Device defect prediction and model training method and device, apparatus, and storage medium

Abstract

Embodiments of the present application disclose a device defect prediction and model training method, device, equipment and storage medium, the method comprising: inputting a sample monitoring sequence to a device defect prediction model to be trained; the device defect prediction model comprising: a learning layer, a prediction layer and an output layer; learning the time-dependent features of the sample monitoring sequence through the learning layer, and generating an adjacency matrix of the sample monitoring sequence based on the time-dependent features; obtaining the graph convolution features of the sample monitoring sequence based on the graph convolution operation according to the adjacency matrix through the prediction layer; obtaining the training prediction result of the device defect based on the graph convolution features through the output layer; and adjusting the model parameters of the device defect prediction model based on the training prediction result of the device defect. Embodiments of the present application can train a device defect prediction model that can accurately predict different types of defects of a device, and accurately predict different types of defects of a device in a timely manner.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a method, apparatus, device, and storage medium for predicting equipment defects and training models. Background Technology

[0002] Industrial equipment used in production processes often has numerous components and complex structures. Some equipment is even internally sealed, such as gas-insulated metal-enclosed switchgear. When such highly complex equipment malfunctions, the types of defects are diverse and difficult to detect promptly and directly, thus affecting the normal and safe operation of production. Existing methods for predicting equipment defects include prediction methods relying on expert experience, methods based on periodic inspection results, and methods using convolutional neural networks. However, existing prediction methods have poor accuracy in identifying different types of defects. Summary of the Invention

[0003] This invention provides a method, apparatus, device, and storage medium for equipment defect prediction and model training, which can train a high-precision equipment defect prediction model to predict different types of equipment defects, and make timely and accurate predictions of different types of equipment defects.

[0004] In a first aspect, embodiments of the present invention provide a method for training a device defect prediction model, comprising:

[0005] The sample monitoring sequence is input into the equipment defect prediction model to be trained. The equipment defect prediction model includes a learning layer, a prediction layer and an output layer. The sample monitoring sequence includes multiple historical monitoring sequences. Each historical monitoring sequence is a time series of historical monitoring data obtained by continuously monitoring parameters related to equipment defects. The continuous monitoring duration corresponding to each historical monitoring sequence is a preset duration.

[0006] The learning layer learns the time-dependent features of the sample monitoring sequence and generates the adjacency matrix of the sample monitoring sequence based on the time-dependent features.

[0007] The prediction layer obtains the graph convolution features of the sample monitoring sequence based on the adjacency matrix using graph convolution operations.

[0008] The training prediction results of device defects are obtained through the output layer based on the graph convolutional features; and

[0009] The value of the training loss function is determined based on the training prediction results of the equipment defects, and the model parameters of the equipment defect prediction model are adjusted according to the value of the training loss function.

[0010] Secondly, embodiments of the present invention provide a method for predicting equipment defects, including:

[0011] The real-time monitoring sequence of the equipment is input into the trained equipment defect prediction model; the equipment defect prediction model includes a learning layer, a prediction layer and an output layer, and the real-time monitoring sequence is a time series of monitoring data obtained from monitoring parameters related to equipment defects from a preset time period to the current time;

[0012] The learning layer learns the time dependency features of the real-time monitoring sequence, and generates the adjacency matrix of the real-time monitoring sequence based on the time dependency features;

[0013] The prediction layer obtains the graph convolution features of the real-time monitoring sequence based on the adjacency matrix using graph convolution operations; and

[0014] The output layer obtains real-time prediction results of device defects based on the graph convolution features.

[0015] Thirdly, embodiments of the present invention provide a training apparatus for a device defect prediction model, comprising:

[0016] The sample input module is used to input the sample monitoring sequence into the equipment defect prediction model to be trained. The equipment defect prediction model includes a learning layer, a prediction layer and an output layer. The sample monitoring sequence includes multiple historical monitoring sequences. Each historical monitoring sequence is a time series of historical monitoring data obtained by continuously monitoring parameters related to equipment defects. The continuous monitoring duration corresponding to each historical monitoring sequence is a preset duration.

[0017] The sample graph structure learning module is used to learn the time dependency features of the sample monitoring sequence through the learning layer, and generate the adjacency matrix of the sample monitoring sequence based on the time dependency features;

[0018] The sample graph convolution module is used to obtain the graph convolution features of the sample monitoring sequence based on the adjacency matrix using the prediction layer and graph convolution operations; and

[0019] The sample prediction module is used to obtain the training prediction results of device defects based on the graph convolutional features through the output layer; and

[0020] The parameter adjustment module is used to determine the function value of the training loss function based on the training prediction results of the equipment defects, and to adjust the model parameters of the equipment defect prediction model according to the function value of the training loss function.

[0021] Fourthly, embodiments of the present invention also provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the device defect prediction model training method or the device defect prediction method as described in any of the embodiments of the present invention.

[0022] Fifthly, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the equipment defect prediction model training method or the equipment defect prediction method as described in any of the embodiments of the present invention.

