Data augmentation of sensor time-series data for failure assessment
By generating artificial samples using TVAR models based on real sensor data, the system addresses the scarcity of anomalous data, enhancing fault detection accuracy and predictive maintenance performance.
Patent Information
- Application Number
- PCT/US2024/032955
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-06-07
- Filing Date
- 2024-06-07
- Publication Date
- 2024-12-12
AI Technical Summary
The scarcity of anomalous and faulty time series data in sensor systems makes it difficult for anomaly detection systems to differentiate between normal and anomalous instances accurately, leading to imbalanced datasets and poor performance in fault detection and predictive maintenance.
The system generates artificial samples of multivariate non-stationary time-series data using mean and covariance estimates from real sensor data, fitting a time-varying autoregressive (TVAR) model to augment the dataset, allowing for improved diagnostics and prognostics in fault detection.
This approach enhances the quality of diagnostics and prognostics solutions by creating synthetic data that improves the performance of machine learning models in identifying faults, particularly in cases where real data is scarce, thereby reducing false alarms and improving predictive maintenance.
Smart Images

Figure US2024032955_12122024_PF_FP_ABST
Abstract
Description
202310877 DATA AUGMENTATION OF SENSOR TIME-SERIES DATA FOR FAILURE ASSESSMENT BACKGROUND
[0001] An anomaly or faulty behavior can generally be defined as an event or occurrence that does not follow expected or normal behavior. In the context of neural networks or machine learning, an anomaly can be difficult to define, but the definition can be critical to the success and effectiveness of a given anomaly detector. An efficient anomaly detector should be capable of differentiating between anomalous and normal instances with high precision and accuracy, so as to detect as many as possible of the anomalous instances while avoiding false alarms. Supervised learning models, for instance models for anomaly detection, typically require large amounts of data to be trained. It is recognized herein, however, that acquiring and storing sensor data is often expensive or cost-prohibitive. Furthermore, instances of time series data corresponding to faulty or anomalous behaviors tend to be scarce. For example, various machines and systems are typically expected to operate normally the vast majority of the time, such that anomalies or faults are a rare exception in healthy environments. Thus, in some cases, sensor time series data corresponding to anomalous and faulty behaviors are misrepresented and anomaly detection datasets are often imbalanced. This lack of data problem, among other technical issues, can make it difficult or impossible to properly characterize faulty and anomalous time series data through, for example, mathematical expressions and machine learning or data science models, thereby limiting or prohibiting the ability of various systems to identify and rectify various anomalous and potentially dangerous or hazardous situations. BRIEF SUMMARY
[0002] Embodiments of the invention address and overcome one or more of the described- herein shortcomings by providing methods, systems, and apparatuses that improve fault detection.
[0003] In an example aspect, a failure diagnostics or prognostics computer system can include obtaining data, for instance from a sensor configured to capture real samples of data from a physical system. The computer system can include a memory storing instructions that, when202310877 executed by the processor, cause the processor to generate, from the real samples of data, mean and covariance estimates. Based on the mean and covariance estimates, the computer system can generate artificial samples of multivariate non-stationary time-series data. Furthermore, the computer system can train a machine learning model with the artificial samples, so as to define a trained machine learning model. The trained machine learning model can be configured to detect or predict a fault within the physical system, based on the real samples of data. The real samples of data can be captured by the sensor so as to define a time-series dataset. In another example aspect, the computer system can fit a stochastic nonstationary model to the real samples of data, so as to generate a time-varying autoregressive (TVAR) model. In particular, the system can fit a first nonstationary model to a mean defined by the real samples of data over time, so as to generate a first TVAR sub-model that characterizes a first order of the real samples of data. Furthermore, the system can fit a second nonstationary model to a covariance defined by the real samples of data over time, so as to generate a second TVAR sub-model that characterizes a second order of the real samples of data. Alternatively, in some cases, a single TVAR model can fit both mean and covariance. BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS
[0004] The foregoing and other aspects of the present invention are best understood from the following detailed description when read in connection with the accompanying drawings. For the purpose of illustrating the invention, there is shown in the drawings embodiments that are presently preferred, it being understood, however, that the invention is not limited to the specific instrumentalities disclosed. Included in the drawings are the following Figures:
[0005] FIG. 1 is a block diagram of an example system that includes a failure diagnostics or prognostics computer system configured to augment time series data, in accordance with an example embodiment.
[0006] FIG. 2 shows example synthetic time series that are generated by the failure diagnostics or prognostics computer system with an example time series dataset, in accordance with an example.
[0007] FIG. 3 illustrates an example data augmentation method that can be performed by the failure diagnostics or prognostics computer system, in accordance with an example embodiment.202310877
[0008] FIG. 4 illustrates a computing environment within which embodiments of the disclosure may be implemented. DETAILED DESCRIPTION
[0009] As an initial matter, variations of anomaly and fault can be used interchangeably herein without limitation, unless otherwise specified. It is recognized herein that in order to improve an anomaly detection system via time series classification, data can be augmented based on multivariate time series modeling. In accordance with various embodiments, a multivariate, nonstationary stochastic model can characterize time series data from underrepresented classes or groups in a dataset, for instance anomalous or faulty time series data. In particular, the model can be fitted to the time series of underrepresented classes, then used to generate new synthetic time series data.
