Method and system for predicting service life of electric energy meter production line bearing, equipment and medium

Abstract

The invention discloses a life prediction method and system for a bearing of an electric energy meter production line, equipment and a medium. According to the life prediction method, time domain features and frequency domain features are extracted through an improved TPA algorithm, the weight of important features can be improved, and the weight of unimportant features can be reduced; the weight proportion can be re-allocated at each time node according to the importance of the features, dynamic adjustment of time-frequency domain feature weights is realized, features having more influence on residual service life prediction can be reserved, redundant features are eliminated, the fused features are more refined and representative, the subsequent calculation amount is reduced, and the prediction efficiency is improved. The interference of unimportant features on the final decision of the model is reduced, and the attention is completely focused on the prediction target, so that the prediction accuracy and robustness of the prediction model are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of service life prediction, and in particular to a method and system for predicting the service life of bearings in an electric energy meter production line, an electronic device, and a computer-readable storage medium. Background Art

[0002] In the large-scale production of smart energy meters, bearings (typically rolling bearings) in meter production lines serve as core power transmission components of the production line's drive system. Their health directly determines the production line's operational efficiency. Given the current annual demand for over 10 million smart energy meters for the power grid, bearings in meter production lines are typically required to maintain continuous high-speed operation (typically reaching 2000-3000 rpm) 24 hours a day. A single bearing failure can cause the entire line to shut down for 4-8 hours, resulting in economic losses exceeding one million yuan per event. Furthermore, micron-level vibration deviations (<50μm) caused by premature bearing wear can be transmitted to precision assembly processes, causing critical parameters (such as metering accuracy and communication module positioning) to exceed specifications and leading to mass quality failures. Therefore, improving the accuracy of predicting the remaining useful life (RUL) of bearings in meter production lines can not only reduce the risk of line downtime but also prevent mass quality failures such as metering accuracy deviations caused by premature bearing wear.

[0003] The RUL prediction problem for bearings in electricity meter production lines is a typical time series problem. Many current studies use attention mechanisms to capture important features of time series data to achieve life prediction. For example, the paper "An Improved CNN-LSTM Model for Rolling Bearing Remaining Life Prediction" uses a Temporal Pattern Attention Mechanism (TPA) to focus on key information and then performs data prediction within an LSTM model. Compared to other models, this model demonstrates higher accuracy and better generalization. Classic TPA emphasizes information at key time nodes, making it more suitable for time series model predictions. However, because classic TPA assigns weights to all time node information without considering their contribution or relevance, the TPA model consumes significant computational resources for irrelevant or unrepresentative data nodes, resulting in a waste of resources. More importantly, it fails to eliminate the interference of unimportant and noisy features on the model's final decision, affecting the model's prediction accuracy. Summary of the Invention

[0004] The present invention provides a life prediction method and system for bearings in an electric energy meter production line, electronic equipment, and a computer-readable storage medium, which can improve the prediction accuracy and robustness of a prediction model.

[0005] According to one aspect of the present invention, a method for predicting the life of bearings in an electric energy meter production line is provided, comprising the following steps: Collect vibration signal data from bearings in the electric energy meter production line and extract time domain and frequency domain signals from the vibration signal data; The improved TPA algorithm is used to extract time domain feature signals and frequency domain feature signals from time domain signals and frequency domain signals respectively, and then fuse them to obtain time-frequency domain fusion signals. The improved TPA algorithm increases the weight of important features and reduces the weight of unimportant features. The time-frequency domain fusion signal is input into the prediction model to predict the remaining service life of the bearings in the electricity meter production line.

[0006] Furthermore, the process of extracting the time domain feature signal from the time domain signal using the improved TPA algorithm includes the following: The convolutional neural network is used to perform convolution operations on past information to obtain a feature matrix, and the scoring function is used to calculate the score of each row in the feature matrix to obtain a scoring matrix; Compare all scores in the scoring matrix with the preset first and second thresholds. If a score is greater than the first threshold, increase the score. If a score is less than the second threshold, decrease the score. If a score is greater than or equal to the second threshold and less than or equal to the first threshold, keep the score unchanged, and generate a new scoring matrix. The new score matrix and the feature matrix are weighted and summed to obtain the feature vector, which is then concatenated with the current information to obtain the final feature vector.