[0023] This invention provides a method, apparatus, device, and storage medium for equipment defect prediction and model training. By training the equipment defect prediction model, the time dependency features of sample monitoring sequences are learned, and an adjacency matrix is ​​generated based on these features. Then, graph convolution features of the sample monitoring sequences are obtained based on the adjacency matrix using graph convolution operations. Further, the training prediction results for equipment defects are determined based on these graph convolution features. This allows for the analysis of the time dependency features of sample monitoring sequences through graph convolution operations to identify potentially complex and irregular relationships between different types of equipment defects and parameter time series, thereby improving the prediction accuracy of the equipment defect prediction model. By using the equipment defect prediction model trained according to this invention to obtain real-time prediction results for equipment defects, the accuracy of real-time prediction of different types of equipment defects can be improved based on the relationship between the time series of different types of equipment defects and parameters. This enables timely prevention or detection of various equipment defects and faults, avoiding impacts on the normal and safe operation of production. Attached Figure Description

[0024] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a flowchart illustrating the equipment defect prediction model training method provided in an embodiment of the present invention;

[0026] Figure 2 This is another flowchart illustrating the equipment defect prediction model training method provided in this embodiment of the invention;

[0027] Figure 3 This is another flowchart illustrating the equipment defect prediction model training method provided in this embodiment of the invention;

[0028] Figure 4This is a flowchart illustrating the equipment defect prediction method provided in an embodiment of the present invention;

[0029] Figure 5 This is a schematic diagram of a device for training a equipment defect prediction model provided in an embodiment of the present invention;

[0030] Figure 6 This is a schematic diagram of the equipment defect prediction device provided in an embodiment of the present invention;

[0031] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0032] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0033] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0034] Figure 1 This is a flowchart illustrating a method for training a device defect prediction model according to an embodiment of the present invention. This method can be executed by a device defect prediction model training device provided in this embodiment, which can be implemented in software and / or hardware. In a specific embodiment, the device can be integrated into an electronic device, such as a computer or server. The following embodiments will illustrate this using the integration of the device into an electronic device as an example. (Reference) Figure 1 The method may specifically include the following steps:

[0035] Step 101: Input the sample monitoring sequence into the equipment defect prediction model to be trained. The equipment defect prediction model includes a learning layer, a prediction layer, and an output layer. The sample monitoring sequence includes multiple historical monitoring sequences. Each historical monitoring sequence is a time series of historical monitoring data obtained by continuously monitoring parameters related to equipment defects. The continuous monitoring duration corresponding to each historical monitoring sequence is a preset duration. This step facilitates the learning of the adjacency matrix of the sample monitoring sequences through the learning layer.

[0036] Specifically, the aforementioned equipment defect-related parameters can be understood as operating parameters related to equipment defects. The aforementioned equipment can be understood as highly complex industrial equipment, specifically internally sealed equipment, such as gas-insulated switchgear (GIS).

[0037] Optionally, equipment defect-related parameters include multiple operating parameters of the equipment, specifically including: temperature, humidity, operating voltage, and operating current during equipment operation.

[0038] Specifically, the aforementioned equipment defect-related parameters may also include only one operating parameter.

[0039] Specifically, the preset duration can be determined based on the monitoring and collection frequency of equipment defect-related parameters. For example, when the monitoring and collection frequency of equipment defect-related parameters is high, the preset duration can be set to a smaller value to avoid excessive monitoring data in the sample monitoring sequence, which could affect the accuracy of model training; when the collection frequency of equipment-related parameters is low, the preset duration can be set to a larger value to obtain sufficient sample monitoring sequences.

[0040] Optionally, the sample monitoring sequence corresponding to a continuous monitoring period includes multiple simultaneous time-segment sequences that correspond one-to-one with multiple operating parameters.

[0041] Specifically, a sample monitoring sequence corresponding to a continuous monitoring period can also include a monitoring sequence that integrates information from the above-mentioned multiple operating parameters, such as a time series composed of matrices.

[0042] Optionally, before step 101, the original historical monitoring data is obtained and preprocessed by deduplication, anomaly removal, etc., to obtain preprocessed historical data, and the above sample monitoring sequence is generated using the preprocessed historical data.

[0043] Step 102 involves learning the time dependency features of the sample monitoring sequences through a learning layer and generating an adjacency matrix for the sample monitoring sequences based on these features. This step determines the graph structure of the sample monitoring sequences by learning their time dependency features and represents it using an adjacency matrix. This facilitates the acquisition of graph convolution features of the sample monitoring sequences by the prediction layer based on the adjacency matrix using graph convolution operations. Consequently, based on the time dependency features of the sample monitoring sequences, the analysis can determine the potentially complex and irregular relationships between the time series of defects and parameters of different types of equipment.

[0044] In an optional specific embodiment of the present invention, the process of generating the adjacency matrix of the sample monitoring sequence based on time-dependent features includes: learning the time-dependent features of the sample monitoring sequence through a gated recurrent unit, and adjusting the self-attention of the time-dependent features through a self-attention mechanism to obtain the adjacency matrix.

[0045] Alternatively, the time-dependent features can be processed using other methods in the prior art, such as correction using correction coefficients, to obtain the aforementioned adjacency matrix.

[0046] Specifically, the dependency characteristics of sample monitoring sequences can also be learned through long short-term memory networks.