[0010] Fault diagnostics and failure prognostics of equipment and systems (predictive maintenance, prognostics and health management (PHM), etc.) have been the subject of active investigation and intense real world solution development during the recent decades. It is recognized herein that avoiding the occurrence of failures or reducing their consequences in general translates various benefits, such as reduced downtime, improved yield, and safety. A large variety of methods have been proposed over the years, ranging from reliability methods based on population statistics to data-driven solutions employing cutting-edge machine learning methods, or physics-based approaches incorporating advanced failure mechanism models. It is recognized herein, however, that one of the most important technical problems limiting the successful development and application of PHM solutions are related to the availability of good quality historical data that can be used for the development and validation of such solutions. It is further recognized herein that failure events may be rare but their impact can be significant for developing related PHM solutions, among others. Furthermore, in some cases, relevant sensor data associated with historical events is not collected, or is collected but corresponding labels are unavailable, or the corresponding labels are not reliable so as to define the actual failure mode, etc.
[0011] Such data limitations can affect data driven approaches, reliability-based approaches, physics-based approaches. For example, reliability solutions are based on historical data, despite simplifications such as the assumption of parametric statistical models or the202310877 combination of data from different but similar equipment. Additionally, physics-based diagnostics and prognostics solutions might also be validated based on sufficient real historical data before they can be deployed to the field. Whenever enough data is not available, the development and validation of diagnostics or prognostics solution may result in poor performance or may not even be feasible. Thus, it is recognized herein that data-centric approaches to PHM can be valuable in terms of addressing the relevant data-related challenges faced in real world both for fault diagnostics and failure prognostics.
[0012] In accordance with various embodiments, multivariate time-series data is augmented based on time-varying autoregressive (TVAR) models. In particular, information is extracted from scarce data and used to create additional samples in a way that can improve the quality of diagnostics and prognostics solutions. One characteristic of the data augmentation described herein that makes it especially suited for failure diagnostics and prognostics problems is the fact that it can directly deal with non-stationary time series data.
[0013] Embodiments address technical problems, among others, of augmenting sensor time series data from the viewpoint of time-series groups or classes. To this end, data can be divided into classes such as anomalous and normal. In some examples, time-series datasets that are augmented include time series data that are sampled equally within the same dataset, and time- series data of the same class that can modeled by the same multi-variate, potentially non- stationary stochastic process. Potentially non-stationary behavior refers to the properties of the stochastic process being permitted to vary over time.
[0014] Referring initially to FIG. 1, an example automation or industrial network or physical system 100 can include one or more plants or production networks 104 that contain control logic, host web servers, and the like. For example, the physical system 100 can include an enterprise or IT network 102 and multiple operational plant or production networks 104 communicatively coupled to the IT network 102. The production network 104 or enterprise network 102 can include one or more failure diagnostics or prognostics computer systems or modules 106 connected within the production network 104. The computer system 106 can define a time-varying autoregressive (TVAR) model 107, as further described herein. An example computer system 106 is connected to the IT network 102. The arrangement of the computer system 106 can vary as desired, and all such arrangements are contemplated as being within the scope of this disclosure. For example, in some cases, data augmentation and model training described herein can be performed on a different system than the system that monitors202310877 a physical system and collects the data from the physical system. In other cases, the system that collects the data can also augment the data and train the models.
[0015] Still referring to FIG. 1, the production network 104 can include various production machines configured to work together to perform one or more manufacturing operations. Example production machines of the production network 104 can include, without limitation, robots 108 and other field devices that can be controlled by a respective PLC 114, such as sensors 110, actuators 112, or other machines, such as automatic guided vehicles (AGVs) 108. The PLC 114 can send instructions to respective field devices. In some cases, a given PLC 114 can be coupled to a human machine interfaces (HMIs) 116. It will be understood that the physical system 100 is simplified for purposes of example. That is, the physical system 100 may include additional or alternative nodes or systems, for instance other network devices, that define alternative configurations, and all such configurations are contemplated as being within the scope of this disclosure.
[0016] The network or system 100, in particular each production network 104, can define a field portion or level 118 and plant level or portion 120. For example, and without limitation, the plant level 120 can define one or more industrial plants or systems that can be geographically and functionally separate from or independent of each other. For example, the plant level 120 can include Brownfield plants and Greenfield plants that are each connected to respective field devices within the field level 118. The field level 118 can include various field devices such as the robots 108, PLC 114, sensors 110, actuators 112, HMIs 116, and AGVs. The sensors 110 can be configured to capture real samples of time series data from the physical system 100. The field portion 118 can define one or more production lines or control zones associated with a given plant in the plant level 120. The PLC 114, sensors 110, actuators 112, and HMI 116 within a given production line can communicate with each other via a respective field bus 122. Each control zone can be defined by a respective PLC 114, such that the PLC 114, and thus the corresponding control zone, can connect to the respective plant portion 120 via an Ethernet connection 124. In some cases, the robots 108 and AGVs can be configured to communicate with other devices within the fieldbus portion 118 via a Wi-Fi connection 126. Similarly, the robots 108 and AGVs can communicate with the Ethernet portion 120, in particular a Supervisory Control and Data Acquisition (SCADA) server 128, via the Wi-Fi connection 126. In various examples, a respective computer system 106 is communicatively coupled between the PLC 114 and the respective plant in the plant level 120, for instance via202310877 the Ethernet connection 124 or the Wi-Fi connection 126. In some examples, the computer system 106 is defined by the PLC 114.
