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 .
Claim Rejections - 35 USC § 112
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefore, subject to the conditions and requirements of this title.
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 1-20 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.
Claims 1, 11, and 16 recite “… and that causes the full-winding rate not meeting not to meet the requirement …”. The phrase “not meeting not to meet” is grammatically incorrect. It is unclear whether “not meeting” is acting as a noun, a typo, or a negative modifier. For the purpose of examination, we consider the phrase to be “to fail to meet”.
Claims 2-10, 12-15, and 17-20 are rejected because they depend from claims 1, 11, and 16, respectively.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-8, and 10- 20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to abstract idea without significantly more. The claim(s) recite(s) abstract idea as discussed below. This judicial exception is not integrated into a practical application because of the reasons discussed below. The claim(s) does/do not include additional elements that are sufficient to amount to significantly more than the judicial exception because of the reasons discussed below.
Step 1 - Statutory Category: Step 1 of the 2019 Guidance requires the examiner to determine if the claims are to one of the statutory categories of invention. Applied to the present application, claims 1-20 are directed to a process (method) and a machine (device/system) and manufacture (non-transitory computer-readable storage medium). Accordingly, claims 1-19 fall within at least one of the four statutory categories of invention (process, machine, manufacture, or composition of matter) under 35 U.S.C. 101.
Claim 1 is reproduced below with the abstract idea underlined.
Claim 1: A parameter processing method, comprising: determining T full-winding rates of a yarn spindle wound by a winding machine, wherein the t-th full-winding rate in the T full-winding rates is obtained based on N first characteristic parameters collected at the t-th moment, and a first characteristic parameter in the N first characteristic parameters is a characteristic parameter that is able to influence on a full-winding rate of the yarn spindle wound by the winding machine; T is a positive integer more than or equal to 1; t is a positive integer more than or equal to 1 and less than or equal to T; and N is a positive integer more than or equal to 1; and determining a candidate characteristic parameter set under the condition where it is determined that M full-winding rates in the T full-winding rates do not meet a requirement of a preset full-winding rate, wherein a candidate characteristic parameter included in the candidate characteristic parameter set is the characteristic parameter that is selected from N first characteristic parameters corresponding to a full-winding rate not meeting the requirement of the preset full-winding rate and that causes the full-winding rate to fail to meet the requirement of the preset full-winding rate; and M is a positive integer more than or equal to 1 and less than or equal to T.
Under Step 2A, Prong 1, Claim 1’s underlined limitations recite comparing data to a threshold (when it is determined that M full-winding rates do not meet a requirement of a preset full winding rate) and identifying a subset (determining a candidate characteristic parameter set) of setting that caused the failure. These steps can technically be performed through human mental logic or simple mathematical calculations (for instance, one could write down 5 hourly intervals (T=5), look at the settings (N parameters), notice that hour 3 and 4 failed to meet a 95% target, and circle “oil pressure” as the likely culprit). Accordingly, claim 1 recites the abstract idea of mathematical calculations and mental process.
Step 2A, Prong 2: examiner needs to determine if the claim(s) recite additional elements that integrate the exception into a practical application of the exception. The additional elements in the claim have been left in normal font. Claim 1 does not integrate the judicial exception into a practical application because of the following reasons:
Claim 1 additional elements recite gathering data, determining T full-winding rates based on N first characteristic parameters collected at the t-th moment, that is merely gathering the information used in the subsequent analysis and it is considered insignificant extra-solution activity. Reciting yarn spindle, winding machines are characterized as field of use limitations. Accordingly, the additional elements, individually and in combination, do not integrate the abstract idea into a practical application.
Claims 11 and 16: the analysis with respect to claim 1 applies analogously to claims 11 and 16. Claim 11 merely recasts the limitations of claim 1 as a system comprising processor-executable instructions. The recited processor, memory, and instructions merely perform the same data gathering, mathematical analysis, and determination recited in claim 1 using generic computer components.
Step 2A, Prong 1 for dependent claims: claims 2-10 depend from claim 1, claims 12-15 depends from claim 11, and claims 17-10 depends from claim 16 and share the same abstract idea as claims 1, 11, and 16, respectively. Claims 5-7, 15 and 20 further recite statistical filtering and linear algebra matrix manipulation and claims 8 and 10 recite virtual iterative optimization loops, the system repeats the simulation adjustments and rejudge the results which amount to mathematical optimization process.
Accordingly, claims 2-10, 12-15, and 17-20 recite the abstract idea of mathematical calculations and mental process.
