Fault Prediction Method and Device for Aging Test Equipment Drive Control Detection Board

Abstract

The embodiments of this specification provide a method and device for fault prediction of an aging test equipment drive control detection board. The method for fault prediction of the aging test equipment drive control detection board includes: obtaining historical signal values corresponding to analog quantity signals of the aging test equipment drive control detection board at multiple historical time series; calculating the similarity between different degradation paths of the aging test equipment drive control detection board, where different degradation paths are composed of historical signal values corresponding to different historical time series; clustering different degradation paths based on the similarity to generate at least two clustering results; respectively using the historical signal values corresponding to the degradation paths included in different clustering results as training data to input into corresponding fault prediction models to be trained for training, and obtaining the trained fault prediction models; the number of fault prediction models to be trained is equal to the number of at least two clustering results, and each fault prediction model to be trained corresponds to each clustering result one by one.

Description

Technical Field

[0001] The embodiments of this specification relate to the technical field of non-destructive reliability screening of integrated circuits, and particularly to a method for predicting faults of a driving control detection board for an aging test device. Background Art

[0002] An integrated circuit high-temperature aging test device accelerates various physical and chemical reaction processes inside components by continuously applying a certain electrical stress to the components for a long time, prompting various potential faults inside the components to be exposed early, so as to eliminate early failure products and enable electronic components to enter a period with low and relatively stable failure rates from the beginning of use.

[0003] At present, the integrated circuit high-temperature aging test bench can achieve long-term monitoring of the test environment by means of increasing in-machine long-term monitoring, over-stress protection mechanisms, etc., and can respond in a timely manner when the machine fails, realizing after-sales maintenance based on fault data. However, due to the lack of active guarantee technology for integrated circuit high-temperature aging test benches at present, it is difficult to achieve the integrity of the test process and the consistency of the test environment stress, and it is extremely easy to cause major property losses such as the destruction of millions of test devices due to the forced interruption of the test process caused by machine failures, or the aging test is recognized as a failure test due to adverse effects such as additional stress introduced during the test period due to machine performance degradation, resulting in ineffective waste of resources. At the same time, ensuring the quality of high-temperature aging tests can avoid over-aging and under-aging faults of integrated circuits, greatly reduce the failure rate of integrated circuits, and is conducive to avoiding the failure shutdown of large-scale electronic systems such as new energy vehicles, civil airliners, and energy storage substations using the same integrated circuits due to integrated circuit failures.

[0004] Since the integrated circuit high-temperature aging test device operates under harsh working conditions such as high temperature for a long time, the failure risk of its key equipment is relatively high. Therefore, the fault prediction and health management of the key equipment of the integrated circuit high-temperature aging test device have become an important research direction. Effective health assessment and fault prediction can promote the development of reliable maintenance plans, thereby preventing potential faults.

[0005] The key equipment of the integrated circuit high-temperature aging test device, such as the driving control detection board, is mainly used for generating secondary power and detecting voltage and current, generating and driving digital signals, generating and driving analog signals, temperature signal feedback circuits, etc. Due to the core role of drive control detection in the system, once it fails, it will have a serious impact on the entire system, not only causing the test to fail, but also possibly causing equipment damage and triggering safety accidents. Therefore, how to predict faults in drive control detection has become an urgent problem to be solved. Summary of the Invention

[0006] In view of this, the embodiments of the present specification provide a method for predicting faults of a driving control detection board of an aging test device. One or more embodiments of the present specification also relate to a device for predicting faults of a driving control detection board of an aging test device, a computing device, a computer-readable storage medium, and a computer program, so as to solve the technical defects existing in the prior art.

[0007] According to the first aspect of the embodiments of the present specification, there is provided a method for predicting faults of a driving control detection board of an aging test device, including:

[0008] Obtaining historical signal values corresponding to analog signals of a driving control detection board of an aging test device at multiple historical time series;

[0009] Calculating the similarity between different degradation paths of the driving control detection board of the aging test device, where different degradation paths are composed of historical signal values corresponding to different historical time series;

[0010] Clustering different degradation paths based on the similarity to generate at least two clustering results;

[0011] Respectively taking the historical signal values corresponding to the degradation paths included in different clustering results as training data and inputting them into corresponding fault prediction models to be trained for training, so as to obtain trained fault prediction models;

[0012] Wherein, the number of fault prediction models to be trained is equal to the number of the at least two clustering results, and each fault prediction model to be trained corresponds to each clustering result one by one.

[0013] Optionally, the fault prediction model to be trained is a multi-dimensional recurrent neural network model, including an input layer, an intermediate layer, and an output layer. The intermediate layer includes a sequentially connected near neighbor component analysis model structure and a multi-dimensional recurrent neural network model structure. The multi-dimensional recurrent neural network model structure includes a fully connected layer, a bidirectional long short-term memory network layer, a bidirectional gated recurrent unit layer, and a splicing layer.

[0014] Optionally, the calculating the similarity between different degradation paths of the driving control detection board of the aging test device includes:

[0015] Processing the historical signal values corresponding to different degradation paths of the driving control detection board of the aging test device through the near neighbor component analysis model structure to obtain the similarity between different degradation paths, where the near neighbor component analysis model structure processes the historical signal values corresponding to different degradation paths of the driving control detection board of the aging test device through a near neighbor component analysis algorithm.

[0016] Optionally, the intermediate layer includes a sequentially connected nearest neighbor component analysis model structure, a Gaussian mixture model structure, and a multi-dimensional recurrent neural network model structure;

[0017] Correspondingly, the clustering of different degradation paths based on the similarity includes:

[0018] Clustering different degradation paths based on the similarity through the Gaussian mixture model structure, where the Gaussian mixture model structure clusters different degradation paths through the Gaussian mixture algorithm and based on the similarity.

[0019] Optionally, the calculating the similarity between different degradation paths of the aging test equipment drive control detection board includes:

[0020] Determining at least two signal parameters of the aging test equipment drive control detection board based on the historical signal values;

[0021] Performing feature extraction on the at least two signal parameters to obtain feature vectors corresponding to different degradation paths;

[0022] Calculating the similarity between the feature vectors corresponding to each degradation path to obtain the similarity between different degradation paths of the aging test equipment drive control detection board.

[0023] Optionally, the at least two signal parameters include: mean, standard deviation, variance, skewness, peak-to-peak value.

[0024] Optionally, the fault prediction method for the aging test equipment drive control detection board further includes:

[0025] Obtaining the signal values corresponding to the analog signals of the aging test equipment drive control detection board in a target time series;

[0026] Determining the target clustering result to which the target degradation path composed of the signal values belongs;

[0027] Determining a target fault prediction model corresponding to the target clustering result, inputting the signal values into the target fault prediction model, and obtaining the fault prediction result of the aging test equipment drive control detection board output by the target fault prediction model.