[0017] The plant level 120 of a given production network 104 can include various computing devices or subsystems communicatively coupled together via the Ethernet connection 124. Example computing devices or subsystems in the plant portion 120 include, without limitation, a mobile data collector 130, HMIs 132, the SCADA server 128, the computing system 106, a wireless router 134, a manufacturing execution system (MES) 136, an engineering system (ES) 138, and a log server 140. The ES 138 can include one or more engineering works stations. In an example, the MES 136, HMIs 132, ES 138, and log server 140 are connected to the production network 104 directly. The wireless router 134 can also connect to the production network 104 directly. Thus, in some cases, mobile users, for instance the mobile data collector 130 and robots 108 (e.g., AGVs), can connect to the production network 104 via the wireless router 134.
[0018] It will be appreciated that the program modules, applications, computer-executable instructions, code, or the like depicted herein are merely illustrative and not exhaustive, and that processing described as being supported by any particular module may alternatively be distributed across multiple modules or performed by a different module. In addition, various program module(s), script(s), plug-in(s), Application Programming Interface(s) (API(s)), or any other suitable computer-executable code may be provided to support functionality provided by the program modules, applications, or computer-executable code depicted herein and / or additional or alternate functionality. Further, functionality may be modularized differently such that processing described as being supported collectively by the collection of program modules depicted herein may be performed by a fewer or greater number of modules, or functionality described as being supported by any particular module may be supported, at least in part, by another module. In addition, program modules that support the functionality described herein may form part of one or more applications executable across any number of systems or devices in accordance with any suitable computing model such as, for example, a client-server model, a peer-to-peer model, and so forth. In addition, any of the functionality described as being supported by any of the program modules depicted herein may be implemented, at least partially, in hardware and / or firmware across any number of devices, for instance computer system 106.202310877
[0019] Referring again FIG. 1, the diagnostic or prognostics computing system 100 can include a stochastic nonstationary model that is fitted to data, so as to define the time-varying autoregressive (TVAR) model 107. The TVAR model 107 exhibits various advantages over other classical stochastic time series models (e.g., AR, ARMA, ARIMA). For example, the TVAR model 107 does not require the time series to be stationary, or that eventual nonstationary dynamics in the data follows some specific patterns, such as linear trend and seasonality, which is explicitly identified by the aforementioned classical models as assumptions. Therefore, it is recognized herein that TVAR models, for instance the TVAR model 107, can be better suited to model a wider range of real-world time series data, which often exhibits a large variety of nonstationary behaviors that cannot be modeled by traditional approaches.
[0020] Unlike traditional time series models, the TVAR model 107 can allow its parameters to change over time, for example, to better capture the time-varying dynamics in the data. The TVAR model 102 can define a plurality of parameters and a set of basis functions, weights, and a model order, which can control how the parameters change over time. For example, the TVAR model 102 can model different nonstationary behaviors by selecting different basis functions, weights, and model orders. In an example, the TVAR model 107 is customized such that the mean and covariance vary over time following an arbitrary functional form.
[0021] Characterizing the time series data of a given data class with the TVAR model 102 means that the time series data from the same class are modeled by the same TVAR mathematical expression, based on the TVAR customization. The TVAR customization according to various embodiments employs the following stochastic expression to model a given multivariate time series ^^(^^)with dimension M (e.g., ^^(^^)∈ R^^):(1)
[0022] Referring to equation (1), ^^( ^^) represents a perturbation vector characterized by a multivariate (dimension M), zero-mean stationary random process with correlation matrix given by ^^; ^^(^^)is an arbitrary functional form characterizing the time-varying behavior of the mean vector and covariance matrix of the time series; ^^1and ^^2are real constants controlling the convergence speed of the mean and covariance of ^^(^^), respectively; λ is a real constant202310877 controlling the gain of the covariance matrix of ^^(^^); and the initial value of ^^(^^)is given by ^^(0) = γ ^^^(0) with ^^ being a real constant controlling the gain of the mean vector of ^^( ^^) and ^^^(0)being an arbitrary constant basis vector.
[0023] The nonstationary mean vector ^^( ^^) and covariance matrix ^^( ^^) of (1) can be defined as: (2)(3)
[0024] In accordance with various embodiments, the TVAR model 107 is defined in accordance with the equations described above, so as to represent the underlying stochastic process of the time series data belonging to a given class. In various examples, such a stochastic process can be described by empirical statistics computed from the data. Thus, fitting the TVAR model 102 can be performed by finding parameters and functional forms that make the expressions for the mean vector and covariance matrix of the TVAR process match the empirical statistics calculated for the data. It is recognized herein that this case differs from the usual time series model fitting paradigm, where model parameters are adjusted to match a single time series. After suitable TVAR parameters and functional forms are determined, the system 100 can plug them into the TVAR process formula in Equation (1) to create randomized augmented data for a given signal class.