[0007] Furthermore, the score is adjusted based on the following formula: in, Indicates the first x Ratings, express The adjusted rating, β Indicates the adjustment parameters, X represents the first threshold, Y Indicates the second threshold.

[0008] Furthermore, all the scores in the score matrix are sorted from small to large, the first quartile is taken as the second threshold, and the third quartile is taken as the first threshold.

[0009] Furthermore, after obtaining the time-frequency domain fusion signal, the following contents are also included: The temporal convolutional network is used to perform multiple dilation operations on the time-frequency domain fusion signal to extract deep features.

[0010] Furthermore, the process of performing multiple dilation operations on the time-frequency domain fusion signal using the time convolutional network includes the following: The temporal convolutional network is used to perform the first dilation operation on the time-frequency domain fusion signal in a signal overlapping manner, and then deep features are extracted through two dilation causal convolution operations.

[0011] Furthermore, the temporal convolutional network includes three convolution kernels of different sizes, which are distributed in three parallel temporal convolution branches to extract short, medium and long scale features respectively.

[0012] In addition, the present invention also provides a life prediction system for bearings in an electric energy meter production line, comprising: The time-frequency domain signal processing module is used to collect the vibration signal data of the bearings in the electric energy meter production line and extract the time domain signal and frequency domain signal of the vibration signal data; The time-frequency domain signal fusion module is used to extract the time domain feature signal and the frequency domain feature signal from the time domain signal and the frequency domain signal respectively using the improved TPA algorithm, and fuse them to obtain the time-frequency domain fusion signal. The improved TPA algorithm increases the weight of important features and reduces the weight of unimportant features. The remaining service life prediction module is used to input the time-frequency domain fusion signal into the prediction model to predict the remaining service life of the bearings in the electricity meter production line.

[0013] In addition, the present invention also provides an electronic device, including a processor and a memory, wherein the memory stores a computer program, and the processor is configured to execute the steps of the above method by calling the computer program stored in the memory.

[0014] In addition, the present invention also provides a computer-readable storage medium for storing a computer program for predicting the bearing life of an electric energy meter production line, wherein the computer program executes the steps of the above-mentioned method when running on a computer.

[0015] The present invention has the following beneficial effects: The life prediction method for bearings in an electric energy meter production line of the present invention extracts time domain features and frequency domain features through an improved TPA algorithm, which can increase the weight of important features and reduce the weight of unimportant features. The weight ratio can be redistributed at each time node according to the importance of the features, thereby realizing dynamic adjustment of the weights of time and frequency domain features, thereby retaining features that are more influential on the remaining service life prediction and excluding redundant features, making the fused features more refined and representative, reducing the subsequent calculation amount, and reducing the interference of unimportant features on the final decision of the model, focusing all attention on the prediction target, thereby improving the prediction accuracy and robustness of the prediction model.

[0016] In addition, the life prediction system for bearings in the electric energy meter production line of the present invention also has the above advantages.

[0017] In addition to the above-described objects, features and advantages, the present invention has other objects, features and advantages. The present invention will be further described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are intended to explain the present invention and do not constitute an undue limitation of the present invention. In the accompanying drawings: Figure 1 This is a flow chart of a method for predicting the life of bearings in an electric energy meter production line according to a preferred embodiment of the present application; Figure 2 yes Figure 1 Schematic diagram of the sub-process of step S2; Figure 3 This is a schematic diagram of the principle of achieving accurate feature memory using the improved TPA algorithm in the preferred embodiment of the present application; Figure 4 This is another flow chart of the method for predicting the life of bearings in an electric energy meter production line according to a preferred embodiment of the present application; Figure 5 1 is a schematic diagram of the comparison results of the prediction method of the present application and the existing prediction method in the preferred embodiment of the present application; Figure 6 This is a schematic diagram of the prediction results before and after adding 5dB noise when performing noise robustness testing in a preferred embodiment of the present application; Figure 7 This is a schematic diagram of the prediction results before and after adding 10dB noise when performing noise robustness testing in a preferred embodiment of the present application; Figure 8 It is a schematic diagram of the module structure of a bearing life prediction system for an electric energy meter production line according to another embodiment of the present application. DETAILED DESCRIPTION