[0028] According to the second aspect of the embodiments of the present specification, there is provided a fault prediction device for an aging test equipment drive control detection board, including:

[0029] An acquisition module, configured to acquire historical signal values corresponding to the analog signals of the aging test equipment drive control detection board in a plurality of historical time series;

[0030] A calculation module, configured to calculate the similarity between different degradation paths of the driving control detection board of the aging test equipment, where the different degradation paths are composed of historical signal values corresponding to different historical time series;

[0031] A clustering module, configured to cluster different degradation paths based on the similarity to generate at least two clustering results;

[0032] A training module, configured to respectively use the historical signal values corresponding to the degradation paths included in different clustering results as training data to input into corresponding to-be-trained fault prediction models for training to obtain trained fault prediction models;

[0033] Wherein, the number of to-be-trained fault prediction models is equal to the number of the at least two clustering results, and each to-be-trained fault prediction model corresponds to each clustering result one by one.

[0034] According to the third aspect of the embodiments of the present specification, a computing device is provided, including:

[0035] A memory and a processor;

[0036] The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the steps of the fault prediction method for the driving control detection board of the aging test equipment according to any one of the above.

[0037] According to the fourth aspect of the embodiments of the present specification, a computer-readable storage medium is provided, which stores computer-executable instructions, and when the instructions are executed by a processor, the steps of the fault prediction method for the driving control detection board of the aging test equipment according to any one of the above are implemented.

[0038] According to the fifth aspect of the embodiments of the present specification, a computer program is provided, wherein when the computer program is executed on a computer, the computer is made to execute the steps of the fault prediction method for the driving control detection board of the aging test equipment described above.

[0039] The fault prediction method for the aging test equipment drive control detection board provided in the embodiments of this specification includes: obtaining the historical signal values corresponding to the analog quantity signals of the aging test equipment drive control detection board at multiple historical time series; calculating the similarity between different degradation paths of the aging test equipment drive control detection board, where different degradation paths are composed of the historical signal values corresponding to different historical time series; clustering different degradation paths based on the similarity to generate at least two clustering results; respectively using the historical signal values corresponding to the degradation paths included in different clustering results as training data and inputting them into the corresponding fault prediction model to be trained for training, so as to obtain the trained fault prediction model; where the number of fault prediction models to be trained is equal to the number of the at least two clustering results, and each fault prediction model to be trained corresponds to each clustering result one by one. Through this processing method, considering the fault mode as a factor, identifying different degradation paths of the aging test equipment drive control detection board, and bringing the historical signal values of the analog quantity signals of the aging test equipment drive control detection board with the same type of degradation paths into the same model for training and prediction is beneficial to improving the accuracy of the fault prediction results of the aging test equipment drive control detection board. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 is a flowchart of a fault prediction method for an aging test equipment drive control detection board provided in an embodiment of this specification;

[0041] Figure 2 is a processing flowchart of the training process of a fault prediction model to be trained provided in an embodiment of this specification;

[0042] Figure 3 is a processing flowchart of the processing process of a fault prediction model provided in an embodiment of this specification;

[0043] Figure 4 is a schematic structural diagram of a fault prediction device for an aging test equipment drive control detection board provided in an embodiment of this specification;

[0044] Figure 5 is a structural block diagram of a computing device provided in an embodiment of this specification. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0045] In the following description, many specific details are set forth in order to provide a thorough understanding of this specification. However, this specification can be implemented in many other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the connotation of this specification. Therefore, this specification is not limited by the specific implementations disclosed below.

[0046] The terms used in one or more embodiments of this specification are for the purpose of describing specific embodiments only and are not intended to limit one or more embodiments of this specification. The singular forms "a", "the", and "said" used in one or more embodiments of this specification and the appended claims are also intended to include the plural forms unless the context clearly dictates otherwise. It should also be understood that the term "and / or" used in one or more embodiments of this specification refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0047] It should be understood that although the terms first, second, etc. may be used in one or more embodiments of this specification to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other. For example, without departing from the scope of one or more embodiments of this specification, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "when" or "while" or "in response to determining".

[0048] First, the noun terms related to one or more embodiments of this specification are explained.

[0049] Neighbourhood Component Analysis (NCA): It is a method of supervised learning mainly used for metric learning and dimensionality reduction. Its core idea is to learn a linear space transformation matrix to maximize the average leave-one-out classification effect in the new transformed space.

[0050] Bidirectional Long Short-Term Memory (BiLSTM): It is a special structure of recurrent neural network that combines two Long Short-Term Memory (LSTM) layers in two directions, one processing the forward direction of the time series (from the past to the future) and the other processing the reverse direction of the time series (from the future to the past), so as to be able to obtain context information before and after simultaneously when processing sequence data.

[0051] Remaining Useful Life (RUL): It mainly refers to the remaining useful life after the system has run for a period of time. Accurately predicting the remaining useful life of the system can greatly reduce the losses caused by system downtime and improve the operating reliability of the system.

[0052] Current high-temperature aging products can achieve long-term monitoring of the test environment by means of increasing in-machine long-term monitoring, over-stress protection mechanisms, etc., and can respond in a timely manner when the product fails, realizing after-sales maintenance based on fault data. However, due to the lack of active guarantee technology for high-temperature aging products, it is difficult for existing aging products to achieve the integrity of the test process and the consistency of the test environment stress, which is extremely likely to lead to major property losses such as the destruction of millions of test devices due to the forced interruption of the test process caused by product failures, or the aging test being recognized as a failure test due to adverse effects such as additional stress introduced during the test due to product performance degradation, resulting in the ineffective waste of resources. At the same time, ensuring the quality of high-temperature aging tests can avoid over-aging and under-aging failures of integrated circuits, greatly reduce the failure rate of integrated circuits, and avoid large-scale electronic system failures and shutdowns such as new energy vehicles, civil airliners, and energy storage substations caused by integrated circuit failures.

[0053] Therefore, the development of an intelligent guarantee system for test quality not only has important economic value, but also can help China's high-end test equipment move towards the forefront of the world.

[0054] Currently, it is difficult for active guarantee technology to adapt to high-temperature aging products, and the key difficulties mainly include: (1) It is difficult to determine the health benchmark due to inconsistent test environments; (2) It is difficult to calculate the performance degradation trends of different levels due to the multi-structural levels of the product; (3) It is difficult to construct a fault self-healing strategy due to the complex composition of the fault source.