[0025] In various examples, the functional form ^^( ^^) is computed via curve interpolation over the nonstationary statistic curves of the data. Thus, ^^( ^^) can be named here in the TVAR interpolation function. In some cases, because ^^( ^^) appears both in the mean and in the covariance expressions (see Equations (2) and (3)), a single TVAR model (based on a single^^( ^^)) to describe time series datasets in which mean and covariance change in distinct forms over time, is insufficient. To account for these cases, in an example, two TVAR sub-models characterize the underlying stochastic process of the signal class in accordance with various embodiments. For example, a first TVAR sub-model can define a mean model (TVARmsub- model) and a second TVAR sub-model can define a covariance model (TVARcsub-model). Continuing with the example, each of the first and second sub-models can define interpolated expressions^^m( ^^)and^^c( ^^), respectively, and a set of constant parameters{ ^^m, ^^m, ^^1m, ^^2m}and{ ^^m, ^^m, ^^1m, ^^2m},respectively, which are initialized with the sub-models. These constants can be202310877 later adjusted to fine-tune the TVAR theoretical mean and covariance to best match the first- and second-order statistics of the data.
[0026] In various examples, the TVARm sub-model captures the first-order dynamics of the time series of a given class, and the TVARc sub-model captures the second-order dynamics. The final augmented time series can be obtained by using the interpolated expressions ^^m(^^)and ^^c( ^^) and the set of parameters for TVARmand TVARcin Equation (1), which will lead to Equations (4) and (5), both of which can be used to generate synthetic time series data:
[0027] Referring to the above equations, Equations (4) and (5) can represent behaviors captured by the mean and covariance, respectively. The augmented signals have length N and are computedl=1,…,Ltimes, with every N and L defining free parameters. The mean vector and covariance matrix of Equations (4) and (5) are shown in Equations (6) and (7), and in Equations (8) and (9), respectively:
[0028] Thus, because analytical expressions can be searched for the augmented time series, for instance such as in Equations (4) and (5), finding the TVAR interpolating functions can stand for interpolating regression expressions for ^^m( ^^) and ^^c( ^^) over the empirical statistics calculated from the data. Next, the empirical statistics can be calculated from the data and the TVAR interpolation functions.202310877
[0029] For example, the interpolation functions ^^m(^^)and ^^c(^^)can be fitted to the empirical first- and second-order statistics computed from the data. In an example, the empirical mean vector and covariance matrix of the signals can be computed with ensemble averaging. In particular, for example, the collection multivariate time series^^( ^^)^^belonging to a given class or group g can be defined as follows: (10) where time serieshave length N and dimension M, and
[0030] Based on ^^( ^^), the empirical mean vector and covariance matrix at time n can be computed via ensemble averaging as:
[0031] In an example, the time series datasets can be univariate (e.g., M=1 in Equation (11)). It will be understood that univariate datasets are described herein for purposes of example, but the embodiments described herein can include multivariate cases without loss of generality. Furthermore, in the example, the superscript (g) characterizing the group or class of the time series is dropped. By doing so, Equations (12) and (13) become the sequences^ഥ^( ^^)and^^̅( ^^), respectively.
[0032] As described herein, sub-models TVARm and TVARc can characterize the first- and second-order statistics of the data separately. Therefore, ^^m(^^)can be interpolated solely to ^ഥ^( ^^), and ^^c(^^)can be interpolated to ^^̅( ^^). In accordance with another embodiment, this solution can be reformulated so that a single model characterizes both first and second-order statistics. In an example, to address the calculation of the interpolation functions ^^m( ^^) and ^^c( ^^), sinusoidal regression via Discrete Fourier Transform (DFT) can be performed, which allows modeling a wide range of sensor time series data. In other cases, other curve fitting techniques can be implemented, such as splines and other forms of linear and nonlinear regressions, as long as they return an analytical expression for the fitted curves.202310877
[0033] A sinusoidal regression for ^^m(^^)) and ^^^^(^^)can be obtained by decomposing ^ഥ^( ^^) and ^^̅( ^^) using the DFT. Below is an example to compute the egression expression only for ^^m( ^^), as the results for^^c( ^^)will be the same (the only difference being replacing^ഥ^( ^^)by^^̅( ^^)in the calculations). In particular, for example, the DFT of ^ഥ^( ^^) at discrete frequency ω^^= 2π ^^ can be represented in Equation 14:
[0034] By expressing^^^^(ω^^)in terms of its magnitude| ^^^^(ω^^)|and phase^^^^( ^^^^), such that ^^^^(^^^^)=|^^^^(^^^^)|, and by computing Equation (14) for ω0, … ,ω^^−1(e.g., for all the N available time points), the system can generate the following vector:Referring to Equation (15), the magnitude and phase spectra of ^ഥ^( ^^) can be computed for frequency valuesand ^ഥ^( ^^) can be reconstructed from Equation (15) by computing the inverse DFT as, for example:
[0035] In an example, the system can sort vector ^^^^in descending order according to the values of magnitude, so as to obtain:where ω′^^is the sorted frequency variable, such that largest and smallest values of magnitude in ^^′^^are at ω′0and ω′^^−1, respectively. The system can recompute Equation (16) by using the values of the sorted vector in Equation (17) up to the ^^^^ℎelement, so as to compute:
[0036] In an example, as an approximation of^ഥ^( ^^)with the^^most significant frequencies of the spectrum (here, described by the^^largest values of magnitude), Equation (18) can be expressed in terms of cosines and sines. Furthermore, by taking the real part of the resulting expression, the system can obtain a final sinusoidal regression formula for ^^m(^^):202310877
[0037] The system can repeat the steps described herein for^^̅( ^^),with| ^^^^(ω′^^)|being its sorted magnitude, so as to generate an expression for ^^c( ^^):