[0047] In an optional instance, the process of learning the time-dependent features of the sample monitoring sequence through the gated recurrent unit includes: obtaining the last hidden state h of the gated recurrent unit corresponding to the sample monitoring sequence. f The dependency characteristics of the above sample monitoring sequences were obtained.

[0048] Optionally, the process of obtaining the adjacency matrix by adjusting the self-attention of time-dependent features through the self-attention mechanism can be based on the following formula:

[0049]

[0050] Where A represents the adjacency matrix, Q represents the query vector, and K represents the key vector. Q and K can be mapped to learnable parameters A in the attention mechanism. Q and A K Generates using linear projection, where d is the size of the hidden dimensions of Q and K.

[0051] Step 103 involves obtaining the graph convolution features of the sample monitoring sequence through the prediction layer based on the adjacency matrix and graph convolution operations. This step obtains the graph convolution features of the sample monitoring sequence based on the adjacency matrix determined by the time dependency features of the sample monitoring sequence. This allows for the analysis and determination of potentially complex and irregular relationships between the time series of different types of equipment defects and parameters, and facilitates the acquisition of training prediction results for equipment defects through the output layer based on the graph convolution features.

[0052] Optionally, the prediction layer may include one or more graph convolutional modules.

[0053] Optionally, the above-mentioned graph convolution modules correspond one-to-one with the above-mentioned operating parameters, and the above-mentioned graph convolution modules can be approximately superimposed in first order.

[0054] In an optional specific embodiment of the present invention, the process of obtaining the graph convolutional features of the sample monitoring sequence by the prediction layer based on the adjacency matrix and graph convolutional operation includes:

[0055] The Laplacian matrix of the sample monitoring sequence is obtained from the adjacency matrix; the graph Fourier transform of the sample monitoring sequence is performed on the sample monitoring sequence based on the Laplacian matrix using the graph convolution module to obtain the spectral data corresponding to the sample monitoring sequence; the convolution operation is performed on the spectral data using the graph convolution module to obtain convolutional spectral data; and graph convolution features are obtained based on the convolutional spectral data.

[0056] Optionally, the above process of obtaining graph convolution features based on convolutional spectral data includes: obtaining convolutional spectral data as graph convolution features.

[0057] Optionally, the above process of obtaining graph convolution features based on convolutional spectral data includes: performing an inverse graph Fourier transform on the above convolutional spectral data according to the above Laplacian matrix to obtain the above graph convolution features.

[0058] Specifically, the graph convolutional features of the sample monitoring sequence can be obtained based on the following formula:

[0059] F=ΘLx=Θ(UΛU T )x=UΘ(Λ)U T x

[0060] Where F represents the graph convolution feature, Θ represents the convolution kernel, L represents the Laplacian matrix, Λ represents the diagonal matrix composed of the eigenvalues ​​of L, U is the orthogonal matrix composed of the eigenvectors of L, and x represents the sample monitoring sequence.

[0061] Optionally, the process described above of obtaining the graph convolutional features of the sample monitoring sequence through the prediction layer based on the adjacency matrix and graph convolutional operations includes:

[0062] The adjacency matrix is ​​processed into a one-way linked form to obtain a one-way linked adjacency matrix; and the graph convolution features of the sample monitoring sequence are obtained based on the one-way linked adjacency matrix using graph convolution operations.

[0063] Specifically, by processing the adjacency matrix into a unidirectional link, it is possible to reduce the computational power required during model training and final prediction, thereby improving efficiency.

[0064] Step 104: Obtain the training prediction results of equipment defects through graph convolutional features in the output layer. This step allows for adjusting the model parameters of the equipment defect prediction model based on the training prediction results.

[0065] Specifically, the training prediction results of the aforementioned device defects can be directly determined through fully connected layers based on graph convolutional features.

[0066] Step 105: Determine the value of the training loss function based on the training prediction results of equipment defects, and adjust the model parameters of the equipment defect prediction model according to the value of the training loss function. This step, building upon steps 101 to 104, learns the time dependency features of the sample monitoring sequences and generates an adjacency matrix based on these features. Then, based on the adjacency matrix, it obtains the graph convolution features of the sample monitoring sequences using graph convolution operations. Further, it determines the training prediction results of equipment defects based on these graph convolution features. This allows for the analysis of the time dependency features of the sample monitoring sequences through graph convolution operations to identify the potentially complex and irregular relationships between different types of equipment defects and the time series of parameters, thereby improving the prediction accuracy of the equipment defect prediction model.

[0067] Optionally, the training loss function mentioned above can be the mean squared error loss function, which can be expressed as:

[0068]

[0069] Among them, W θ Here, y represents the adjustable model parameters, and x represents the training prediction result of the equipment defect. t+1 This represents the true value of the equipment defect.

[0070] Specifically, the training loss function mentioned above can also be other loss functions in the prior art, such as the cross-entropy loss function.

[0071] The following further describes the equipment defect prediction model training method provided in the embodiments of the present invention.