[0019] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0020] Reference Figure 1 The preferred embodiment of the present application provides a method for predicting the life of bearings in an electric energy meter production line, comprising the following contents: Step S1: collecting vibration signal data of bearings in an electric energy meter production line, and extracting time domain signals and frequency domain signals of the vibration signal data; Step S2: using an improved TPA algorithm to extract time domain feature signals and frequency domain feature signals from the time domain signal and the frequency domain signal respectively, and fusing them to obtain a time-frequency domain fusion signal, wherein the improved TPA algorithm increases the weight of important features and reduces the weight of unimportant features; Step S3: Input the time-frequency domain fusion signal into the prediction model to predict the remaining service life of the bearings of the electric energy meter production line.

[0021] It can be understood that the life prediction method for bearings in the electricity meter production line of this embodiment, through the extraction of time domain features and frequency domain features by the improved TPA algorithm, can increase the weight of important features and reduce the weight of unimportant features, and can redistribute the weight ratio according to the importance of the features at each time node, thereby realizing dynamic adjustment of the weights of time and frequency domain features, thereby retaining features that are more influential on the remaining service life prediction and excluding redundant features, making the fused features more refined and representative, reducing the subsequent calculation amount, and reducing the interference of unimportant features on the final decision of the model, focusing all attention on the prediction target, thereby improving the prediction accuracy and robustness of the prediction model.

[0022] In step S1, vibration sensor data is first collected from the bearings of the electricity meter production line, including horizontal and vertical vibration signal data. The time domain signal is then extracted from the vibration signal data using the EEMD algorithm, and the frequency domain signal is extracted using the FFT algorithm. Compared to the time domain signal, the frequency domain signal contains more bearing degradation information and can more accurately reflect the periodic characteristics of the bearing degradation process. Both the EEMD algorithm and the FFT algorithm are state-of-the-art, and their specific principles are not further described here. Optionally, after collecting the vibration signal data, it can also be subjected to noise reduction processing to filter out some noise interference.

[0023] In addition, due to the complex components and large amount of information of the bearing vibration signal, the traditional TPA algorithm does not consider the importance of the current information when processing data, and assigns weights to all the current information indiscriminately, resulting in a significant waste of computing resources due to the computing power consumed for unimportant features. Therefore, in step S2, the present application uses the improved TPA algorithm to extract time domain feature signals and frequency domain feature signals from the time domain signal and frequency domain signal respectively, increases the weight of important features and reduces the weight of unimportant features, and fuses the extracted time domain feature signals and frequency domain feature signals to obtain a time-frequency domain fusion signal. Among them, as Figure 2 As shown, the process of extracting the time domain feature signal from the time domain signal using the improved TPA algorithm includes the following: Step S21: Perform convolution operation on the past information through the convolutional neural network to obtain a feature matrix, and use the scoring function to calculate the score of each row in the feature matrix to obtain a scoring matrix; Step S22: Compare all scores in the scoring matrix with the preset first and second thresholds. If a score is greater than the first threshold, increase the score. If a score is less than the second threshold, decrease the score. If a score is greater than or equal to the second threshold and less than or equal to the first threshold, keep the score unchanged, and generate a new scoring matrix. Step S23: Perform weighted summation on the new scoring matrix and the feature matrix to obtain a feature vector, and concatenate the feature vector with the current information to obtain a final feature vector.