[0038] Referring now to FIG. 2, examples of augmented signals are shown to an example empirical mean curve 202 and an example empirical variance curve 204 that are computed with an example time series dataset. In particular, a first sinusoidal mean interpolation ^^m( ^^) 206a and a first sinusoidal variance interpolation ^^^^(^^)206b are shown for a first order ( ^^ = 4), a second sinusoidal mean interpolation ^^m( ^^) 208a and a second sinusoidal variance interpolation ^^^^(^^)208b are shown for a second order ( ^^ = 10), and a third sinusoidal mean interpolation ^^m( ^^) 210a and a third sinusoidal variance interpolation ^^^^( ^^) 210b are shown for a third order ( ^^ = 14). It will be understood that the orders 4, 10, and 14 are presented as an example, and the orders can vary as desired in accordance with various embodiments. As shown but without being bound by theory, the models can fit empirical statistic curves well, particularly for higher orders. Referring now to FIG. 3, an example data augmentation method 300 illustrates data augmentation described above, in accordance with an example embodiment. The data augmentation method 300 can be repeated for any class of data that a user wants to augment, but some classes of data are presented below by way of examples. The data augmentation method 300 can improve the performance of various machine learning (ML) systems in classification settings in which data from abnormal operations are needed. In particular, for example, specific failures (classifications) are detected. In an example, data augmentation in accordance with embodiments described herein is performed on bearing vibration signals. In particular, the real samples of data can contain vibration signals collected from a drive-end bearing of an electrical motor operating under different loads. In the example, the selected vibration signals are sampled at 12 kHz for motor speed of 1797 RPM. The chosen signals are from a normal bearing (Class 0), a bearing with a crack in the inner race (Class 1), and another one with a crack in the ball (Class 2). The cracks of the faulty bearings have 0.007 inches. In the example, the system 106 can generate signal snippets of 0.1 second from the vibration data of the bearings. To simulate challenging real-world scenarios, training samples from the faulty classes have been under-sampled to make the training data imbalanced.
[0039] In other example, data augmentation is performed on a second dataset that represents a collection of signals for anomaly detection. In particular, the second example dataset consists202310877 of noise signals collected from sensors installed on automotive engines that are either normal (Class 0), or present failure symptoms (Class 1).
[0040] In yet another example, data augmentation is performed in accordance with embodiments described herein on a third example dataset that represents a set of measurements from piezoelectric (PZT) sensors. In particular, the example dataset contains Lamb waves from fatigue experiments on aluminum lap joints. The selected Lamb wave signals are measured for eight specimens (specimen T1 to T8). For each specimen, different loads are applied to the testing material, and a pair of signals from two sensors are recorded (each signal with 4000 samples). Then, the crack lengths are measured for each specimen, applied load, and pair of signals. In this example, a given signal sample corresponds to a damaged or faulty sample (Class 1) if the observed crack length exceeds 4 mm. Otherwise, the sample is considered as normal or non-anomalous (Class 0). Specimens T3 to T8 represent the training data and the remaining specimens represent the training data. Thus, both train and test partitions can have data from Class 1.
[0041] With respect to the example datasets described above, augmentation can be evaluated in the context of time series classification in association with various machine learning models, for instance a convolutional neural network (CNN) and a random forest (RF). For example, the CNN architecture can define model hyperparameters (number of filters, batch and kernel sizes, etc.) that are determined with the KerasTuner module, and the resulting architecture can be trained for 20 epochs. The RF model can define, for example, a standard random forest implementation offered by Python scikit-learn library. For example, the RF architecture can define 50 estimators and a max depth of 5. The RF model can be fitted to a tabular feature representation extracted from the time series data with the Python package tsfresh. This module can allow for extracting typical features from time series via a systematic framework.
[0042] The augmentation described herein can generate new synthetic faulty samples belong belonging to the training partitions of various datasets, for instance the example datasets described above. In an example, the signal samples are augmented with the following augmentation fractions: 0.1, 0.25, 0.5 and 1. The CNN and RF models can be trained in the original and augmented train data partitions, and tested with the available test data. The evaluation in the test data can be performed by computing the accuracy score of classifying the time series samples into the considered Faulty / Normal categories. Without being bound by theory, it is recognized herein that the application of data augmentation improves the overall202310877 detection performances of the ML models, particularly in cases that are more challenging, for each of the example datasets and augmentation fractions.
[0043] Thus, as described herein, a failure diagnostics or prognostics computer system can include obtaining data, for instance from a sensor configured to capture real samples of data from a physical system. The computer system can include a memory storing instructions that, when executed by the processor, cause the processor to generate, from the real samples of data, mean and covariance estimates. Based on the mean and covariance estimates, the computer system can generate artificial samples of multivariate non-stationary time-series data. Furthermore, the computer system can train a machine learning model with the artificial samples, so as to define a trained machine learning model. The trained machine learning model can be configured to detect or predict a fault within the physical system, based on the real samples of data. The real samples of data can be captured by the sensor so as to define a time- series dataset. In another example aspect, the computer system can fit a stochastic nonstationary model to the real samples of data, so as to generate a time-varying autoregressive (TVAR) model. In particular, the system can fit a first nonstationary model to a mean defined by the real samples of data over time, so as to generate a first TVAR sub-model that characterizes a first order of the real samples of data. Furthermore, the system can fit a second nonstationary model to a covariance defined by the real samples of data over time, so as to generate a second TVAR sub-model that characterizes a second order of the real samples of data. Alternatively, in some cases, a single TVAR model can fit both mean and covariance.