[0072] Optionally, the sample monitoring sequence corresponding to a continuous monitoring period includes multiple simultaneous time-segment sequences corresponding one-to-one with multiple operating parameters, and the prediction layer includes multiple graph convolutional modules corresponding one-to-one with the aforementioned multiple operating parameters. When multiple graph convolutional modules are first-order approximately superimposed, such as... Figure 2 As shown, it may include:

[0073] Step 201: Input multiple simultaneous time sequences of the equipment into the equipment defect prediction model to be trained.

[0074] Step 202: By using gated loop units, the correlation between the parameter data and the running time in each simultaneous sequence is learned, and the time dependency features of each simultaneous sequence are obtained.

[0075] Step 203: Adjust the temporal dependency features of each simultaneous sequence using a self-attention mechanism to obtain the adjacency matrix of each simultaneous sequence.

[0076] Step 204: Using the corresponding graph convolution module, obtain the graph convolution features of each simultaneous sequence based on the corresponding adjacency matrix and graph convolution operation, and obtain the graph convolution features of the sample monitoring sequence based on the graph convolution features of each simultaneous sequence.

[0077] Step 205: Obtain the training prediction results of device defects through the output layer based on graph convolutional features.

[0078] Step 206: Determine the value of the training loss function based on the training prediction results of equipment defects, and adjust the model parameters of the equipment defect prediction model according to the value of the training loss function.

[0079] Specifically, in this embodiment of the invention, graph structure learning and corresponding graph convolution features are obtained by performing graph structure learning on multiple simultaneous sequences that correspond one-to-one with multiple operating parameters. Then, the depth spatial information of the sample monitoring sequence is recovered by a first-order approximately superimposed graph convolution module. This can expand the existing equipment defect data, enabling the training of a high-precision equipment defect prediction model without the need for a large amount of equipment defect data. This solves the technical problem in the prior art where the equipment defect dataset, especially the defect dataset of gas-insulated metal-enclosed switchgear, is small and cannot meet the large-scale requirements of deep learning models.

[0080] The following further describes the equipment defect prediction model training method provided in the embodiments of the present invention, such as... Figure 3 As shown, that is Figure 3 Step 104 may include the following steps:

[0081] Step 1041: Perform a discrete Fourier transform on the graph convolutional features to obtain frequency basis data.

[0082] Step 1042: The frequency basis data features are obtained by learning the feature representation of the frequency basis data through gated linear units.

[0083] Step 1043: Perform inverse discrete Fourier transform and inverse graphical Fourier transform on the frequency basis data features to obtain the output layer features.

[0084] Step 1044: Determine the training prediction results of device defects based on the output layer features through the fully connected layer.

[0085] Specifically, the feature representation of the above graph convolution features can be learned directly through gated linear learning units to obtain the above output layer features.

[0086] Specifically, by further learning the feature representations in graph convolution features through gated linear units (LLUs), the temporally relevant features in graph convolution features can be learned more effectively. Furthermore, after performing a Discrete Fourier Transform on the graph convolution features, learning the transformed data through LLUs can accelerate the learning process of the LLUs without sacrificing computational accuracy.

[0087] Figure 4 This is a flowchart illustrating a device defect prediction method provided in an embodiment of the present invention. The method can be executed by the device defect prediction method provided in this embodiment, and the device can be implemented in software and / or hardware. In a specific embodiment, the device can be integrated into an electronic device, such as a computer or server. The following embodiments will illustrate this using the integration of the device into an electronic device as an example. (Reference) Figure 4 The method may specifically include the following steps:

[0088] Step 401: Input the real-time monitoring sequence of the equipment into the trained equipment defect prediction model. The equipment defect prediction model includes a learning layer, a prediction layer and an output layer. The real-time monitoring sequence is a time series of monitoring data obtained from monitoring parameters related to equipment defects from a preset time period to the current time.

[0089] Optionally, the aforementioned equipment defect-related parameters include multiple operating parameters of the equipment, specifically including: temperature, humidity, operating voltage, and operating current during equipment operation.

[0090] Optionally, the above real-time monitoring sequence may include multiple real-time parameter sequences that correspond one-to-one with the above multiple operating parameters.

[0091] Step 402: Learn the time dependency features of the real-time monitoring sequence through the learning layer, and generate the adjacency matrix of the real-time monitoring sequence based on the time dependency features.

[0092] Optionally, the process of generating the adjacency matrix of the real-time monitoring sequence based on the time-dependent features mentioned above includes: learning the time-dependent features of the real-time monitoring sequence through a gated recurrent unit, and obtaining the adjacency matrix by adjusting the self-attention of the time-dependent features through a self-attention mechanism.

[0093] Optionally, the above process of learning the time dependency features of the real-time monitoring sequence through a gated recurrent unit and adjusting the self-attention of the time dependency features to obtain the adjacency matrix includes: learning the correlation between each parameter data and the running time in each simultaneous sequence through a gated recurrent unit to obtain the time dependency features of each simultaneous sequence, and adjusting the time dependency features of each simultaneous sequence through a self-attention mechanism to obtain the adjacency matrix of each simultaneous sequence.

[0094] Step 403: Obtain the graph convolution features of the real-time monitoring sequence based on the adjacency matrix and graph convolution operation through the prediction layer.

[0095] Optionally, the prediction layer may include one or more graph convolutional modules.