[0055] In addition, for integrated circuit high-temperature aging test benches, the current fault prediction mainly focuses on the individual differences and degradation characteristics of equipment or systems, and there is less research on the impact of different degradation paths of drive control detection boards on the prediction results. However, due to the unstable situations such as multiple degradation paths of drive control detection boards in general, the current prediction methods will have two problems: (1) Multiple degradation paths: There are multiple degradation paths for drive control detection boards, which are reflected as multiple performance degradation paths on the aging test bench. The impacts of different performance degradation paths on the remaining service life of the equipment are different, that is, they will affect the prediction results; (2) The voltage parameter time series of drive control detection boards is relatively long, and it is difficult to capture and learn the degradation information in the series, which will also affect the prediction results. Due to the existence of the above two problems, it is currently impossible to carry out accurate fault prediction for the drive control detection boards of aging test equipment.

[0056] In view of the above problems, the embodiments of this specification utilize a large language model for fault diagnosis and prediction of integrated circuit high-temperature aging equipment to ensure that this technology has the capabilities of processing and analyzing multi-source heterogeneous data, and dynamically mining degradation and fault characteristics, so as to solve the problems of the obliteration of fault characteristics under the coupling interference of thermal-electric multi-physical fields, the fuzziness of failure paths, and the inaccurate prediction of degradation trends caused by environmental stress fluctuations, and realize system self-diagnosis and self-warning, thereby solving the problem that it is difficult to calculate the performance degradation trends of different levels in the case of multiple product structural levels in real time.

[0057] The embodiments of this specification provide a method for fault prediction of a drive control detection board of an aging test equipment under multiple degradation paths to solve the problem that accurate fault prediction cannot be carried out on the drive control detection board.

[0058] In this specification, a method for fault prediction of a drive control detection board of an aging test equipment is provided. This specification also relates to a device for fault prediction of a drive control detection board of an aging test equipment, a computing device, a computer-readable storage medium, and a computer program, which will be described in detail one by one in the following embodiments.

[0059] Figure 1 The flowchart of a method for fault prediction of a drive control detection board of an aging test equipment provided according to an embodiment of this specification is shown, which specifically includes the following steps.

[0060] Step 102: Obtain the historical signal values corresponding to the analog signals of the drive control detection board of the aging test equipment in multiple historical time series.

[0061] Specifically, the aging test equipment is an integrated circuit high-temperature aging test bench, which accelerates various physical and chemical reaction processes inside the components by continuously applying a certain electrical stress to the components for a long time, promotes the early exposure of various potential faults inside the components, thereby eliminating early failure products, and enabling the electronic components to enter a period with low and relatively stable failure rates from the beginning of use. The drive control detection board is mainly used to provide the following functions: generation of secondary power supply and voltage and current detection, generation and drive of digital signals, generation and drive of analog signals, and temperature signal feedback circuit.

[0062] Since the drive control detection board of the aging test equipment generates voltage monitoring signals during operation, therefore, the analog signals described in the embodiments of this specification can be voltage signals.

[0063] In addition, each historical time series described in the embodiments of this specification can be composed of time points in the time period corresponding to each complete use process of the drive control detection board.

[0064] Step 104: Calculate the similarity between different degradation paths of the drive control detection board of the aging test equipment, where the different degradation paths are composed of historical signal values corresponding to different historical time series.

[0065] In an alternative embodiment, the calculating the similarity between different degradation paths of the drive control detection board of the aging test equipment includes:

[0066] Determine at least two signal parameters of the drive control detection board of the aging test equipment based on the historical signal values;

[0067] Extract features from the at least two signal parameters to obtain feature vectors corresponding to different degradation paths;

[0068] Calculate the similarity between the feature vectors corresponding to each degradation path to obtain the similarity between different degradation paths of the drive control detection board of the aging test equipment.

[0069] Among them, the at least two signal parameters include: mean, standard deviation, variance, skewness, peak-to-peak value.

[0070] In addition, the embodiment of the present specification trains the to-be-trained fault prediction model through the obtained historical signal values, and the to-be-trained fault prediction model is a multi-dimensional recurrent neural network model, including an input layer, an intermediate layer and an output layer. The intermediate layer includes a sequentially connected neighborhood component analysis model structure and a multi-dimensional recurrent neural network model structure. The multi-dimensional recurrent neural network model structure includes a fully connected layer, a bidirectional long short-term memory network layer, a bidirectional gated recurrent unit layer and a splicing layer.

[0071] Among them, the bidirectional long short-term memory network (BiLSTM) is used to process multi-sensor monitoring signals, the splicing layer is used to fuse the outputs of the bidirectional long short-term memory network and the bidirectional gated recurrent unit (BiGRU), and the output layer is a fully connected layer, that is, a Dense layer, which is used to predict the remaining useful life RUL. Therefore, the multi-dimensional recurrent neural network uses the bidirectional LSTM (BiLSTM layer) and the bidirectional GRU (BiGRU layer) to extract time series features, and then splices the outputs of the two, and finally outputs the predicted remaining useful life.

[0072] Correspondingly, the calculating the similarity between different degradation paths of the drive control detection board of the aging test equipment includes:

[0073] Process the historical signal values corresponding to different degradation paths of the drive control detection board of the aging test equipment through the near neighbor component analysis model structure to obtain the similarity between different degradation paths, where the near neighbor component analysis model structure processes the historical signal values corresponding to different degradation paths of the drive control detection board of the aging test equipment through the near neighbor component analysis algorithm.

[0074] The processing flow chart of the training process of a to-be-trained fault prediction model provided by an embodiment of this specification is as Figure 2 shown.

[0075] Specifically, during the actual operation of the drive control detection board of the aging test equipment, its available usage duration does not linearly decrease with the degradation process (usage process). Instead, after being used for a period of time and reaching its lifespan limit, the remaining available usage duration gradually decreases further. Therefore, when predicting the fault (available usage duration prediction) of the drive control detection board of the aging test equipment, it can be considered that the initial degradation of the drive control detection board of the aging test equipment can be ignored. During the initial degradation process, its remaining available usage duration is a constant value, and after reaching the lifespan critical value, its remaining available usage duration linearly decreases.

[0076] Based on this, according to the principle of piecewise linear degradation, after obtaining the historical signal values corresponding to multiple historical time series in an embodiment of this specification, at least two signal parameters of the drive control detection board of the aging test equipment can also be determined based on the historical signal values. Specifically, the mean, standard deviation, variance, skewness, peak-to-peak value, etc. corresponding to the analog signal can be determined, so as to train the near neighbor component analysis model structure in the to-be-trained fault prediction model based on the signal parameters.

[0077] In practical applications, after determining at least two signal parameters, feature extraction can also be performed on the at least two signal parameters to obtain feature vectors corresponding to different degradation paths, and then similarity calculation is performed on the feature vectors corresponding to each degradation path to obtain the similarity between different degradation paths of the drive control detection board of the aging test equipment.