[0044] FIG. 4 illustrates an example of a computing environment within which embodiments of the present disclosure may be implemented. A computing environment 500 includes a computer system 510 that may include a communication mechanism such as a system bus 521 or other communication mechanism for communicating information within the computer system 510. The computer system 510 further includes one or more processors 520 coupled with the system bus 521 for processing the information. The failure diagnostics or prognostics computer system 106 may include, or be coupled to, the one or more processors 520.
[0045] The processors 520 may include one or more central processing units (CPUs), graphical processing units (GPUs), or any other processor known in the art. More generally, a processor as described herein is a device for executing machine-readable instructions stored on a computer readable medium, for performing tasks and may comprise any one or combination of, hardware and firmware. A processor may also comprise memory storing machine-readable202310877 instructions executable for performing tasks. A processor acts upon information by manipulating, analyzing, modifying, converting or transmitting information for use by an executable procedure or an information device, and / or by routing the information to an output device. A processor may use or comprise the capabilities of a computer, controller or microprocessor, for example, and be conditioned using executable instructions to perform special purpose functions not performed by a general purpose computer. A processor may include any type of suitable processing unit including, but not limited to, a central processing unit, a microprocessor, a Reduced Instruction Set Computer (RISC) microprocessor, a Complex Instruction Set Computer (CISC) microprocessor, a microcontroller, an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), a System-on-a-Chip (SoC), a digital signal processor (DSP), and so forth. Further, the processor(s) 520 may have any suitable microarchitecture design that includes any number of constituent components such as, for example, registers, multiplexers, arithmetic logic units, cache controllers for controlling read / write operations to cache memory, branch predictors, or the like. The microarchitecture design of the processor may be capable of supporting any of a variety of instruction sets. A processor may be coupled (electrically and / or as comprising executable components) with any other processor enabling interaction and / or communication there-between. A user interface processor or generator is a known element comprising electronic circuitry or software or a combination of both for generating display images or portions thereof. A user interface comprises one or more display images enabling user interaction with a processor or other device.
[0046] The system bus 521 may include at least one of a system bus, a memory bus, an address bus, or a message bus, and may permit exchange of information (e.g., data (including computer-executable code), signaling, etc.) between various components of the computer system 510. The system bus 521 may include, without limitation, a memory bus or a memory controller, a peripheral bus, an accelerated graphics port, and so forth. The system bus 821 may be associated with any suitable bus architecture including, without limitation, an Industry Standard Architecture (ISA), a Micro Channel Architecture (MCA), an Enhanced ISA (EISA), a Video Electronics Standards Association (VESA) architecture, an Accelerated Graphics Port (AGP) architecture, a Peripheral Component Interconnects (PCI) architecture, a PCI-Express architecture, a Personal Computer Memory Card International Association (PCMCIA) architecture, a Universal Serial Bus (USB) architecture, and so forth.202310877
[0047] Continuing with reference to FIG. 4, the computer system 510 may also include a system memory 530 coupled to the system bus 521 for storing information and instructions to be executed by processors 520. The system memory 530 may include computer readable storage media in the form of volatile and / or nonvolatile memory, such as read only memory (ROM) 531 and / or random access memory (RAM) 532. The RAM 532 may include other dynamic storage device(s) (e.g., dynamic RAM, static RAM, and synchronous DRAM). The ROM 531 may include other static storage device(s) (e.g., programmable ROM, erasable PROM, and electrically erasable PROM). In addition, the system memory 530 may be used for storing temporary variables or other intermediate information during the execution of instructions by the processors 520. A basic input / output system 533 (BIOS) containing the basic routines that help to transfer information between elements within computer system 510, such as during start-up, may be stored in the ROM 531. RAM 532 may contain data and / or program modules that are immediately accessible to and / or presently being operated on by the processors 520. System memory 530 may additionally include, for example, operating system 534, application programs 535, and other program modules 536. Application programs 535 may also include a user portal for development of the application program, allowing input parameters to be entered and modified as necessary.
[0048] The operating system 534 may be loaded into the memory 530 and may provide an interface between other application software executing on the computer system 510 and hardware resources of the computer system 510. More specifically, the operating system 534 may include a set of computer-executable instructions for managing hardware resources of the computer system 510 and for providing common services to other application programs (e.g., managing memory allocation among various application programs). In certain example embodiments, the operating system 534 may control execution of one or more of the program modules depicted as being stored in the data storage 540. The operating system 534 may include any operating system now known or which may be developed in the future including, but not limited to, any server operating system, any mainframe operating system, or any other proprietary or non-proprietary operating system.
[0049] The computer system 510 may also include a disk / media controller 543 coupled to the system bus 521 to control one or more storage devices for storing information and instructions, such as a magnetic hard disk 541 and / or a removable media drive 542 (e.g., floppy disk drive, compact disc drive, tape drive, flash drive, and / or solid state drive). Storage devices 540 may202310877 be added to the computer system 510 using an appropriate device interface (e.g., a small computer system interface (SCSI), integrated device electronics (IDE), Universal Serial Bus (USB), or FireWire). Storage devices 541, 542 may be external to the computer system 510.
[0050] The computer system 510 may also include a field device interface 565 coupled to the system bus 521 to control a field device 566, such as a device used in a production line. The computer system 510 may include a user input interface or GUI 561, which may comprise one or more input devices, such as a keyboard, touchscreen, tablet and / or a pointing device, for interacting with a computer user and providing information to the processors 520.