[0024] Specifically, such as Figure 3 As shown, we first perform convolution operation on the past information nodes of the convolutional neural network CNN to obtain the feature matrix H c CNN extracts features from input data through convolution and pooling operations. The convolution operation uses convolution kernels to extract local features of the data. The pooling operation performs downsampling based on the features extracted by the convolution layer, thereby reducing the feature dimension, reducing the number of parameters, and preventing overfitting. Then, the scoring function is used to calculate the score of each row in the feature matrix to obtain a scoring matrix. The scoring function can evaluate the degree of influence of past information on the present. The greater the influence, the higher the score. The specific scoring function and its principle belong to the existing technology and will not be repeated here. Then, all scores in the scoring matrix are compared with the preset first threshold and second threshold. If a score is greater than the first threshold, the score is increased, that is, its weight value is increased, so that the model can pay more attention to important features, which helps the model capture key time patterns. If a score is less than the second threshold, the score is reduced, that is, its weight value is reduced, so that the model can reduce the impact on unimportant or noise features, thereby reducing their interference with the final decision. If a score is greater than or equal to the second threshold and less than or equal to the first threshold, the score is kept unchanged to generate a new scoring matrix, that is, the precise memory matrix. Finally, the new scoring matrix With the characteristic matrix H c Perform weighted sum calculation to obtain the context vector containing the importance of past information to the current node information P t , and then the context vector P t With current information h t Splice to get the final feature vector ,The final feature vector not only contains the current node information, but also contains the importance of past information to the current node information.

[0025] Optional, adjust the score based on the following formula: in, Indicates the first x Ratings, express The adjusted rating, β Indicates the adjustment parameter, usually 0.1, X represents the first threshold, Y Indicates the second threshold.

[0026] Optionally, all scores in the scoring matrix are sorted from smallest to largest, and the first quartile is used as the second threshold, and the third quartile is used as the first threshold. It can be understood that by selecting quartiles as thresholds, the thresholds have stable statistical properties, are insensitive to outliers, can reflect the overall distribution of the data, and avoid abnormal weights interfering with the model. In addition, the quartile division can also achieve automatic threshold adjustment, retain key degradation patterns, and ensure the rationality and interpretability of the threshold setting.

[0027] It can be understood that the present invention has made improvements based on the traditional TPA algorithm. By introducing a threshold screening mechanism after the scoring function, the output of the scoring function is adaptively adjusted, avoiding the situation where the output results of the scoring function are distributed too concentrated or dispersed, and focusing the algorithm's attention on the most relevant and important parts. While deepening the memory of important information, unimportant information is completely forgotten, achieving precise memory, which not only improves the algorithm accuracy and saves computing resources, but also helps capture data degradation characteristics and enhance the robustness of the algorithm.

[0028] In addition, the process of extracting the frequency domain feature signal from the frequency domain signal using the improved TPA algorithm is consistent with the above and will not be repeated here. In addition, after extracting the time domain feature signal and the frequency domain feature signal using the improved TPA algorithm, the time domain feature signal and the frequency domain feature signal can be fused by weighted fusion or splicing to obtain a time-frequency domain fused signal. The specific fusion process belongs to the existing technology and will not be repeated here.

[0029] In step S3, the prediction model uses a BiLSTM model. The BiLSTM model consists of a bidirectional LSTM layer and a fully connected layer. The forward LSTM inputs data into the model in a forward direction, while the reverse LSTM inputs data in a reverse direction. The fully connected layer is responsible for concatenating the hidden states of the forward and reverse LSTMs at each moment. The training process and prediction principles of the BiLSTM model are state-of-the-art and will not be elaborated here.

[0030] Optional, such as Figure 4 As shown, the life prediction method for bearings in the electric energy meter production line further includes the following contents after obtaining the time-frequency domain fusion signal: Step S20: Use a temporal convolutional network to perform multiple dilation operations on the time-frequency domain fusion signal to extract deep features.

[0031] Specifically, while retaining the powerful feature extraction capabilities of CNNs, temporal convolutional networks (TCNs) are more suitable for processing time series data. They can extract deep features, which helps further improve prediction accuracy. However, traditional TCNs only perform a single dilation convolution on time series data, which can easily cause the loss of important information. Therefore, this application uses a temporal convolutional network to perform multiple dilation operations on the time-frequency domain fusion signal, which can more comprehensively learn the important features of the vibration signal without omissions and ensure that important information is not lost.