[0096] Optionally, the above-mentioned graph convolution modules correspond one-to-one with the above-mentioned operating parameters, and the above-mentioned graph convolution modules can be approximately superimposed in first order.

[0097] Optionally, the process of obtaining graph convolutional features of the real-time monitoring sequence based on the adjacency matrix using the prediction layer includes: obtaining the Laplacian matrix of the real-time monitoring sequence based on the adjacency matrix; performing a graph Fourier transform on the real-time monitoring sequence based on the Laplacian matrix using the graph convolution module to obtain the spectral data corresponding to the real-time monitoring sequence; performing a convolution operation on the spectral data using the graph convolution module to obtain convolutional spectral data; and obtaining graph convolutional features based on the convolutional spectral data.

[0098] Optionally, the above process of obtaining graph convolution features based on convolutional spectral data includes: obtaining convolutional spectral data as graph convolution features.

[0099] Optionally, the process described above of obtaining graph convolutional features of the real-time monitoring sequence based on graph convolutional operations using the adjacency matrix through the prediction layer includes: processing the adjacency matrix into a one-way linked form to obtain a one-way linked adjacency matrix, and obtaining the graph convolutional features of the real-time monitoring sequence based on graph convolutional operations using the one-way linked adjacency matrix.

[0100] Optionally, the process of obtaining the graph convolution features of the real-time monitoring sequence by the prediction layer based on the adjacency matrix and graph convolution operation includes: obtaining the graph convolution features of each simultaneous sequence based on the corresponding adjacency matrix by the corresponding graph convolution module, and obtaining the graph convolution features of the real-time monitoring sequence based on the graph convolution features of each simultaneous sequence.

[0101] Step 404: Obtain real-time prediction results of equipment defects through graph convolution features in the output layer.

[0102] Optionally, the process of obtaining real-time prediction results of equipment defects based on graph convolutional features through the output layer includes: performing discrete Fourier transform on the graph convolutional features to obtain frequency basis data; learning the feature representation of the frequency basis data through gated linear units to obtain frequency basis data features; performing inverse discrete Fourier transform and inverse graph Fourier transform on the frequency basis data features to obtain output layer features; and determining the training prediction results of equipment defects based on the output layer features through a fully connected layer.

[0103] The equipment defect prediction method provided in this invention obtains real-time prediction results of equipment defects by using the equipment defect prediction model trained in this invention. Based on the relationship between the time series of different types of equipment defects and parameters, it can provide high accuracy for real-time prediction of different types of equipment defects, thereby timely preventing or detecting various equipment defects and failures and avoiding impact on the normal and safe operation of production.

[0104] Figure 5 This is a structural diagram of a device for training an equipment defect prediction model provided in an embodiment of the present invention. This device is suitable for executing the equipment defect prediction model training method provided in an embodiment of the present invention. Figure 5 As shown, the device may specifically include:

[0105] The sample input module 501 is used to input sample monitoring sequences into the equipment defect prediction model to be trained. The equipment defect prediction model includes a learning layer, a prediction layer, and an output layer. The sample monitoring sequences include multiple historical monitoring sequences, each of which is a time series of historical monitoring data obtained by continuously monitoring parameters related to equipment defects. The continuous monitoring duration corresponding to each historical monitoring sequence is a preset duration. This module can facilitate the learning of the adjacency matrix of the sample monitoring sequences through the learning layer.

[0106] Optionally, the aforementioned equipment defect-related parameters include multiple operating parameters of the equipment, specifically including: temperature, humidity, operating voltage, and operating current during equipment operation.

[0107] Optionally, the above real-time monitoring sequence may include multiple real-time parameter sequences that correspond one-to-one with the above multiple operating parameters.

[0108] Optionally, the sample monitoring sequence corresponding to a continuous monitoring period includes multiple simultaneous time-segment sequences that correspond one-to-one with multiple operating parameters.

[0109] The sample graph structure learning module 502 is used to learn the time dependency features of sample monitoring sequences through a learning layer and generate an adjacency matrix of the sample monitoring sequences based on these features. This module can determine the graph structure of the sample monitoring sequences by learning their time dependency features and representing it with an adjacency matrix. This facilitates the acquisition of graph convolution features of the sample monitoring sequences through graph convolution operations using the adjacency matrix in the prediction layer. Therefore, based on the time dependency features of the sample monitoring sequences, it can analyze and determine the potentially complex and irregular relationships between the time series of defects and parameters of different types of equipment.

[0110] Optionally, the sample graph structure learning module 502 described above can be specifically used to learn the time-dependent features of the sample monitoring sequence through a gated recurrent unit, and to obtain the adjacency matrix by adjusting the self-attention of the time-dependent features through a self-attention mechanism.

[0111] Optionally, the sample graph structure learning module 502 can be specifically used to learn the correlation between the parameter data and the running time in each simultaneous sequence through the gated recurrent unit, so as to obtain the time dependency features of each simultaneous sequence; and to adjust the time dependency features of each simultaneous sequence through the self-attention mechanism to obtain the adjacency matrix of each simultaneous sequence.