[0051] The computer system 510 may perform a portion or all of the processing steps of embodiments of the invention in response to the processors 520 executing one or more sequences of one or more instructions contained in a memory, such as the system memory 530. Such instructions may be read into the system memory 530 from another computer readable medium of storage 540, such as the magnetic hard disk 541 or the removable media drive 542. The magnetic hard disk 541 (or solid state drive) and / or removable media drive 542 may contain one or more data stores and data files used by embodiments of the present disclosure. The data store 540 may include, but are not limited to, databases (e.g., relational, object-oriented, etc.), file systems, flat files, distributed data stores in which data is stored on more than one node of a computer network, peer-to-peer network data stores, or the like. The data stores may store various types of data such as, for example, skill data, sensor data, or any other data generated in accordance with the embodiments of the disclosure. Data store contents and data files may be encrypted to improve security. The processors 520 may also be employed in a multi-processing arrangement to execute the one or more sequences of instructions contained in system memory 530. In alternative embodiments, hard-wired circuitry may be used in place of or in combination with software instructions. Thus, embodiments are not limited to any specific combination of hardware circuitry and software.
[0052] As stated above, the computer system 510 may include at least one computer readable medium or memory for holding instructions programmed according to embodiments of the invention and for containing data structures, tables, records, or other data described herein. The term “computer readable medium” as used herein refers to any medium that participates in providing instructions to the processors 520 for execution. A computer readable medium may take many forms including, but not limited to, non-transitory, non-volatile media, volatile media, and transmission media. Non-limiting examples of non-volatile media include optical202310877 disks, solid state drives, magnetic disks, and magneto-optical disks, such as magnetic hard disk 541 or removable media drive 542. Non-limiting examples of volatile media include dynamic memory, such as system memory 530. Non-limiting examples of transmission media include coaxial cables, copper wire, and fiber optics, including the wires that make up the system bus 521. Transmission media may also take the form of acoustic or light waves, such as those generated during radio wave and infrared data communications.
[0053] Computer readable medium instructions for carrying out operations of the present disclosure may be assembler instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state- setting data, or either source code or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C++ or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The computer readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet Service Provider). In some embodiments, electronic circuitry including, for example, programmable logic circuitry, field-programmable gate arrays (FPGA), or programmable logic arrays (PLA) may execute the computer readable program instructions by utilizing state information of the computer readable program instructions to personalize the electronic circuitry, in order to perform aspects of the present disclosure.
[0054] Aspects of the present disclosure are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, may be implemented by computer readable medium instructions.
[0055] The computing environment 500 may further include the computer system 510 operating in a networked environment using logical connections to one or more remote202310877 computers, such as remote computing device 580. The network interface 570 may enable communication, for example, with other remote devices 580 or systems and / or the storage devices 541, 542 via the network 571. Remote computing device 580 may be a personal computer (laptop or desktop), a mobile device, a server, a router, a network PC, a peer device or other common network node, and typically includes many or all of the elements described above relative to computer system 510. When used in a networking environment, computer system 510 may include modem 572 for establishing communications over a network 571, such as the Internet. Modem 572 may be connected to system bus 521 via user network interface 570, or via another appropriate mechanism.
[0056] Network 571 may be any network or system generally known in the art, including the Internet, an intranet, a local area network (LAN), a wide area network (WAN), a metropolitan area network (MAN), a direct connection or series of connections, a cellular telephone network, or any other network or medium capable of facilitating communication between computer system 510 and other computers (e.g., remote computing device 580). The network 571 may be wired, wireless or a combination thereof. Wired connections may be implemented using Ethernet, Universal Serial Bus (USB), RJ-6, or any other wired connection generally known in the art. Wireless connections may be implemented using Wi-Fi, WiMAX, and Bluetooth, infrared, cellular networks, satellite or any other wireless connection methodology generally known in the art. Additionally, several networks may work alone or in communication with each other to facilitate communication in the network 571.
[0057] It should be appreciated that the program modules, applications, computer-executable instructions, code, or the like depicted in FIG. 4 as being stored in the system memory 530 are merely illustrative and not exhaustive and that processing described as being supported by any particular module may alternatively be distributed across multiple modules or performed by a different module. In addition, various program module(s), script(s), plug-in(s), Application Programming Interface(s) (API(s)), or any other suitable computer-executable code hosted locally on the computer system 510, the remote device 580, and / or hosted on other computing device(s) accessible via one or more of the network(s) 571, may be provided to support functionality provided by the program modules, applications, or computer-executable code depicted in FIG. 4 and / or additional or alternate functionality. Further, functionality may be modularized differently such that processing described as being supported collectively by the collection of program modules depicted in FIG. 4 may be performed by a fewer or greater202310877 number of modules, or functionality described as being supported by any particular module may be supported, at least in part, by another module. In addition, program modules that support the functionality described herein may form part of one or more applications executable across any number of systems or devices in accordance with any suitable computing model such as, for example, a client-server model, a peer-to-peer model, and so forth. In addition, any of the functionality described as being supported by any of the program modules depicted in FIG. 4 may be implemented, at least partially, in hardware and / or firmware across any number of devices.