Step 2A, Prong 2 for dependent claims: claims 2, 12, and 17 additional elements recite limiting data collection to a preset time interval over T different moments triggered by a first detection instruction which is considered insignificant extra-solution activity. Claims 3-4, 13-14, and 18-19 recite a target winding prediction model, which is treated as a tool to do calculations faster. It does not improve the operation of the computer itself, nor does it physically change the winding machine. Claims 5-8, 10, 15 and 20 do not recite any additional elements beyond the abstract idea.
Claim 9 recites generating an instruction to physically adjust each candidate characteristic parameter in the candidate characteristic parameter set of the winding machine according to the target adjustment mode, which solves the real-world technological problem in textile manufacturing by avoiding manual, experience-based machine troubleshooting and replacing it with autonomous machine calibration. Accordingly, claim 9 integrates the abstract idea into a practical application.
Accordingly, claims 2-8, 10, 12-15, and 17-20 do not integrate the abstract idea into a practical application.
Step 2B: Claims 1-8, and 10- 20: the additional elements, considered individually and in combination, do not amount to significantly more than the abstract idea for the same reasons set forth with respect to Step 2A, Prong 2.
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1-2, 11-12, and 16-17 are rejected under 35 U.S.C. 102 (a)(1) and 102 (a)(2) as being anticipated by Samoto (JP H07133534 A), see the English Translation.
Regarding claim 1, Samoto teaches a parameter processing method (a management system for winding machine, ¶ [1]), comprising: determining T full-winding rates of a yarn spindle wound by a winding machine (calculates the full ball forming rate for each spindle over time, ¶ [5]. Figure 7 shows the determined full ball rate for T=30, 30 different days, from October 12 to November 10) , wherein the t-th full-winding rate (individual daily dates points on the x-axis, figure 7) in the T full-winding rates (full ball rate versus dates on horizontal axis of figure 7, the chronological sequence of intervals over which the winder spindle performance is tracked, figure 7) is obtained based on N first characteristic parameters collected at the t-th moment (figure 4 shows that characteristic parameters like yarn speed, tensions and temperatures are associated with the specific time. By installing … a yarn breakage sensor (checks the yarn breakage status that is a characteristic parameter) to calculate the full-bodied yarn formation rate, long-term data on … full-bodied yarn formation rate can be used to estimate the areas that need maintenance. ¶ [18]), and a first characteristic parameter in the N first characteristic parameters is a characteristic parameter that is able to influence on a full-winding rate of the yarn spindle wound by the winding machine (… it shows the ratio of the number of packages that reached full pack without any string breakage during the process, relative to the total number of production packages, (the full package formation rate is based on the absence of the yarn breaks), ¶ [21]); T is a positive integer more than or equal to 1 (T =30 , for 30 different days (T is total duration), figure 7); t is a positive integer more than or equal to 1 and less than or equal to T (t is specific timestamped entry. t=1 corresponds to the first plotted day, Oct 12, t=15 corresponds to the middle of the timeline, Oct 26, t=30 corresponds to the final day, Nov 1. This shows that t lies within the range of
1
≤
t
≤
T
, figure 7 and ¶ [20]); and N is a positive integer more than or equal to 1 (figure 4 discloses 5 characteristic parameters, (T2 tension, Yarn speed, Velocity ratio, Temp H1, and Temp H2 ), so N=5 which is a positive integer greater than 1, figure 4).
Samoto further teaches determining a candidate characteristic parameter set (T2 defect report including: T2 tension, Yarn speed, Velocity ratio, Temp H1, and Temp H2 , ¶ [18]. (Host computer 54 isolates the abnormal parameters when the full ball formation rate drops, ¶ [6 & 15])) under the condition where it is determined that M full-winding rates in the T full-winding rates do not meet a requirement of a preset full-winding rate (the host computer calculates and tracks the ball formation rate continuously. As shown in figure 7, if this rate is monitored over a predetermined period falls below the lower limit of 90%, it acts as the automatic trigger indicating that some abnormality that causes yarn breakage in the specific spindle is beginning to occur, figure 7 and ¶ [6 & 20-21]. M represents the number of days (or moments t) during that month-long period where the full ball formation rate dipped below that 90% requirement, figure 7), wherein a candidate characteristic parameter included in the candidate characteristic parameter set is the characteristic parameter that is selected from N first characteristic parameters (extracting the monitored process variables (tension, speed, VR, H1, H2) for the specific failing spindle, figure 4 & ¶ [18]) corresponding to a full-winding rate not meeting the requirement of the preset full-winding rate (initiated when full ball (package) formation rate falls below the lower limit of 90%, figure 7 &¶ [6 & 21]) and that causes the full-winding rate to fail to meet the requirement of the preset full-winding rate (this (knowing the parameters for the specific spindle at specific time when the fall ball rate dropped below lower limit) allows you to estimate the cause of the problem, ¶ [18]) ; and M is a positive integer more than or equal to 1 and less than or equal to T (figure 7 shows that the rate fall below the lower limit passing rate, suggesting that some machine trouble has occurred from Oct 26 to Nov 3, so total failing moments M equals to 9 days and consequently
1
≤
9
≤
30
or
1
≤
M
≤
T
,
figure 7) .