[0078] Specifically, multiple historical signal values included in the same degradation path can be determined first, then the signal parameters corresponding to these multiple historical signal values are determined, and then a feature vector corresponding to this degradation path is constructed based on the parameter values corresponding to the signal parameters, so as to realize dimensionality reduction feature extraction of multiple historical signal values.

[0079] After feature extraction is completed, the health state features and degradation state features can be divided based on the feature extraction results, and the health state features (including mean, standard deviation, variance, skewness, peak-to-peak value) are regarded as the original input of the health set of the nearest neighbor component analysis model structure, and the degradation state features (including mean, standard deviation, variance, skewness, peak-to-peak value) are regarded as the original input of the degradation set, and the health and degradation labels are used as the model output.

[0080] Among them, the degradation state features can be expressed as , where , i is the number of usage cycles of the detection board of the aging test equipment, and j drive control is the dimension of the degradation state feature (each time the drive control detection board is used, the dimension of its degradation state feature is fixed, but the value of the degradation state feature will change with the change of the usage cycle).

[0081] If the first 10% of the features in the feature extraction results of the drive control detection board of the aging test equipment are used as the health set, and the latter 90% of the features are used as the degradation set, then the health set and the degradation set can be expressed as:

[0082]

[0083]

[0084] Among them, and are the processed features.

[0085]

[0086]

[0087] Based on the division of the health set and the degradation set, the setting of their labels is carried out. The state label can be expressed as , where is the state label corresponding to the health state, is the state label corresponding to the degradation state, and each degradation feature of the drive control detection board of the aging test equipment corresponds to a state label. Therefore, the state labels corresponding to the health set and the degradation set can be expressed as:

[0088]

[0089]

[0090] Using the features of the health set and the degradation set as the input of the nearest neighbor component analysis model, and the state labels corresponding to the features As the model output, by setting the maximum number of iterations of the model, the model parameters of the nearest neighbor component analysis model are optimized, so as to effectively divide and identify the health and degradation states of the drive control detection board of the aging test equipment, and obtain the learned optimized Mahalanobis distance matrix Q.

[0091] Among them, in the adaptive mapping of degradation features, the core is to adaptively map the input features to a new space that conforms to the feature state according to the nearest neighbor component analysis model, so as to reduce the aliasing of health and fault features caused by data fluctuations or working condition changes, making the metric calculation between health and degradation more stable, more robust, and the distance representation more accurate.

[0092] Map the features of the training sample set and the test sample set The one-dimensional mapped features of the drive control detection board of the aging test equipment can be expressed as:

[0093]

[0094]

[0095] Among them, the mapped features of each drive control detection board of the aging test equipment can also be expressed as a set of health mapped features and degradation mapped features , .

[0096] Calculate the similarity of the characteristic time series trajectories such as mean, standard deviation, variance, skewness, and peak-to-peak value between different degradation paths through the nearest neighbor component analysis algorithm. Subsequently, clustering of the time series trajectories can be achieved through the clustering method, so as to classify the drive control detection board of the aging test equipment according to the performance degradation trajectory.

[0097] Based on the performance parameters such as mean, standard deviation, variance, skewness, and peak-to-peak value extracted from the dimensionality reduction of the voltage time series signal of the drive control detection board in the embodiments of this specification, the advantages of the nearest neighbor component analysis (NCA) algorithm are adopted. By scaling the time axis of the unknown sequence to be the same length as the template sequence, the similarity between the sample performance degradation paths can be better grasped.

[0098] Step 106: Cluster different degradation paths based on the similarity to generate at least two clustering results.

[0099] In an alternative embodiment, on the basis that the to-be-trained fault prediction model is a multi-dimensional recurrent neural network model and the model includes an input layer, an intermediate layer, and an output layer, the intermediate layer includes a nearest neighbor component analysis model structure, a Gaussian mixture model structure, and a multi-dimensional recurrent neural network model structure connected in sequence;

[0100] Accordingly, clustering different degradation paths based on the similarity includes:

[0101] The different degradation paths are clustered based on the similarity by using the Gaussian mixture model structure, wherein the Gaussian mixture model structure clusters the different degradation paths based on the similarity by using a Gaussian mixture algorithm.

[0102] Specifically, the embodiment of this specification takes advantage of the time used for cluster centroid selection in Gaussian mixture and the complexity of cluster overlap space, and uses the optimal point that actually exists in the data set as its centroid, so that the trajectory corresponding to the centroid can be found.

[0103] Among them, the implementation process of the Gaussian mixture algorithm is as follows:

[0104] Algorithm principle: Assume a point in a high-dimensional space (dimension n) If the distribution of these points is approximately an ellipsoid, a single Gaussian density function can be used To describe the probability density function that produces these data:

[0105]

[0106] in is the mean, the center point of the density function; is the covariance matrix of the density function.

[0107] However, if the distribution of these points is not ellipsoidal, it is not appropriate to use a single Gaussian probability density function to describe the probability density function of the distribution of these points. At this time, a workaround can be used to express it by the weighted average of several single Gaussian probability density functions, that is, the Gaussian mixture model (GMM). The Gaussian mixture model is defined as follows:

[0108]

[0109] Where m is the mixing number of the model; is the weight coefficient of the mixed model, and ; is the i-th single Gaussian probability density function.

[0110] The unknown parameters in the mixed Gaussian distribution can usually be evaluated by the expectation maximization EM algorithm (Expectation Maximization), and the main steps are divided into two stages: E and M.

[0111] The Gaussian mixture method is used to cluster system samples through the performance degradation path similarity matrix of the aging test equipment drive control detection board, so that the aging test equipment drive control detection board can be identified according to the performance degradation trajectory time series.

[0112] Step 108: Respectively take the historical signal values corresponding to the degradation paths included in different clustering results as training data and input them into the corresponding fault prediction models to be trained for training, so as to obtain the trained fault prediction models.

[0113] Among them, the number of fault prediction models to be trained is equal to the number of the at least two clustering results, and each fault prediction model to be trained corresponds to each clustering result one by one.

[0114] Specifically, after clustering different degradation paths to generate at least two clustering results, the number of clustering results can be determined first, and then the corresponding number of fault prediction models to be trained can be constructed, and the multi-dimensional recurrent neural network model structures in the constructed fault prediction models to be trained are respectively trained by using the historical signal values corresponding to the degradation paths included in each clustering result; among them, the historical signal values corresponding to the degradation paths included in each clustering result are used to train a fault prediction model to be trained, and the fault prediction models trained by the historical signal values corresponding to the degradation paths included in each clustering result are different.