[0058] It should further be appreciated that the computer system 510 may include alternate and / or additional hardware, software, or firmware components beyond those described or depicted without departing from the scope of the disclosure. More particularly, it should be appreciated that software, firmware, or hardware components depicted as forming part of the computer system 510 are merely illustrative and that some components may not be present or additional components may be provided in various embodiments. While various illustrative program modules have been depicted and described as software modules stored in system memory 530, it should be appreciated that functionality described as being supported by the program modules may be enabled by any combination of hardware, software, and / or firmware. It should further be appreciated that each of the above-mentioned modules may, in various embodiments, represent a logical partitioning of supported functionality. This logical partitioning is depicted for ease of explanation of the functionality and may not be representative of the structure of software, hardware, and / or firmware for implementing the functionality. Accordingly, it should be appreciated that functionality described as being provided by a particular module may, in various embodiments, be provided at least in part by one or more other modules. Further, one or more depicted modules may not be present in certain embodiments, while in other embodiments, additional modules not depicted may be present and may support at least a portion of the described functionality and / or additional functionality. Moreover, while certain modules may be depicted and described as sub-modules of another module, in certain embodiments, such modules may be provided as independent modules or as sub-modules of other modules.
[0059] Although specific embodiments of the disclosure have been described, one of ordinary skill in the art will recognize that numerous other modifications and alternative embodiments are within the scope of the disclosure. For example, any of the functionality202310877 and / or processing capabilities described with respect to a particular device or component may be performed by any other device or component. Further, while various illustrative implementations and architectures have been described in accordance with embodiments of the disclosure, one of ordinary skill in the art will appreciate that numerous other modifications to the illustrative implementations and architectures described herein are also within the scope of this disclosure. In addition, it should be appreciated that any operation, element, component, data, or the like described herein as being based on another operation, element, component, data, or the like can be additionally based on one or more other operations, elements, components, data, or the like. Accordingly, the phrase “based on,” or variants thereof, should be interpreted as “based at least in part on.”
[0060] Although embodiments have been described in language specific to structural features and / or methodological acts, it is to be understood that the disclosure is not necessarily limited to the specific features or acts described. Rather, the specific features and acts are disclosed as illustrative forms of implementing the embodiments. Conditional language, such as, among others, “can,” “could,” “might,” or “may,” unless specifically stated otherwise, or otherwise understood within the context as used, is generally intended to convey that certain embodiments could include, while other embodiments do not include, certain features, elements, and / or steps. Thus, such conditional language is not generally intended to imply that features, elements, and / or steps are in any way required for one or more embodiments or that one or more embodiments necessarily include logic for deciding, with or without user input or prompting, whether these features, elements, and / or steps are included or are to be performed in any particular embodiment.
[0061] The flowchart and block diagrams in the Figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the Figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and / or flowchart illustration, and combinations of blocks202310877 in the block diagrams and / or flowchart illustration, can be implemented by special purpose hardware-based systems that perform the specified functions or acts or carry out combinations of special purpose hardware and computer instructions.
Claims
202310877 CLAIMS What is claimed is:
1. A failure diagnostics or prognostics computer system, comprising: a sensor configured to capture real samples of data from a physical system; a memory storing instructions that, when executed by the processor, cause the processor to: generate, from the real samples of data, mean and covariance estimates; based on the mean and covariance estimates, generate artificial samples of multivariate non-stationary time-series data; and train a machine learning model with the artificial samples, so as to define a trained machine learning model, wherein the failure diagnostics or prognostics computer system further comprises the trained machine learning model that is configured to detect or predict a fault within the physical system, based on the real samples of data.
2. The system as recited in claim 1, wherein the real samples of data captured by the sensor define a time-series dataset.
3. The system as recited in claim 1, the memory further storing instructions that, when executed by the processor, cause the processor to: fit a stochastic nonstationary model to the real samples of data, so as to generate a time- varying autoregressive (TVAR) model.
4. The system as recited in claim 3, the memory further storing instructions that, when executed by the processor, cause the processor to: fit a first nonstationary model to a mean defined by the real samples of data over time, so as to generate a first TVAR sub-model that characterizes a first order of the real samples of data.
5. The system as recited in claim 4, the memory further storing instructions that, when executed by the processor, cause the processor to:202310877 fit a second nonstationary model to a covariance defined by the real samples of data over time, so as to generate a second TVAR sub-model that characterizes a second order of the real samples of data.
6. A method performed by a failure diagnostics or prognostics computer system, the method comprising: obtaining real samples of data from a physical system; generating, from the real samples of data, mean and covariance estimates; based on the mean and covariance estimates, generating artificial samples of multivariate non-stationary time-series data; training a machine learning model with the artificial samples, so as to define a trained machine learning model; and the trained machine learning model detecting or predicting a fault within the physical system, based on the real samples of data.
7. The method as recited in claim 6, the method further comprising: a sensor capturing the real samples of data that define a time series dataset.
8. The method as recited in claim 6, the method further comprising: fitting a stochastic nonstationary model to the real samples of data, so as to generate a time-varying autoregressive (TVAR) model.
9. The method as recited in claim 6, the method further comprising: fitting a first nonstationary model to a mean defined by the real samples of data over time, so as to generate a first TVAR sub-model that characterizes a first order of the real samples of data.
10. The method as recited in claim 6, the method further comprising: fitting a second nonstationary model to a covariance defined by the real samples of data over time, so as to generate a second TVAR sub-model that characterizes a second order of the real samples of data.
Citation Information
Cited By
Pressure sensor fault detection, analysis and processing system
CN120403965A