Claims 11 and 16 recite an electronic device and computer-readable medium comprising a physical processor and memory to execute parameter processing steps of claim 1. Samoto discloses a management system comprising a Host computer 54, sub computers (44, 52), and personal computers (55, 56). A physical computer cannot read program instructions or perform mathematical calculations without inherently possessing a physical processor and a physical memory to store those instructions. Because these physical hardware components are technically necessary for the operation of Samoto’s disclosed management system, Samoto inherently discloses the processor and memory of claims 11 and 16. Claims 11 and 16 are rejected for the same reasons as set forth with respect to rejection of claim 1.
Regarding claim 2, Samoto teaches the method of claim 1 as set forth with respect to rejection of claim 1.
Samoto further teaches wherein determining the T full-winding rates of the yarn spindle wound by the winding machine comprises: performing a winding prediction step in response to a first detection instruction (a winding event acts as the automatic instruction to transmit and log the spindle’s parameters to the host computer 54, ¶ [19]. Alternatively, an abnormality limit breach acts as a detection trigger to capture parameter snapshots, ¶ [18]); and performing a next winding prediction step for T times, under the condition where a preset time interval is reached, to obtain full-winding rates of the winding machine corresponding to T different moments (system tracks and updates the full package formation rate temporally over consecutive intervals (days) across predetermined period (T=30 times (Oct 12 to Nov 10)),figure 7); wherein the t-th winding prediction step comprises: obtaining N first characteristic parameters of the winding machine at the t-th moment (system retrieves N=5 parameters (tension, speed, VR, H1, H2) captured by sensor 27 and 28 at a timestamped moment t. figure 4 visually maps this parameter snapshot at a specific moment, figure 4 and ¶ [18]); and obtaining through estimation the t-th full-winding rate of the yarn spindle wound by the winding machine based on the N first characteristic parameters at the t-th moment (calculates the full ball rate based on occurrence or absence of yarn breaks caused by parameter abnormalities, ¶ [21]).
Claims 12 and 17 are rejected for the same reasons as set forth with respect to rejection of claim 11, 16 and 2.
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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or non-obviousness.
Claims 3, 8, 10, 13, and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Samoto (JP H07133534 A), see the English Translation, and further in view of Cheng (CN-114611235-A), see the English Translation.
Regarding claim 3, Samoto teaches the method of claim 2 as set forth with respect to rejection of claim 2.
Samoto further teaches obtaining through estimation the t-th full-winding rate of the yarn spindle wound by the winding machine based on the N first characteristic parameters at the t-th moment (system retrieves N=5 parameters (tension, speed, VR, H1, H2) captured by sensor 27 and 28 at a timestamped moment t. figure 4 visually maps this parameter snapshot at a specific moment, figure 4 and ¶ [18] and calculates the full ball rate based on occurrence or absence of yarn breaks caused by parameter abnormalities, ¶ [21]). Samato teaches the first result corresponding to the t-th moment includes a first value which represents a full-winding rate of the yarn spindle wound by the winding machine (calculates a value for each t that represents full-ball rate for that timestamp, figure 7).
However, Samoto does not teach utilizing any winding prediction model. Samoto does not teach inputting the N first characteristic parameters at the t-th moment into a target winding prediction model, to at least obtain a first prediction result output from a first branch of the target winding prediction model and corresponding to the t-th moment, wherein the first branch of the target winding prediction model is configured to predict the full-winding rate of the yarn spindle wound by the winding machine; the first prediction result corresponding to the t-th moment includes a first value which represents a predicted full-winding rate of the yarn spindle wound by the winding machine.