[0112] The sample graph convolution module 503 is used to obtain the graph convolution features of the sample monitoring sequence based on the adjacency matrix through graph convolution operations in the prediction layer. This module can obtain the graph convolution features of the sample monitoring sequence based on the adjacency matrix determined by the time dependency features of the sample monitoring sequence. It can analyze and determine the complex and irregular relationships that may exist between the time series of different types of equipment defects and parameters based on the time dependency features of the sample monitoring sequence, and facilitate the acquisition of the training prediction results of equipment defects through the output layer based on the graph convolution features.

[0113] Optionally, the above-mentioned sample graph convolution module 503 can be specifically used to: obtain the Laplacian matrix of the sample monitoring sequence based on the adjacency matrix; perform a graph Fourier transform on the sample monitoring sequence based on the Laplacian matrix using the graph convolution module to obtain the spectral data corresponding to the sample monitoring sequence; perform a convolution operation on the spectral data using the graph convolution module to obtain convolutional spectral data; and obtain graph convolution features based on the convolutional spectral data.

[0114] Optionally, the above-mentioned sample graph convolution module 503 can be specifically used to obtain convolutional spectrum data as graph convolutional features.

[0115] Optionally, the prediction layer may include one or more graph convolutional modules.

[0116] Optionally, the above-mentioned graph convolution modules correspond one-to-one with the above-mentioned operating parameters, and the above-mentioned graph convolution modules can be approximately superimposed in first order.

[0117] Optionally, the above-mentioned sample graph convolution module 503 can be specifically used to process the adjacency matrix into a one-way linked form to obtain a one-way linked adjacency matrix; and to obtain the graph convolution features of the sample monitoring sequence based on the one-way linked adjacency matrix using graph convolution operations.

[0118] Optionally, the above-mentioned sample graph convolution module 503 can be specifically used to obtain the graph convolution features of each simultaneous sequence based on the graph convolution operation through the corresponding graph convolution module according to the corresponding adjacency matrix, and obtain the graph convolution features of the sample monitoring sequence based on the graph convolution features of each simultaneous sequence.

[0119] The sample prediction module 504 is used to obtain the training prediction results of equipment defects through graph convolutional features in the output layer. This module can help adjust the model parameters of the equipment defect prediction model based on the training prediction results of equipment defects.

[0120] Optionally, the sample prediction module 504 can be specifically used to: perform discrete Fourier transform on the graph convolution features to obtain frequency basis data; learn the feature representation of the frequency basis data through gated linear units to obtain frequency basis data features; perform inverse discrete Fourier transform and inverse graph Fourier transform on the frequency basis data features to obtain output layer features; and determine the training prediction results of equipment defects based on the output layer features through a fully connected layer.

[0121] The parameter adjustment module 505 is used to determine the value of the training loss function based on the training prediction results of equipment defects, and to adjust the model parameters of the equipment defect prediction model according to the value of the training loss function. This module, in conjunction with modules 501 to 503, learns the time-dependency features of the sample monitoring sequences and generates an adjacency matrix based on these features. Then, based on the adjacency matrix, it obtains the graph convolution features of the sample monitoring sequences using graph convolution operations. Furthermore, it determines the training prediction results of equipment defects based on these graph convolution features. This allows for the analysis of the time-dependency features of the sample monitoring sequences through graph convolution operations to identify the potentially complex and irregular relationships between different types of equipment defects and the time series of parameters, thereby improving the prediction accuracy of the equipment defect prediction model.

[0122] Figure 6 This is a structural diagram of an equipment defect prediction device provided in an embodiment of the present invention. This device is suitable for executing the equipment defect prediction method provided in an embodiment of the present invention. Figure 6 As shown, the device may specifically include:

[0123] The real-time input module 601 is used to input the real-time monitoring sequence of the device into the trained device defect prediction model. The device defect prediction model includes a learning layer, a prediction layer and an output layer. The real-time monitoring sequence is a time series of monitoring data obtained from monitoring parameters related to device defects from a preset time period to the current time.

[0124] The real-time graph structure learning module 602 is used to learn the time dependency features of the real-time monitoring sequence through the learning layer, and generate the adjacency matrix of the real-time monitoring sequence based on the time dependency features.

[0125] Real-time graph convolution module 603 is used to obtain graph convolution features of the real-time monitoring sequence based on the adjacency matrix using graph convolution operations through the prediction layer; and

[0126] The real-time prediction module 604 is used to obtain real-time prediction results of equipment defects through graph convolution features of the output layer.

[0127] The equipment defect prediction device provided in this embodiment of the invention can obtain real-time prediction results of equipment defects by using the equipment defect prediction model trained in this embodiment of the invention. It can provide high accuracy for real-time prediction of different types of equipment defects based on the relationship between the time series of different types of equipment defects and parameters, thereby timely preventing or detecting various equipment defects and failures and avoiding impact on the normal and safe operation of production.

[0128] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is merely an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the functional modules described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0129] This invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the device defect prediction model training method or the device defect prediction method provided in any of the above embodiments.

[0130] This invention also provides a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the equipment defect prediction model training method or the equipment defect prediction method provided in any of the above embodiments.

[0131] This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the equipment defect prediction model training method or the equipment defect prediction method as described in any of the embodiments of this invention.