[0115] For example, if two clustering results are generated, two fault prediction models to be trained are constructed, and the multi-dimensional recurrent neural network model structure in one fault prediction model to be trained can be trained by using the historical signal values corresponding to the degradation paths included in the first clustering result, and the multi-dimensional recurrent neural network model structure in the other fault prediction model to be trained can be trained by using the historical signal values corresponding to the degradation paths included in the second clustering result.

[0116] In practical applications, the multi-dimensional recurrent neural network model structure uses momentum and adaptive learning rate to accelerate the convergence speed, and is proved to be superior to the classical stochastic gradient descent optimizer, so as to obtain higher processing efficiency and prediction quality.

[0117] The specific implementation process is as follows:

[0118] 1) Establish at least two fault prediction models to be trained, corresponding to multiple performance degradation paths. According to the performance degradation paths, respectively take the historical signal values corresponding to different degradation paths of the drive control detection board as the input of the multi-dimensional recurrent neural network model structure in the fault prediction model to be trained.

[0119] In practical applications, after determining at least two signal parameters of the drive control detection board of the aging test equipment, the at least two signal parameters can also be used as the model input. Construct a training set , where K is the number of training samples, represents the input data matrix obtained from the monitoring data time series according to the time window, with a size of N, where is the sensor measurement value of s selected sensors at time t.

[0120] 2) Concatenate the operation condition data in series to construct a high - order vector . Then, feed the vector into the stacked BiLSTM layer and BiGRU layer to obtain hidden feature maps of different dimensions.

[0121] 3) Set the number of BiLSTM layers and BiGRU layers to be the same, that is, the total number of stacked BiGRU layers is also M. For the last BiLSTM layer and BiGRU layer, use the output of the last BiLSTM layer and BiGRU layer at the last time as the final output of the last BiLSTM layer and BiGRU layer.

[0122] 4) Concatenate the hidden feature vectors into a combined feature vector. Finally, input the combined feature vector into two other linear regression dense layers to generate the predicted RUL.

[0123] 5) Calculate the loss using the error between the predicted RUL value and the actual RUL value as the loss function, update the relevant parameters in the network through backpropagation, and repeat the training until the loss meets the requirements. The prediction accuracy is used to evaluate the performance of the fault prediction model. In the embodiments of this specification, the definition of the prediction accuracy is as follows:

[0124]

[0125] where RUL is the actual RUL (actual remaining available usage duration), is the predicted RUL (predicted remaining available usage duration) estimated by the fault prediction model, and TUL is the actual total usage duration of the target circuit, which is defined as the total number of test cycles when the aging test equipment drives the control detection board to reach the failure threshold.

[0126] In an alternative embodiment, the fault prediction method for the aging test equipment drive control detection board further includes:

[0127] Obtain the signal values corresponding to the analog signal of the aging test equipment drive control detection board in the target time series;

[0128] Determine the target clustering result to which the target degradation path composed of the signal values belongs according to the signal values;

[0129] Determine the target fault prediction model corresponding to the target clustering result, input the signal values into the target fault prediction model, and obtain the fault prediction result of the aging test equipment drive control detection board output by the target fault prediction model.

[0130] Specifically, after the model training is completed, the signal values corresponding to the analog signal of the aging test equipment drive control detection board in the target time series can be obtained, and the signal values are input into the fault prediction model. Then, the signal values are processed through the nearest neighbor component analysis model structure in the fault prediction model, and the feature vectors corresponding to the target degradation path formed by the signal values are determined according to the processing results. Next, the similarity between the target degradation path and the aforementioned multiple different degradation paths is calculated based on the feature vectors. Then, the target clustering result to which the target degradation path belongs is determined through the Gaussian mixture model structure in the fault prediction model, where the target clustering result belongs to one of the aforementioned at least two clustering results. After that, the target fault prediction model corresponding to the target clustering result is determined, the signal value is input into the target fault prediction model, and the signal value or the mean, standard deviation, variance, skewness, and peak-to-peak value determined based on the signal value are processed through the multi-dimensional recurrent neural network model structure in the target fault prediction model to generate the fault prediction result of the aging test equipment drive control detection board.

[0131] In the case where it is determined according to the fault prediction result that the aging test equipment drive control detection board is in a degraded state, the method of automatically optimizing the weights of multi-distance metric fusion is used to quantitatively calculate the health state of the signal value, obtain the health degree of the degraded state at the current moment, and use the health degree of the degraded state at the current moment as the final fault detection result of the aging test equipment drive control detection board.

[0132] In an alternative embodiment, using the method of automatically optimizing the weights of multi-distance metric fusion to quantitatively calculate the health state of the signal value and obtain the health degree of the degraded state at the current moment includes:

[0133] Calculate the Mahalanobis distance, cosine similarity, and Manhattan distance between the signal value and the pre-acquired healthy baseline data respectively;

[0134] Through multi-distance metric fusion with automatically optimized weights for the Mahalanobis distance, the cosine similarity, and the Manhattan distance, the health degree of the degraded state at the current moment is obtained.

[0135] Among them, the healthy baseline data is constructed based on the historical signal values corresponding to all fault types.

[0136] In an alternative embodiment, through multi-distance metric fusion of the Mahalanobis distance, the cosine similarity, and the Manhattan distance, obtaining the health degree of the degraded state at the current moment includes:

[0137]

[0138] Among them, represents the health degree of the degraded state at the current moment represents the Mahalanobis distance; represents the cosine similarity; represents the Manhattan distance; , and respectively represent weights, and + + = 1.

[0139] Alternatively, when it is determined according to the fault prediction result that the drive control detection board of the aging test equipment is in a degraded state, based on the probability space point after mapping of the multi-dimensional feature sequence and the health baseline in the residual convolutional neural network model, the state degradation curve of the drive control detection board of the aging test equipment is determined, wherein the state degradation curve can characterize the health state of the drive control detection board of the aging test equipment; the multi-dimensional feature sequence is generated by extracting features from historical signal values.

[0140] In an alternative embodiment, the residual convolutional neural network model is trained as follows:

[0141] Obtain the training time-series degradation feature samples of the drive control detection board of the aging test equipment; wherein each training time-series degradation feature sample is labeled with a degradation feature state;

[0142] Input the training time-series degradation feature samples into a pre-established residual convolutional neural network model to obtain the first probability space point after mapping;

[0143] Discriminate the degradation starting point of the drive control detection board of the aging test equipment according to the two-dimensional first probability space point, and generate the health baseline of the drive control detection board of the aging test equipment.

[0144] In the actual implementation process, the training time-series degradation feature samples of the drive control detection board of the aging test equipment can be expressed as , the maximum number of iterations max_iter. Mark the degradation feature state corresponding to the training drive control detection board of the aging test equipment , including the healthy state and the degraded state.