Cheng teaches inputting the N first characteristic parameters at the t-th moment into a target winding prediction model (the fault neural network prediction model uses LSTM neural network that takes time series x(t) of the equipment fault characteristic parameters Nc as input vector to predict winding failures, ¶ [49]) to at least obtain a first prediction result output from a first branch of the target winding prediction model and corresponding to the t-th moment (utilizes a dual-prediction pipeline where a neural network prediction branch (Vi,j(t)) and a digital twin branch (Wi,j(t)) run in parallel to output predictions at moment t, ¶ [50]) wherein the first branch of the target winding prediction model is configured to predict the winding [quality] of the yarn spindle wound by the winding machine (a quality traceability module that takes winding process parameters to predict winding quality, ¶ [39]).
It would have been obvious to one ordinary skill in the art before the effective filling date of the claimed invention to modify Samoto’s monitoring system by incorporating Cheng’s prediction model and train the model to predict full-winding rate. Because a full-winding rate is the fundamental mathematical expression of winding quality, utilizing a known neural network to predict this specific parameter represents an adaptation of Cheng’s predictive software to Samoto’s established textile metrics, yielding the predictable result of feedforward spindle tracking. It should also be noted that Samoto’s figure 6 and 7 provide visual proof that winding quality and full-winding rate are perfectly correlated and functionally interchangeable. The combined system would predict the failures before they physically occur, thereby enhancing the efficiency of equipment management.
Regarding claim 8, Samoto in view of Cheng teaches the method of claim 3 as set forth with respect to rejection of claim 3.
Samoto in view of Cheng further teaches determining the candidate characteristic parameter set ( Samoto, system retrieves N=5 parameters (tension, speed, VR, H1, H2) captured by sensor 27 and 28 at a timestamped moment t. figure 4 visually maps this parameter snapshot at a specific moment, figure 4 and ¶ [18] and calculates the full ball rate based on occurrence or absence of yarn breaks caused by parameter abnormalities, ¶ [21]).
However Samoto does not teach the method further comprises: performing simulation adjustment on each candidate characteristic parameter in the candidate characteristic parameter set; obtaining through re-estimation a new full-winding rate of the yarn spindle wound by the winding machine at least based on each candidate characteristic parameter after the simulation adjustment; and obtaining a target adjustment mode under the condition where the new full-winding rate of the yarn spindle wound by the winding machine meets the requirement of the preset full-winding rate.
Cheng teaches the method further comprises: performing simulation adjustment on each candidate characteristic parameter in the candidate characteristic parameter set (Cheng uses real-time data from the winding machines to update and run the digital twin model, then takes the output of these virtual simulations to optimize operational decisions on the physical workshop floor, ¶ [21]).
Cheng teaches obtaining through re-estimation (using the digital twin model to dynamically analyze and predict product parameters in virtual space, ¶ [39]) a new winding [quality] of the yarn spindle wound by the winding machine (predicts the winding quality, ¶ [39]) at least based on each candidate characteristic parameter after the simulation adjustment (updates the virtual model parameters to run new simulations, ¶ [21] and predict the winding quality using digital twin model simulation, ¶ [39]).
Cheng further teaches and obtaining a target adjustment mode (A fault maintenance service module, which provides corresponding maintenance plans to maintenance personnel based on fault diagnosis results, ¶ [36]) under the condition where the new winding [quality] of the yarn spindle wound by the winding machine meets the requirement of the preset winding [quality] (the simulation runs recursively. If a trial fails, it repeats, but once a configuration clears the evaluation standards, it is deemed feasible triggering the final adjustment outputs, ¶ [53]).
It would have been obvious to one ordinary skill in the art before the effective filling date of the claimed invention to utilize Cheng’s iterative optimization feedback gate to verify that simulated parameter changes will successfully restore the full-winding rate before exit. Employing Cheng’s virtual feasibility loop to define the target physical adjustments once quality standards are met represents a standard engineering practice to prevent physical machine damage and yarn waste, yielding the expected result of a verified, safe parameter adjustment profile.
Regarding claim 10, Samoto in view of Cheng teaches the method of claim 8 as set forth with respect to rejection of claim 8.
Samoto in view of Cheng further teaches the method of claim 8, further comprising: under the condition where the new full-winding rate of the yarn spindle wound by the winding machine does not meet the requirement of the preset full-winding rate (Samoto, defines the lower limit pass rate for quality/full-ball rate, falling below this limit triggers the diagnostic event, figure 6 and 7. Cheng, the gating condition is when a simulated adjustment is deemed not feasible (meaning it fails to satisfy preset standards), ¶ [53]), performing the simulation adjustment on each candidate characteristic parameter in the candidate characteristic parameter set again to determine the new full-winding rate (Cheng, if a repair method is deemed infeasible, the system allows for the iterative selection of alternative repair methods for continued simulation and evaluation until a feasible method is found ¶ [53]), and judging whether the new full-winding rate meets the requirement of the preset full-winding rate (Samoto teaches full ball rate/quality requirements, figure 6 and 7. Cheng, the system inputs the newly modified parameters back into the digital twin model and recursively re-runs the simulation and quality prediction, ¶ [39]).