[0032] The process of performing multiple dilation operations on the time-frequency domain fusion signal using a temporal convolutional network includes the following: The temporal convolutional network is used to perform the first dilation operation on the time-frequency domain fusion signal in a signal overlapping manner, and then deep features are extracted through two dilation causal convolution operations.

[0033] Specifically, assuming that the size of the time-frequency domain fusion signal is j×3, TCN first divides the time-frequency domain fusion signal into three segments, each of which is j / 3×3 in size, and then performs overlapping processing so that there is signal stacking between each segment. Finally, the three stacked signals are spliced together to form a stacked signal of size 3j / 2×3, where j represents the length of the time-frequency domain fusion signal sequence. The stacked signal is then subjected to two dilated causal convolution operations. Dilated causal convolution is a basic operation in traditional TCNs. Dilated convolution amplifies the receptive field by filling holes with "0"s without increasing the model's computational cost. Causal convolution appropriately pads the input data to ensure that the convolution output depends only on past and present inputs and not on future information. The first dilated convolution has an input of 3j / 2×3 and an output of 3j×3. The first causal convolution has an input of 3j×3 and an output of (3j+2)×3. The second dilated convolution has an input of (3j+2)×3 and an output of (6j+4)×3. The second causal convolution has an input of (6j+4)×3 and an output of (6j+6)×3. Furthermore, the TCN's input and output matrices both have 3 columns, representing time information, horizontal vibration signals, and vertical vibration signals, respectively.

[0034] As can be understood, the present invention first segments and overlaps the signal, ensuring that each signal segment overlaps with other segments. This overlapping portion provides more contextual information, ensuring that important signal features are repeatedly captured across multiple segments. This increases the likelihood that important information will be retained, enabling the model to better predict signal trends and further improving its predictive accuracy. Furthermore, through two dilated causal convolution operations, more in-depth features can be extracted, further improving predictive accuracy.

[0035] In addition, the convolution kernel size of the traditional TCN is fixed, so the size of its receptive field is also fixed, which results in the traditional TCN being able to extract only single-scale features, limiting the ability of TCN to capture features of different scales. Therefore, optionally, the temporal convolution network of the present application includes three convolution kernels of different sizes, which are distributed in three parallel temporal convolution branches to extract short, medium and long scale features respectively. Specifically, the TCN of the present application includes three parallel TCN branches, which realize the extraction of long, medium and short scale features of time series by setting convolution kernels of three different sizes: large, medium and small, so that the model can capture data dependencies more comprehensively.

[0036] Preferably, the three convolution kernel sizes are set to 2, 3, and 5 respectively. The calculation formula for the data length that the convolution kernel can cover is: l = k +( k-1)×( d -1), l Represents the receptive field length of the convolution kernel, that is, the length of the input data that the convolution kernel can cover. k is the original size of the convolution kernel, that is, the width or height of the convolution kernel without dilation, d represents the dilation rate, which defines the spacing between elements in the convolution kernel, d The size of is generally set to 4. Therefore, the receptive field can be calculated l They are 5, 9 and 17 respectively. This parameter setting fully considers the multi-scale characteristics of the bearing vibration signal. Smaller kernel sizes (k=2, k=3) can effectively capture the high-frequency transient features in the signal, while the larger kernel size (k=5) is more suitable for extracting low-frequency trend information. By adopting a dilation rate of 4, while ensuring that the receptive field (5, 9, 17) is sufficient to cover the complete signal features, the computational redundancy problem brought by traditional large convolution kernels is avoided. This parameter setting not only achieves a good balance between computational efficiency and feature extraction capability, but also significantly improves the prediction performance of the model.