[0145] The processing procedure flowchart of a fault prediction model provided by the embodiments of this specification is as Figure 3 shown, and specifically includes the following steps.

[0146] Step 302: Obtain historical signal values.

[0147] Step 304: Construct a fault prediction model to be trained.

[0148] Step 306: Initialize network parameters.

[0149] Step 308: Calculate the loss function.

[0150] Step 310: Model training.

[0151] Step 312: Determine whether one training cycle is completed.

[0152] Specifically, after model training, it can be determined whether one training cycle is completed; if so, execute Step 314; if not, return to execute Step 308.

[0153] Step 314: Determine whether all training cycles are completed.

[0154] Specifically, after determining that one training cycle is completed, it can continue to determine whether all training cycles are completed; if so, execute Step 318; if not, return to execute Step 308.

[0155] Step 316: Obtain the target signal value.

[0156] Step 318: Complete model training.

[0157] After obtaining the target signal value, the target signal value can be input into the trained fault prediction model for processing. Among them, the target signal value is the signal value corresponding to the analog signal of the drive control detection board of the aging test equipment in the target time series.

[0158] Step 320: Predict the remaining service life.

[0159] Step 322: True data of the remaining service life.

[0160] Step 324: Calculate the model prediction error.

[0161] Specifically, after processing the target signal value through the fault prediction model and obtaining the predicted remaining service life output by the fault prediction model, the actual remaining service life (i.e., the true data of the remaining service life) of the drive control detection board of the aging test equipment can also be obtained, and then the prediction error of the fault prediction model is calculated based on the predicted remaining service life and the actual remaining service life.

[0162] Step 326: Evaluate the model prediction result.

[0163] In the embodiments of this specification, the similarity between different degradation paths of the drive control detection board of aging test equipment with multiple different degradation rates and different fault starting degrees is evaluated through the nearest neighbor component analysis algorithm. On this basis, the Gaussian mixture clustering algorithm is introduced to achieve rapid and accurate classification of different degradation paths of the drive control detection board of the aging test bench. In addition, the embodiments of this specification propose a fault prediction model for the drive control detection board of aging test equipment based on a multi-dimensional recurrent neural network model. By comprehensively considering the influence of the degradation path on the prediction result of the remaining available service life of the drive control detection board of the aging test equipment, high-precision fault prediction is achieved.

[0164] The embodiments of this specification provide an active guarantee technology for integrated circuit high-temperature aging test benches to achieve the integrity of the test process and the consistency of test environmental stress, and can reduce the significant property losses caused by the destruction of millions of test devices due to forced interruption during the test process caused by machine failures, or can reduce the adverse effects such as the introduction of additional stress during the test due to machine performance degradation, so as to avoid the aging test being recognized as a failure test, which is conducive to avoiding waste of resources. At the same time, ensuring the quality of high-temperature aging tests can avoid over-aging and under-aging faults of integrated circuits, significantly reduce the failure rate of integrated circuits, and is conducive to avoiding the failure shutdown of large-scale electronic systems such as new energy vehicles, civil airliners, and energy storage power transmission stations using the same integrated circuits due to integrated circuit failures.

[0165] In addition, the embodiments of this specification utilize a large language model for fault diagnosis and prediction for integrated circuit high-temperature aging equipment to ensure that this technology has the ability to process and analyze multi-source heterogeneous data and the ability to dynamically mine degradation and fault characteristics, so as to solve the problems of the disappearance of fault characteristics, the ambiguity of failure paths, and the inaccurate prediction of degradation trends caused by thermal-electric multi-physical field coupling interference, and achieve system self-diagnosis and self-warning.

[0166] The fault prediction method for the aging test equipment drive control detection board provided by the embodiments of this specification obtains the historical signal values corresponding to the analog signals of the aging test equipment drive control detection board at multiple historical time series; calculates the similarity between different degradation paths of the aging test equipment drive control detection board, where different degradation paths are composed of historical signal values corresponding to different historical time series; clusters different degradation paths based on the similarity to generate at least two clustering results; respectively inputs the historical signal values corresponding to the degradation paths included in different clustering results as training data into the corresponding fault prediction models to be trained for training, and obtains the trained fault prediction models; where the number of fault prediction models to be trained is equal to the number of the at least two clustering results, and each fault prediction model to be trained corresponds to each clustering result one by one. Through this processing method, taking the fault mode as a consideration factor, different degradation paths of the aging test equipment drive control detection board are identified, and the historical signal values of the analog signals of the aging test equipment drive control detection board with the same type of degradation paths are brought into the same model for training and prediction, which is beneficial to improving the accuracy of the fault prediction results of the aging test equipment drive control detection board.

[0167] Corresponding to the above method embodiment, this specification also provides an embodiment of a fault prediction device for an aging test equipment drive control detection board. Figure 4 FIG. shows a schematic structural diagram of a fault prediction device for an aging test equipment drive control detection board provided by an embodiment of this specification. As Figure 4 shown, the device includes:

[0168] An acquisition module 402, configured to acquire the historical signal values corresponding to the analog signals of the aging test equipment drive control detection board at multiple historical time series;

[0169] A calculation module 404, configured to calculate the similarity between different degradation paths of the aging test equipment drive control detection board, where different degradation paths are composed of historical signal values corresponding to different historical time series;

[0170] A clustering module 406, configured to cluster different degradation paths based on the similarity to generate at least two clustering results;

[0171] A training module 408, configured to respectively input the historical signal values corresponding to the degradation paths included in different clustering results as training data into the corresponding fault prediction models to be trained for training, and obtain the trained fault prediction models;

[0172] Where the number of fault prediction models to be trained is equal to the number of the at least two clustering results, and each fault prediction model to be trained corresponds to each clustering result one by one.

[0173] Optionally, the fault prediction model to be trained is a multi-dimensional recurrent neural network model, including an input layer, an intermediate layer, and an output layer. The intermediate layer includes a sequentially connected nearest neighbor component analysis model structure and a multi-dimensional recurrent neural network model structure. The multi-dimensional recurrent neural network model structure includes a fully connected layer, a bidirectional long short-term memory network layer, a bidirectional gated recurrent unit layer, and a splicing layer.

[0174] Optionally, the calculation module 404 is further configured to:

[0175] Process the historical signal values corresponding to different degradation paths of the aging test equipment drive control detection board through the nearest neighbor component analysis model structure to obtain the similarity between different degradation paths, where the nearest neighbor component analysis model structure processes the historical signal values corresponding to different degradation paths of the aging test equipment drive control detection board through the nearest neighbor component analysis algorithm.