It would have been obvious to one ordinary skill in the art before the effective filling date of the claimed invention to build a recursive optimization loop as taught by Samoto in view of Cheng. If an initial adjustment proposal is modeled in virtual space and predicted to fail (below the quality threshold), it is logical to repeatedly replace the trial parameters and continue simulation until the threshold is cleared. This prevents pushing a flawed correction to the physical winder, ensuring continuous winding-line operation without human in the loop delay or physical thread waste.
Claim 13 and 18 are rejected for the same reasons as set forth with respect to rejection of claims 11, 16 and 3.
Claim 9 is rejected under 35 U.S.C. 103 as being unpatentable over Samoto (JP H07133534 A), see the English Translation, Cheng (CN-114611235-A), see the English Translation and further in view of Luan et al. (CN- 116631769-A) hereinafter Luan, see the English Translation.
Regarding claim 9, Samoto in view of Cheng teaches the method of claim 8 as set forth with respect to rejection of claim 8.
Samoto in view of Cheng further teaches the method of claim 8, further comprising: generating a first adjustment instruction based on the target adjustment mode (Cheng teaches a digital twin architecture that maps a physical winding workshop into a 3D virtual space and discloses the controlling the real with the virtual (figure 2 top of the page), where optimized operational and repair decisions generated through virtual model simulations are used to update real-world workshop equipment, figure 1 and figure 2)
Samoto in view of Cheng teaches that maintenance suggestions or repair strategies is presented to a human technician on a screen.
Samoto in view of Cheng doesn’t teach the first adjustment instruction is configured to instruct to physically adjust each candidate characteristic parameter in the candidate characteristic parameter set of the winding machine according to the target adjustment mode.
Luan teaches deploying its machine learning models directly onto a low-power embedded microcontroller (MCU) on the winding machine, ¶[12]. The winder-embedded MCU runs the prediction engine, calculates deviation values, and feeds it to the machine’s local PID circuits. The winder then automatically adjusts its own actuators according to the calculated u(t), ¶ [20-26].
It would have been obvious to one ordinary skill in the art before the effective filling date of the claimed invention to combine the virtual-to-physical digital twin architecture of Samoto in view of Cheng with the real-time embedded PID controllers of Luan to allow the winder to automatically execute the simulated adjustment instruction directly at the local PLC registers without human delay and create a self-healing winding line.
Claims 4, 14, and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Samoto (JP H07133534 A), see the English Translation, Cheng (CN-114611235-A), see the English Translation and further in view of Hu et al. (CN-113191824-A) hereinafter Hu, see the English Translation.
Regarding claim 4, Samoto in view of Cheng teaches the method of claim 3 as set forth with respect to rejection of claim 3.
Samoto in view of Cheng further teaches wherein the t-th winding prediction step further comprises: under the condition where the first prediction result which is output from the first branch of the target winding prediction model and corresponds to the t-th moment is obtained in the t-th winding prediction step (Cheng, utilizes a dual-prediction pipeline where a neural network prediction branch (Vi,j(t)) and a digital twin branch (Wi,j(t)) run in parallel to output predictions at moment t, ¶ [50]. The product quality traceability results are recorded in the storage module 108 and they can be used to trigger fault diagnosis module to analyze the physical sensors, figure 1 and ¶ [28 & 39]).
Samoto in view of Cheng further teaches obtaining a second prediction result output from a second branch of the target winding prediction model and corresponding to the t-th moment (the physical digital-twin simulation pipeline that outputs a simulation-driven value Wi,j(t) representing the state of the j-th sensor of the i-th machine at time t, ¶ [50]) wherein the second branch of the target winding prediction model is configured to predict a degree of influence of inputting each of the N first characteristic parameters at the t-th moment on the full-winding rate (Cheng, the subscript and the superscript indices of the time series input vector X that is fed into LSTM neural network represents the feature type and the temporal states, respectively. The network then outputs a vector Y where each element
y
i
t
represents the predicted value of the i-th feature type at future moment t,¶ [49]. Because Cheng’s LSTM is mathematically segregated by feature types (subscript i) the model calculates a distinct prediction stream for each individual parameter. Predicting the future state or deviation of each individual feature type is analogous to outputting the degree of influence of the specific parameter on the final quality failure.)