[0132] The following is for reference. Figure 7 It shows a schematic diagram of the structure of a computer system 700 suitable for implementing an electronic device according to embodiments of the present invention. Figure 7 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0133] like Figure 7 As shown, the computer system 700 includes a central processing unit (CPU) 701, which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) 702 or programs loaded from storage section 708 into random access memory (RAM) 703. The RAM 703 also stores various programs and data required for the operation of the system 700. The CPU 701, ROM 702, and RAM 703 are interconnected via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0134] The following components are connected to the I / O interface 705: an input section 706 including a keyboard, mouse, etc.; an output section 707 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 708 including a hard disk, etc.; and a communication section 709 including a network interface card such as a LAN card, modem, etc. The communication section 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I / O interface 705 as needed. A removable medium 711, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 710 as needed so that computer programs read from it can be installed into the storage section 708 as needed.

[0135] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 709, and / or installed from removable medium 711. When the computer program is executed by central processing unit (CPU) 701, it performs the functions defined above in the system of this invention.

[0136] It should be noted that the computer-readable medium shown in this invention can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0137] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0138] The modules and / or units described in the embodiments of the present invention can be implemented in software or hardware. The described modules and / or units can also be housed in a processor. For example, a processor may be described as including a sample input module, a sample graph structure learning module, a sample graph convolution module, a sample prediction module, and a parameter adjustment module; or, a processor may be described as including a real-time input module, a real-time graph structure learning module, a real-time graph convolution module, and a real-time prediction module. The names of these modules do not necessarily constitute a limitation on the module itself under certain circumstances.

[0139] In another aspect, the present invention also provides a computer-readable medium, which may be included in the device described in the above embodiments; or it may exist independently and not assembled into the device. The computer-readable medium carries one or more programs, which, when executed by the device, cause the device to include: inputting sample monitoring sequences into a device defect prediction model to be trained; the device defect prediction model includes: a learning layer, a prediction layer, and an output layer; the sample monitoring sequences include multiple historical monitoring sequences, each historical monitoring sequence being a time series of historical monitoring data obtained by continuously monitoring parameters related to device defects, and the continuous monitoring duration corresponding to each historical monitoring sequence being a preset duration; learning the time dependency features of the sample monitoring sequences through the learning layer, and generating an adjacency matrix of the sample monitoring sequences based on the time dependency features; obtaining graph convolution features of the sample monitoring sequences through the prediction layer based on graph convolution operations according to the adjacency matrix; obtaining the training prediction result of the device defect through the output layer based on the graph convolution features; and determining the function value of a training loss function based on the training prediction result of the device defect, and adjusting the model parameters of the device defect prediction model according to the function value of the training loss function.

[0140] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for training a device defect prediction model, characterized in that, include: Input the sample monitoring sequence into the equipment defect prediction model to be trained; The equipment defect prediction model includes a learning layer, a prediction layer, and an output layer. The sample monitoring sequence includes multiple historical monitoring sequences. Each historical monitoring sequence is a time series of historical monitoring data obtained by continuously monitoring parameters related to equipment defects. The continuous monitoring duration corresponding to each historical monitoring sequence is a preset duration. The learning layer learns the time-dependent features of the sample monitoring sequence and generates the adjacency matrix of the sample monitoring sequence based on the time-dependent features. The prediction layer obtains the graph convolution features of the sample monitoring sequence based on the adjacency matrix using graph convolution operations. The training prediction results of device defects are obtained through the output layer based on the graph convolutional features; and The value of the training loss function is determined based on the training prediction results of the equipment defects, and the model parameters of the equipment defect prediction model are adjusted according to the value of the training loss function. The step of generating the adjacency matrix of the sample monitoring sequence based on the time-dependent features includes: The temporal dependence features of the sample monitoring sequence are learned through a gated recurrent unit, and the adjacency matrix is ​​obtained by adjusting the self-attention of the temporal dependence features through a self-attention mechanism. The equipment defect-related parameters include multiple operating parameters. The sample monitoring sequence corresponding to a continuous monitoring period includes multiple simultaneous sequences that correspond one-to-one with the multiple operating parameters. The prediction layer includes multiple graph convolutional modules that correspond one-to-one with the multiple operating parameters. The multiple graph convolutional modules are approximately superimposed in first order. The step of learning the time-dependent features of the sample monitoring sequence through a gated recurrent unit and obtaining the adjacency matrix by adjusting the self-attention of the time-dependent features through a self-attention mechanism includes: The gated loop unit learns the correlation between each parameter data and running time in each simultaneous sequence, thereby obtaining the time dependency characteristics of each simultaneous sequence; and The adjacency matrix of each simultaneous sequence is obtained by adjusting the temporal dependency features of each simultaneous sequence through a self-attention mechanism. The step of obtaining the graph convolutional features of the sample monitoring sequence through the prediction layer based on the adjacency matrix using graph convolutional operations includes: The graph convolution module obtains the graph convolution features of each simultaneous sequence based on the corresponding adjacency matrix and graph convolution operation, and obtains the graph convolution features of the sample monitoring sequence based on the graph convolution features of each simultaneous sequence.