[0176] Optionally, the intermediate layer includes a sequentially connected nearest neighbor component analysis model structure, a Gaussian mixture model structure, and a multi-dimensional recurrent neural network model structure;

[0177] Correspondingly, the clustering module is further configured to:

[0178] Cluster different degradation paths based on the similarity through the Gaussian mixture model structure, where the Gaussian mixture model structure clusters different degradation paths through the Gaussian mixture algorithm and based on the similarity.

[0179] Optionally, the calculation module 404 is further configured to:

[0180] Determine at least two signal parameters of the aging test equipment drive control detection board based on the historical signal values;

[0181] Extract features from the at least two signal parameters to obtain feature vectors corresponding to different degradation paths;

[0182] Calculate the similarity between the feature vectors corresponding to each degradation path to obtain the similarity between different degradation paths of the aging test equipment drive control detection board.

[0183] Optionally, the at least two signal parameters include: mean, standard deviation, variance, skewness, peak-to-peak value.

[0184] Optionally, the fault prediction device of the aging test equipment drive control detection board further includes a processing module 410, configured to:

[0185] Obtain the signal values corresponding to the analog quantity signal of the aging test equipment drive control detection board in the target time series;

[0186] Determine a target clustering result to which a target degradation path composed of the signal values belongs according to the signal values;

[0187] Determine a target fault prediction model corresponding to the target clustering result, input the signal values into the target fault prediction model, and obtain a fault prediction result of the drive control detection board of the aging test equipment output by the target fault prediction model.

[0188] The above is a schematic solution of a fault prediction device for a drive control detection board of an aging test equipment. It should be noted that the technical solution of the fault prediction device for the drive control detection board of the aging test equipment belongs to the same concept as the above-mentioned fault prediction method for the drive control detection board of the aging test equipment. For the details not described in detail in the technical solution of the fault prediction device for the drive control detection board of the aging test equipment, reference can be made to the description of the technical solution of the above-mentioned fault prediction method for the drive control detection board of the aging test equipment.

[0189] Figure 5 FIG. shows a structural block diagram of a computing device 500 according to an embodiment of the present specification. The components of the computing device 500 include, but are not limited to, a memory 510 and a processor 520. The processor 520 is connected to the memory 510 through a bus 530, and a database 550 is used to store data.

[0190] The computing device 500 further includes an access device 540, and the access device 540 enables the computing device 500 to communicate via one or more networks 560. Examples of these networks include a public switched telephone network (PSTN), a local area network (LAN), a wide area network (WAN), a personal area network (PAN), or a combination of communication networks such as the Internet. The access device 540 may include one or more of any type of wired or wireless network interface (for example, a network interface card (NIC)), such as an IEEE802.11 wireless local area network (WLAN) wireless interface, a Worldwide Interoperability for Microwave Access (Wi-MAX) interface, an Ethernet interface, a Universal Serial Bus (USB) interface, a cellular network interface, a Bluetooth interface, a Near Field Communication (NFC) interface, and so on.

[0191] In an embodiment of the present specification, the above components of the computing device 500 and Figure 5 other components not shown therein may also be connected to each other, for example, through a bus. It should be understood that Figure 5 the shown structural block diagram of the computing device is only for example purposes and is not a limitation on the scope of the present specification. Those skilled in the art can add or replace other components as needed.

[0192] The computing device 500 can be any type of stationary or mobile computing device, including mobile computers or mobile computing devices (e.g., tablet computers, personal digital assistants, laptop computers, notebook computers, netbooks, etc.), mobile phones (e.g., smartphones), wearable computing devices (e.g., smartwatches, smart glasses, etc.) or other types of mobile devices, or stationary computing devices such as desktop computers or PCs. The computing device 500 can also be a mobile or stationary server.

[0193] Wherein, the processor 520 is used to execute the following computer-executable instructions, and when the computer-executable instructions are executed by the processor, the steps of the above-mentioned fault prediction method for the aging test equipment drive control detection board are implemented.

[0194] The above is a schematic solution of a computing device according to this embodiment. It should be noted that the technical solution of this computing device and the technical solution of the above-mentioned fault prediction method for the aging test equipment drive control detection board belong to the same concept. For the details not described in detail in the technical solution of the computing device, reference can be made to the description of the technical solution of the above-mentioned fault prediction method for the aging test equipment drive control detection board.

[0195] This specification also provides a computer-readable storage medium in one embodiment, which stores computer-executable instructions, and when the computer-executable instructions are executed by the processor, the steps of the above-mentioned fault prediction method for the aging test equipment drive control detection board are implemented.

[0196] The above is a schematic solution of a computer-readable storage medium according to this embodiment. It should be noted that the technical solution of this storage medium and the technical solution of the above-mentioned fault prediction method for the aging test equipment drive control detection board belong to the same concept. For the details not described in detail in the technical solution of the storage medium, reference can be made to the description of the technical solution of the above-mentioned fault prediction method for the aging test equipment drive control detection board.

[0197] This specification also provides a computer program in one embodiment, wherein when the computer program is executed on a computer, the computer is made to execute the steps of the above-mentioned fault prediction method for the aging test equipment drive control detection board.

[0198] The above is a schematic solution of a computer program according to this embodiment. It should be noted that the technical solution of this computer program and the technical solution of the above-mentioned fault prediction method for the aging test equipment drive control detection board belong to the same concept. For the details not described in detail in the technical solution of the computer program, reference can be made to the description of the technical solution of the above-mentioned fault prediction method for the aging test equipment drive control detection board.

[0199] The above describes specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the acts or steps recited in the claims may be performed in a different order than in the embodiments and still achieve the desired result. Additionally, the processes depicted in the figures do not necessarily require the particular order or sequential order shown to achieve the desired result. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0200] The computer instructions include computer program code, which may be in source code form, object code form, executable files, or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, mobile hard disks, magnetic disks, optical disks, computer memories, read-only memories (ROMs), random access memories (RAMs), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium may be appropriately increased or decreased according to the requirements of legislation and patent practice within the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0201] It should be noted that for the foregoing method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the embodiments of this specification are not limited by the described order of actions, because according to the embodiments of this specification, certain steps may be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the embodiments of this specification.

[0202] In the above embodiments, the descriptions of the various embodiments have their own focuses. For the parts not detailed in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

[0203] The preferred embodiments of this specification disclosed above are only used to help explain this specification. The alternative embodiments do not elaborate on all the details and do not limit the invention to only the specific embodiments described. Obviously, many modifications and variations can be made according to the content of the embodiments of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the embodiments of this specification, so that those skilled in the art can well understand and utilize this specification. This specification is only limited by the claims and their full scope and equivalents.