Samoto in view of Cheng teaches the second prediction result corresponding to the t-th moment includes N second values (Cheng, the LSTM neural network outputs
Y
=
y
1
t
,
y
2
t
,
…
y
N
c
t
, which is multidimensional vector of predicted parameters corresponding to target moment t, where
N
c
is the total number of physical feature types monitored, ¶ [102]), and the n-th element in the N second values (Cheng, the subscripts in LSTM represent the specific feature type, the i-th element
y
i
t
in output vector represents the future value of the i-th feature type, ¶ [49]) represents a degree of influence of inputting the n-th first characteristic parameter in the N first characteristic parameters at the t-th moment on the full-winding rate (Cheng, the system tracks physical sensor indices j and uses its product quality traceability services module to evaluate how these parameters predict winding quality, ¶ [39 & 49-50]).
Samoto in view of Cheng teaches the system tracks physical sensor indices j and uses its product quality traceability services module to evaluate how these parameters predict winding quality, (Cheng, ¶ [39 & 49-50]). However, Samoto in view of Cheng doesn’t teach the degree of influence associated with feature type.
Samoto in view of Cheng doesn’t teach the n-th element in the N second values represents a degree of influence of inputting the n-th first characteristic parameter in the N first characteristic parameters at the t-th moment on the full-winding rate.
Hu teaches training a feature analysis model respectively based on sample datasets to determine feature parameters used for representing the influence degree of each candidate feature on the product index, ¶ [65 & 132].
It would have been obvious to one ordinary skill in the art before the effective filling date of the claimed invention to configure Samoto in view of Cheng’s LSTM model to output its input weight parameters as degree of influence, as taught by Hu. It is well established in the art of computer science that neural networks mathematically process inputs by assigning and optimizing numerical weights. These weights represent the degree of influence that each input variable has on the final output. Doing so represents utilizing a known programming option to provide interpretable diagnostics, yielding the predictable result of identifying which running sensor is causing a winding quality drop.
Claims 14 and 19 are rejected for the same reasons as set forth with respect to rejection of claims 11, 16 and 4.
Claims 5, 6, 15, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Samoto (JP H07133534 A) in view of Cheng (CN-114611235-A), Hu et al. (CN-113191824-A) hereinafter Hu, see the English Translation, and further in view of von Grunigen et al. (US 20210342705 A1) hereinafter von Grunigen.
Regarding claim 5, Samoto in view of Cheng and Hu teaches the method of claim 4 as set forth with respect to rejection of claim 4.
Samoto in view of Cheng and Hu further teaches wherein determining the candidate characteristic parameter set (Samoto, T2 defect report including: T2 tension, Yarn speed, Velocity ratio, Temp H1, and Temp H2 , ¶ [18]. (Host computer 54 isolates the abnormal parameters when the full ball formation rate drops, ¶ [6 & 15])).
Samoto in view of Cheng and Hu teaches a conditional diagnostic filter to isolate the candidate characteristic parameter set in the specific moments (M) that the full ball rate fell below a lower limit (
Samoto, the host computer calculates and tracks the ball formation rate continuously. As shown in figure 7, if this rate is monitored over a predetermined period falls below the lower limit of 90%, it acts as the automatic trigger indicating that some abnormality that causes yarn breakage in the specific spindle is beginning to occur, figure 7 and ¶ [6 & 20-21]. M represents the number of days (or moments t) during that month-long period where the full ball formation rate dipped below that 90% requirement, figure 7).
Samoto in view of Cheng and Hu teaches utilizing a predictive model to estimate the full winding rate (winding quality) for various parameter sets and outputs a vector Y, which is a set of predicted parameters with specific timestamps and feature types (Cheng, the LSTM neural network outputs
Y
=
y
1
t
,
y
2
t
,
…
y
N
c
t
, which is multidimensional vector of predicted parameters corresponding to target moment t, where
N
c
is the total number of physical feature types monitored, ¶ [102]).
However, Samoto in view of Cheng and Hu doesn’t teach selecting specific outputs corresponding only to those M moments to isolate parameters.
Samoto in view of Cheng and Hu doesn’t teach selecting second prediction results corresponding to the M full-winding rates which do not meet the requirement of the preset full-winding rate from T second prediction results, wherein the T second prediction results are obtained after T times of the winding prediction step; and obtaining the candidate characteristic parameter set based on the second prediction results corresponding to the M full-winding rates which do not meet the requirement of the preset full-winding rate.