2. The equipment defect prediction model training method according to claim 1, characterized in that, The step of obtaining the graph convolutional features of the sample monitoring sequence based on the adjacency matrix using graph convolutional operations includes: The adjacency matrix is ​​processed into a unidirectional link form to obtain a unidirectional link adjacency matrix; The graph convolution features of the sample monitoring sequence are obtained based on the unidirectional link adjacency matrix and graph convolution operation.

3. The equipment defect prediction model training method according to claim 1, characterized in that, The step of obtaining the graph convolutional features of the sample monitoring sequence through the prediction layer based on the adjacency matrix using graph convolutional operations includes: The Laplace matrix of the sample monitoring sequence is obtained based on the adjacency matrix; The graph convolution module performs a graph Fourier transform on the sample monitoring sequence based on the Laplacian matrix to obtain the spectral data corresponding to the sample monitoring sequence; The graph convolution module performs a convolution operation on the spectral data to obtain convolutional spectral data; and The graph convolution features are obtained based on the convolutional spectral data.

4. The equipment defect prediction model training method according to claim 3, characterized in that, The step of obtaining the graph convolutional features based on the convolutional spectral data includes: The convolutional spectral data is obtained as the graph convolutional feature; The step of obtaining the training prediction result of device defects through the output layer based on the graph convolutional features includes: The frequency basis data is obtained by performing a discrete Fourier transform on the convolutional features of the graph. The frequency basis data features are obtained by learning the feature representation of the frequency basis data through a gated linear unit; The output layer features are obtained by performing inverse discrete Fourier transform and inverse graphical Fourier transform on the frequency basis data features; and The training prediction results of the device defects are determined by the fully connected layer based on the features of the output layer.

5. A method for predicting equipment defects, characterized in that, include: Input the real-time monitoring sequence of the equipment into the trained equipment defect prediction model; The equipment defect prediction model is trained using the equipment defect prediction model training method described in any one of claims 1 to 4; The equipment defect prediction model includes a learning layer, a prediction layer, and an output layer. The real-time monitoring sequence is a time series of monitoring data obtained from monitoring parameters related to equipment defects from a preset time period to the current time. The learning layer learns the time dependency features of the real-time monitoring sequence, and generates the adjacency matrix of the real-time monitoring sequence based on the time dependency features; The prediction layer obtains the graph convolution features of the real-time monitoring sequence based on the adjacency matrix using graph convolution operations; and The output layer obtains real-time prediction results of device defects based on the graph convolution features.

6. A training device for a equipment defect prediction model, characterized in that, include: The sample input module is used to input the sample monitoring sequence into the equipment defect prediction model to be trained; The equipment defect prediction model includes a learning layer, a prediction layer, and an output layer. The sample monitoring sequence includes multiple historical monitoring sequences. Each historical monitoring sequence is a time series of historical monitoring data obtained by continuously monitoring parameters related to equipment defects. The continuous monitoring duration corresponding to each historical monitoring sequence is a preset duration. The sample graph structure learning module is used to learn the time dependency features of the sample monitoring sequence through the learning layer, and generate the adjacency matrix of the sample monitoring sequence based on the time dependency features; The sample graph convolution module is used to obtain the graph convolution features of the sample monitoring sequence based on the adjacency matrix using the prediction layer and graph convolution operations; and The sample prediction module is used to obtain the training prediction results of device defects based on the graph convolutional features through the output layer; and The parameter adjustment module is used to determine the function value of the training loss function based on the training prediction results of the equipment defects, and to adjust the model parameters of the equipment defect prediction model according to the function value of the training loss function; The sample graph structure learning module is specifically used to learn the time-dependent features of the sample monitoring sequence through a gated recurrent unit, and to adjust the self-attention of the time-dependent features through a self-attention mechanism to obtain the adjacency matrix. The equipment defect-related parameters include multiple operating parameters. The sample monitoring sequence corresponding to a continuous monitoring period includes multiple simultaneous sequences that correspond one-to-one with the multiple operating parameters. The prediction layer includes multiple graph convolutional modules that correspond one-to-one with the multiple operating parameters. The multiple graph convolutional modules are approximately superimposed in first order. The step of learning the time-dependent features of the sample monitoring sequence through a gated recurrent unit and obtaining the adjacency matrix by adjusting the self-attention of the time-dependent features through a self-attention mechanism includes: The gated loop unit learns the correlation between each parameter data and running time in each simultaneous sequence, thereby obtaining the time dependency characteristics of each simultaneous sequence; and The adjacency matrix of each simultaneous sequence is obtained by adjusting the temporal dependency features of each simultaneous sequence through a self-attention mechanism. The step of obtaining the graph convolutional features of the sample monitoring sequence through the prediction layer based on the adjacency matrix using graph convolutional operations includes: The graph convolution module obtains the graph convolution features of each simultaneous sequence based on the corresponding adjacency matrix and graph convolution operation, and obtains the graph convolution features of the sample monitoring sequence based on the graph convolution features of each simultaneous sequence.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the equipment defect prediction model training method as described in any one of claims 1 to 4 or the equipment defect prediction method as described in claim 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the equipment defect prediction model training method as described in any one of claims 1 to 4 or the equipment defect prediction method as described in claim 5.

Images