[0037] In addition, in order to verify the effectiveness of the prediction method of this application, this application also conducted a comparative verification with some existing prediction methods, using MAE, MSE and RMSE as evaluation indicators. The comparison results are as follows: Figure 5 As shown. Figure 5 It can be seen that the three evaluation indicators of the prediction method of the present application are all better than the existing eight algorithms. This is because the present application adopts a more advanced architectural design. It not only distributes proportional weights to the time-frequency signals through the improved TPA algorithm, retains the features that are more influential on the remaining useful life prediction, and eliminates redundant features, but also combines the TCN with multiple expansion and multi-core mechanisms to deeply mine and extract features, which can effectively capture the key information in the data and extract features. Therefore, the error value of the prediction method of the present application is smaller.

[0038] In addition, in order to verify the robustness of the prediction method of this application, this application also conducted a noise robustness test, and the test results are as follows: Figure 6 and Figure 7 As shown, Figure 6 Comparison of prediction results before and after adding 5dB noise to the model. Figure 7 Comparison of prediction results before and after adding 10dB noise to the model. It can be seen that the model's prediction results are very close to the real data without adding noise, indicating that the model has high prediction accuracy in a noise-free environment. Furthermore, the model's prediction results remain highly consistent with the real data when adding 5dB and 10dB noise, indicating that the model has strong noise resistance.

[0039] In addition, if Figure 8As shown, another embodiment of the present invention further provides a life prediction system for bearings in an electric energy meter production line, preferably using the above-mentioned life prediction method for bearings in an electric energy meter production line, comprising: The time-frequency domain signal processing module is used to collect the vibration signal data of the bearings in the electric energy meter production line and extract the time domain signal and frequency domain signal of the vibration signal data; The time-frequency domain signal fusion module is used to extract the time domain feature signal and the frequency domain feature signal from the time domain signal and the frequency domain signal respectively using the improved TPA algorithm, and fuse them to obtain the time-frequency domain fusion signal. The improved TPA algorithm increases the weight of important features and reduces the weight of unimportant features. The remaining service life prediction module is used to input the time-frequency domain fusion signal into the prediction model to predict the remaining service life of the bearings in the electricity meter production line.

[0040] It can be understood that the life prediction system of the bearings of the electricity meter production line in this embodiment extracts time domain features and frequency domain features through the improved TPA algorithm, which can increase the weight of important features and reduce the weight of unimportant features. The weight ratio can be redistributed at each time node according to the importance of the features, thereby realizing dynamic adjustment of the weights of time and frequency domain features, thereby retaining features that are more influential on the remaining service life prediction and excluding redundant features, making the fused features more refined and representative, reducing the subsequent calculation amount, and reducing the interference of unimportant features on the final decision of the model, focusing all attention on the prediction target, thereby improving the prediction accuracy and robustness of the prediction model.

[0041] In addition, the life prediction system for bearings in the electric energy meter production line also includes: The multiple dilation operation module is used to perform multiple dilation operations on the time-frequency domain fusion signal using a temporal convolutional network to extract deep features.

[0042] It can be understood that the various modules of the system embodiment correspond to the various steps of the above method embodiment, so the working principles of the various modules will not be repeated here, and the corresponding references can be made to the various steps of the above method embodiment.

[0043] In addition, another embodiment of the present invention further provides an electronic device, including a processor and a memory, wherein the memory stores a computer program, and the processor is configured to execute the steps of the above method by calling the computer program stored in the memory.

[0044] In addition, another embodiment of the present invention further provides a computer-readable storage medium for storing a computer program for predicting the bearing life of an electric energy meter production line, wherein the computer program executes the steps of the above-described method when running on a computer.

[0045] Common forms of computer-readable storage media include: floppy disks, flexible disks, hard disks, magnetic tape, any other magnetic media, CD-ROMs, any other optical media, punch cards, paper tape, any other physical medium with a pattern of holes, random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), flash-EPROM, any other memory chip or cartridge, or any other medium that can be read by a computer. Instructions can further be transmitted or received via a transmission medium. The term transmission medium may include any tangible or intangible medium that can be used to store, encode, or carry instructions for execution by a machine, including digital or analog carrier communication signals or other intangible media that facilitate communication of such instructions. Transmission media include coaxial cables, copper wire, and fiber optics, including the wires of a bus used to transmit a computer data signal.