Von Grunigen teaches that when a fault is detected (by identifying parameter deviations from reference threshold, ¶ [26]), the electronic device selects the specific parameter and configuration data associated with those fault occurrences (the M anomalous moments) and applies them to machine learning algorithms to estimate the exact sources of the fault (figure 4 and ¶ [98]).
It would have been obvious to one ordinary skill in the art before the effective filling date of the claimed invention to incorporate von Grunigen event-triggered diagnostic filtering into Samoto in view of Cheng and Hu system. Filtering the T total running outputs down to the M anomalous moments and analyzing those specific moments “model parameters/influence degrees” to establish a candidate characteristic parameter set (the fault sources), helps the system to reduce CPU overhead and memory bandwidth allowing advance diagnostic model to run locally and cost-effectively on standard workshop hardware.
Claims 15 and 20 are rejected for the same reasons as set forth with respect to rejection of claims 11, 16 and 5.
Regarding claim 6, Samoto in view of Cheng, Hu and von Grunigen teaches the method of claim 5 as set forth with respect to rejection of claim 5.
Samoto in view of Cheng, Hu, and von Grunigen further teaches wherein obtaining the candidate characteristic parameter set based on the second prediction results(Cheng, the LSTM neural network outputs
Y
=
y
1
t
,
y
2
t
,
…
y
N
c
t
, which is multidimensional vector of predicted parameters corresponding to target moment t, where
N
c
is the total number of physical feature types monitored, ¶ [102]), corresponding to the M full-winding rates which do not meet the requirement of the preset full-winding rate (Samoto, T2 defect report including: T2 tension, Yarn speed, Velocity ratio, Temp H1, and Temp H2 , ¶ [18]. (Host computer 54 isolates the abnormal parameters when the full ball formation rate drops, ¶ [6 & 15] and the host computer calculates and tracks the ball formation rate continuously. As shown in figure 7, if this rate is monitored over a predetermined period falls below the lower limit of 90%, it acts as the automatic trigger indicating that some abnormality that causes yarn breakage in the specific spindle is beginning to occur, figure 7 and ¶ [6 & 20-21]. M represents the number of days (or moments t) during that month-long period where the full ball formation rate dipped below that 90% requirement, figure 7) comprises: obtaining M groups of initial sets based on the second prediction results corresponding to the M full-winding rates which do not meet the requirement of the preset full-winding rate (von Grunigen teaches that when a fault is detected (by identifying parameter deviations from reference threshold, ¶ [26]), the electronic device selects the specific parameter and configuration data associated with those fault occurrences (the M anomalous moments) and applies them to machine learning algorithms to estimate the exact sources of the fault (figure 4 and ¶ [98])).
However, Samoto in view of Cheng, and von Grunigen does not teach an initial set in the M groups of initial sets contains a first characteristic parameter having a second value larger than a preset numerical value in the second prediction results corresponding to the full-winding rates which do not meet the requirement of the preset full-winding rate; and selecting the candidate characteristic parameters from the M groups of initial sets to obtain the candidate characteristic parameter set.
Hu teaches an initial set in the M groups of initial sets contains (a target feature set that acts as the initial set compiling the candidate variables that are determined to have high importance, ¶ [43]) a first characteristic parameter having a second value (the feature parameters are used to characterize the degree of influence of each candidate feature on the product indicators [(like winding rate)], ¶ [9]) larger than a preset numerical value (in response to the feature parameters satisfying a predetermined condition, adding the corresponding candidate feature to the target feature set ¶ [10 & 18]) in the second prediction results (predicted numerical value output by the feature analysis model, ¶ [67]) corresponding to the product feature indexes which do not meet the requirement of the preset product feature indexes (extracting importance indexes is triggered to optimize the product and improve user experience when metrics fall below expectations, ¶ [121]); and selecting the candidate characteristic parameters (evaluating the feature parameters of candidate features and selecting those that represent high importance to be included as target features, ¶ [132]) from the M groups of initial sets (iteratively evaluating candidate feature subsets (e.g., set {a, b, c, d} updating to {b, c, d}) and extracting the highest ranking features from these groups,¶ [132]) to obtain the candidate characteristic parameter set (in response to the number of target features in the target feature set reaching a preset number, outputting the target feature set, figure 1 S160).
It would have been obvious to one ordinary skill in the art before the effective filling date of the claimed invention to combine Samoto in view of Cheng, and von Grunigen’s system, which continuously monitors rates but trigger diagnostics when the quality rate violates a preset lower threshold, with Hu’s teaching, which calculates feature parameters and applies a threshold condition to isolate high-impact feature, to eliminate the strong subjectivity and inconsistency of manual human interference.