Claims

1. A fault prediction method for a drive control detection board of an aging test equipment, comprising: Obtaining signal values corresponding to a target time series of the drive control detection board of the aging test equipment and a corresponding target fault prediction model, inputting the signal values into the target fault prediction model, and obtaining a fault prediction result; In the case where the fault prediction result is a degradation state, respectively calculating the Mahalanobis distance, cosine similarity, and Manhattan distance between the signal values and pre-acquired healthy baseline data, and performing multi-distance metric fusion with automatic weight optimization on the Mahalanobis distance, the cosine similarity, and the Manhattan distance to obtain a final fault detection result of the drive control detection board of the aging test equipment, which includes: ； Among them, represents the final fault detection result, represents the Mahalanobis distance; represents the cosine similarity; represents the Manhattan distance; , and respectively represent weights, and + + = 1; Alternatively, in the case where the fault prediction result is a degradation state, based on the probability space points mapped from a multi-dimensional feature sequence generated by feature extraction of historical signal values and the healthy baseline in the residual convolutional neural network model, determining a state degradation curve for characterizing the health state of the drive control detection board of the aging test equipment; Wherein, the residual convolutional neural network model is trained in the following manner: obtaining a training time series degradation feature sample of the drive control detection board of the aging test equipment; inputting the training time series degradation feature sample into a pre-established residual convolutional neural network model to obtain a first probability space point after mapping; discriminating the degradation starting point of the drive control detection board of the aging test equipment according to the two-dimensional first probability space point, and generating a healthy baseline of the drive control detection board of the aging test equipment.

2. The fault prediction method for the drive control detection board of the aging test equipment according to claim 1, further comprising: Obtain the Multiple historical signal values of the aging test equipment drive control detection board and calculate the similarity between different degradation paths composed of different historical signal values; Clustering different degradation paths based on the similarity between different degradation paths composed of different historical signal values to generate at least two clustering results, and respectively using the historical signal values corresponding to the degradation paths included in different clustering results as training data to input into corresponding fault prediction models to be trained, and obtaining trained fault prediction models; The fault prediction model to be trained is a multi-dimensional recurrent neural network model, including an input layer, an intermediate layer, and an output layer. The intermediate layer includes a sequentially connected nearest neighbor component analysis model structure and a multi-dimensional recurrent neural network model structure. The multi-dimensional recurrent neural network model structure includes a fully connected layer, a bidirectional long short-term memory network layer, a bidirectional gated recurrent unit layer, and a splicing layer.

3. The fault prediction method for the drive control detection board of the aging test equipment according to claim 2, wherein calculating the similarity between different degradation paths of the drive control detection board of the aging test equipment includes: Processing the historical signal values corresponding to different degradation paths of the drive control detection board of the aging test equipment through the nearest neighbor component analysis model structure to obtain the similarity between different degradation paths, wherein the nearest neighbor component analysis model structure processes the historical signal values corresponding to different degradation paths of the drive control detection board of the aging test equipment through the nearest neighbor component analysis algorithm.

4. The fault prediction method for the aging test equipment drive control detection board according to claim 2 or 3, wherein the middle layer includes a sequentially connected nearest neighbor component analysis model structure, a Gaussian mixture model structure, and a multi-dimensional recurrent neural network model structure; Correspondingly, the clustering of different degradation paths based on the similarity includes: Clustering different degradation paths based on the similarity through the Gaussian mixture model structure, wherein the Gaussian mixture model structure clusters different degradation paths through the Gaussian mixture algorithm and based on the similarity.

5. The fault prediction method for the aging test equipment drive control detection board according to claim 1, wherein the calculation of the similarity between different degradation paths of the aging test equipment drive control detection board includes: Determining at least two signal parameters of the aging test equipment drive control detection board based on the historical signal values; Performing feature extraction on the at least two signal parameters to obtain feature vectors corresponding to different degradation paths; Calculating the similarity between the feature vectors corresponding to each degradation path to obtain the similarity between different degradation paths of the aging test equipment drive control detection board.

6. The fault prediction method of the aging test equipment drive control detection board according to claim 5, wherein the at least two signal parameters include: Mean, standard deviation, variance, skewness, peak-to-peak value.

7. The fault prediction method for the aging test equipment drive control detection board according to claim 1, wherein the obtaining of the signal values corresponding to the target time series of the aging test equipment drive control detection board and its corresponding target fault prediction model, and inputting the signal values into the target fault prediction model to obtain the fault prediction result includes: Obtaining the signal values corresponding to the analog signal of the aging test equipment drive control detection board in the target time series; Determining the target clustering result to which the target degradation path composed of the signal values belongs according to the signal values; Determining the target fault prediction model corresponding to the target clustering result, inputting the signal values into the target fault prediction model, and obtaining the fault prediction result of the aging test equipment drive control detection board output by the target fault prediction model.

8. A fault prediction device for an aging test equipment drive control detection board, comprising: An acquisition module, configured to acquire the signal values corresponding to the target time series of the aging test equipment drive control detection board and its corresponding target fault prediction model, input the signal values into the target fault prediction model to obtain a fault prediction result, and in the case that the fault prediction result is a degradation state, respectively calculate the Mahalanobis distance, cosine similarity, and Manhattan distance between the signal values and the pre-acquired healthy baseline data, and obtain the final fault detection result of the aging test equipment drive control detection board through multi-distance metric fusion with automatic weight optimization of the Mahalanobis distance, the cosine similarity, and the Manhattan distance, which includes: ； Among them, represents the final fault detection result, represents the Mahalanobis distance; represents the cosine similarity; represents the Manhattan distance; , and respectively represent weights, and + + = 1; The acquisition module is further configured to determine a state degradation curve for characterizing the health state of the drive control detection board of the aging test equipment based on the probability space point after mapping of the multi-dimensional feature sequence generated by feature extraction of historical signal values and the health baseline in the residual convolutional neural network model when the fault prediction result is a degradation state; Wherein, the residual convolutional neural network model is trained in the following manner: acquiring a training time-series degradation feature sample of the drive control detection board of the aging test equipment; inputting the training time-series degradation feature sample into a pre-established residual convolutional neural network model to obtain a first probability space point after mapping; discriminating the degradation starting point of the drive control detection board of the aging test equipment according to the two-dimensional first probability space point, and generating a health baseline of the drive control detection board of the aging test equipment.

9. A computing device, comprising: A memory and a processor; The memory is used for storing computer-executable instructions, and the processor is used for executing the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the fault prediction method for the drive control detection board of the aging test equipment according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing computer-executable instructions, and when the computer-executable instructions are executed by a processor, the steps of the fault prediction method for the drive control detection board of the aging test equipment according to any one of claims 1 to 7 are implemented.

Images