am to 6 pm.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Andrew Schechter can be reached on (571) 272-2302. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
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.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free).
Mohamed Charioui
/MOHAMED CHARIOUI/Primary Examiner, Art Unit 2857
Prosecution Insights