Claim 7 is rejected under 35 U.S.C. 103 as being unpatentable over Samoto, Cheng, Hu, and von Grunigen and further in view of Wang Guilan et al. (Plasma Spray Coating Quality Forecast Based on Principal Component Analysis, JOURNAL OF MECHANICAL SCIENCE AND TECHNOLOGY, VOL. 25, NO. 12, DEC. 2006) hereinafter Wang Guilan, see the English Translation and Toda et al. (JP 2012021253 A) hereinafter Toda, see the English Translation.
Regarding claim 7, Samoto in view of Cheng, Hu and von Grunigen teaches the method of claim 5 as set forth with respect to rejection of claim 5.
Samoto in view of Cheng, Hu and von Grunigen teaches obtaining the candidate characteristic parameter set based on the second prediction results corresponding to the M full-winding rates which do not meet the requirement of the preset full-winding rate (as set forth with respect to rejection of claim 5).
Samoto in view of Cheng, Hu and von Grunigen doesn’t teach obtaining a characteristic parameter matrix of M rows × N columns based on the second prediction results corresponding to the M full-winding rates which do not meet the requirement of the preset full-winding rate, wherein each row in the characteristic parameter matrix corresponds to N second values, and the second values of the same first characteristic parameter corresponding to different full-winding rates are located in the same column; weighting the columns in the characteristic parameter matrix to obtain a vector of 1 row × N columns, wherein the vector of 1 row × N columns represents N third values; and obtaining the candidate characteristic parameter set based on the vector of 1 row × N columns.
Wang Guilan teaches obtaining a characteristic parameter matrix of M rows × N columns (formula No. 1 where the parameters are organized as a multi-row, multi-column matrix representing different variables across test samples, page 2 ) and weighting the columns in the characteristic parameter matrix to obtain a vector of 1 row × N columns (uses Principal Component Analysis (PCA) to calculate eigenvectors (pi) and eigenvalues (λi), then it multiplies (weights) the standardized column variables (xj*) by calculated coefficients (αij) to obtain a single principal component vector score, formula 9, page 5. This represents the analogous dimensional reduction (M × N to 1 × N) of temporal observations into a single, pooled priority vector).
It would have been obvious to one ordinary skill in the art before the effective filling date of the claimed invention to organize winding parameters as taught by Samoto in view of Cheng, Hu and von Grunigen into Wang Guilan’s M × N matrix format and execute column -weighted compression to smooth out transient, single moment sensor spikes. This yields the predictable benefit of a clean, stable 1 × N priority vector of persistent mechanical anomalies, optimizing edge-processor efficiency on the factory floor.
Samoto in view of Cheng, Hu, von Grunigen and Wang Guilan doesn’t teach obtaining the candidate characteristic parameter set based on the vector of 1 row × N columns.
Toda teaches the vector of 1 row × N columns (fitting a multivariate regression equation representing a yarn quality target index,
R
=
k
1
x
+
k
2
Y
+
k
3
, where the regression coefficients
k
=
k
1
,
k
2
,
⋯
k
N
are calculated for each of the N running textile machine process parameters¶ [27]. This array of calculated coefficients forms the mathematical analogous of the 1 row × N columns vector of third values). Toda teaches obtaining the candidate characteristic parameter set based on the vector (the diagnostic controller automatically evaluates the calculated coefficient vector K. It isolates the specific process parameter (column) associated with the largest absolute weighted regression coefficient Ki and extracts and displays it as the primary factor causing yarn quality deterioration, ¶ [28]).
It would have been obvious to one ordinary skill in the art before the effective filling date of the claimed invention to combine the teachings of Samoto in view of Cheng, Hu, von Grunigen and Wang Guilan with Toda. By automatically isolating the parameters with the largest weights from the compressed vector to form the candidate characteristic parameter set, the combined system eliminates the subjectivity of manual inspections, allowing the winding workshop edge-controllers to reliably pinpoint the exact mechanical actuator driving a drop in the full-winding rate.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SAEEDE NAFOOSHE whose telephone number is (571)272-8629. The examiner can normally be reached Monday-Friday 8:00 am -5:00pm.
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, Andrew Schechter can be reached at 571-272-2302. 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.
/SAEEDE NAFOOSHE/ Examiner, Art Unit 2857
/ANDREW SCHECHTER/ Supervisory Patent Examiner, Art Unit 2857
Prosecution Insights