[0046] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of a fully hardware embodiment, a fully software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present application may be implemented in various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0047] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0048] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0049] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0050] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.

[0051] Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include these modifications and variations.

[0052] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A method for predicting the life of bearings in an electric energy meter production line, characterized in that: Includes the following: Collect vibration signal data from bearings in the electric energy meter production line and extract time domain and frequency domain signals from the vibration signal data; The improved TPA algorithm is used to extract time domain feature signals and frequency domain feature signals from time domain signals and frequency domain signals respectively, and then fuse them to obtain time-frequency domain fusion signals. The improved TPA algorithm increases the weight of important features and reduces the weight of unimportant features. The time-frequency domain fusion signal is input into the prediction model to predict the remaining service life of the bearings in the electricity meter production line.

2. The life prediction method for bearings in an electric energy meter production line according to claim 1, characterized in that: The process of extracting the time domain feature signal from the time domain signal using the improved TPA algorithm includes the following: The convolutional neural network is used to perform convolution operations on past information to obtain a feature matrix, and the scoring function is used to calculate the score of each row in the feature matrix to obtain a scoring matrix; Compare all scores in the scoring matrix with the preset first and second thresholds. If a score is greater than the first threshold, increase the score. If a score is less than the second threshold, decrease the score. If a score is greater than or equal to the second threshold and less than or equal to the first threshold, keep the score unchanged, and generate a new scoring matrix. The new score matrix and the feature matrix are weighted and summed to obtain the feature vector, which is then concatenated with the current information to obtain the final feature vector.

3. The life prediction method for bearings in an electric energy meter production line according to claim 2, characterized in that: Adjust the score based on the following formula: in, Indicates the first x Ratings, express The adjusted rating, β Indicates the adjustment parameters, X represents the first threshold, Y Indicates the second threshold.

4. The life prediction method for bearings in an electric energy meter production line according to claim 3, characterized in that: Sort all the scores in the score matrix from small to large, take the first quartile as the second threshold, and take the third quartile as the first threshold.

5. The life prediction method for bearings in an electric energy meter production line according to claim 1, characterized in that: After obtaining the time-frequency domain fusion signal, the following contents are also included: The temporal convolutional network is used to perform multiple dilation operations on the time-frequency domain fusion signal to extract deep features.

6. The life prediction method for bearings in an electric energy meter production line according to claim 5, characterized in that: The process of performing multiple dilation operations on the time-frequency domain fusion signal using a temporal convolutional network includes the following: The temporal convolutional network is used to perform the first dilation operation on the time-frequency domain fusion signal in a signal overlapping manner, and then deep features are extracted through two dilation causal convolution operations.

7. The life prediction method for bearings in an electric energy meter production line according to claim 6, characterized in that: The temporal convolutional network includes three convolution kernels of different sizes, which are distributed in three parallel temporal convolution branches to extract short, medium and long scale features respectively.

8. A life prediction system for bearings in an electric energy meter production line, characterized in that: include: The time-frequency domain signal processing module is used to collect the vibration signal data of the bearings in the electric energy meter production line and extract the time domain signal and frequency domain signal of the vibration signal data; The time-frequency domain signal fusion module is used to extract the time domain feature signal and the frequency domain feature signal from the time domain signal and the frequency domain signal respectively using the improved TPA algorithm, and fuse them to obtain the time-frequency domain fusion signal. The improved TPA algorithm increases the weight of important features and reduces the weight of unimportant features. The remaining service life prediction module is used to input the time-frequency domain fusion signal into the prediction model to predict the remaining service life of the bearings in the electricity meter production line.

9. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores a computer program, and the processor is configured to execute the steps of the method according to any one of claims 1 to 7 by calling the computer program stored in the memory.

10. A computer-readable storage medium for storing a computer program for predicting the life of bearings in an electric energy meter production line, characterized in that: When the computer program is run on a computer, the computer program executes the steps of the method according to any one of claims 1 to 7.

Images