Intelligent regulation and control system and method for sealing element production line

Abstract

The invention belongs to the technical field of intelligent regulation and control, and particularly relates to an intelligent regulation and control system and method for a sealing element production line, prediction is carried out through a neural network module based on a codec, and richer time sequence information, context and positive and negative sequence features are captured by using the neural network module of the codec, so that the reliability of the sealing element production line is improved. In the continuous iteration and unit state calculation process, the BiLSTM carries out vector splicing on prediction results of two LSTM models in the structure through historical data of each time step and inflow and outflow information of a control forgetting gate, an input gate and an output gate to obtain a first fault prediction result sequence, prediction is carried out through a BP neural network module, and a fault prediction result is obtained. And the first fault prediction result sequence and the second fault prediction result sequence are trained for weighted summation, so that the capture capability of data information is greatly improved, and the prediction precision is improved.

Description

Technical Field

[0001] This application belongs to the field of intelligent control technology, specifically relating to an intelligent control system and method for a sealing component production line. Background Technology

[0002] With the continuous improvement of industrial automation, mechanical equipment is being used more and more widely in the production process. The reliability and stability of this equipment are crucial for ensuring production efficiency and safety. However, mechanical equipment is susceptible to malfunctions due to various factors during long-term operation, leading to equipment downtime and severely impacting production progress and economic benefits. Therefore, forecasting the equipment production line and adjusting controls based on the forecast results is of significant practical importance. Summary of the Invention

[0003] To address the technical problems mentioned above, this invention provides an intelligent control system and method for a sealing component production line.

[0004] In a first aspect, the present invention provides an intelligent control system for a sealing component production line, comprising: Terminal equipment layer: used to collect operating status data, operating image data, and noise data of the sealing production line; The edge computing layer is used to predict production line failures and formulate corresponding control strategies based on the production line operation status data, operation image data and noise data collected by the terminal device layer. The cloud-based monitoring platform is used to visualize the production line's operating status and issue control commands to the edge computing layer. The edge computing layer includes: a data receiving module, a data preprocessing module, and a fault prediction model; The data receiving module is used to receive production line operating status data, operating image data, and noise data collected by the terminal equipment layer; The data preprocessing module is used to process the collected production line operating status data, operating image data, and noise data to generate cross-modal time series feature sequences. The fault prediction model includes: an extended Kalman filter, a codec-based neural network module, a backpropagation neural network module, and a fully connected module; Extended Kalman filter is used to remove noise and outliers from cross-modal time series feature sequences; The encoder-decoder-based neural network module includes an encoder and a decoder, which are connected through a recursive structure. Both the encoder and decoder use a BiLSTM module as the core to capture the temporal information, context and forward and reverse sequence features in the cross-modal temporal feature sequence to obtain the first fault prediction result sequence. The BP neural network module is used to obtain the second fault prediction result sequence based on the cross-modal temporal feature sequence; The fully connected module is used to sum the first fault prediction result sequence and the second fault prediction result sequence according to preset weights to obtain the final prediction result sequence.

[0005] Furthermore, the edge computing layer also includes a first training module for training a codec-based neural network module; The training of the encoder-decoder-based neural network module includes: The cross-modal temporal feature sequence X is input into the encoder; The BiLSTM module in the encoder performs mapping analysis and feature extraction on the cross-modal temporal feature sequence X and obtains the hidden state after recursive update. And output sequence Y, and obtain the hidden state Provided to the BiLSTM module in the decoder; The decoder obtains the hidden state Then, hide the state. Transform into the decoder's initial state And based on the encoder's output sequence Y, the prediction result sequence for the next time step is obtained; The predicted sequence for the next time step is recursively fed back into the encoder-decoder-based neural network module. This process is repeated multiple times to obtain an output sequence of length T, which is the first fault prediction result sequence.

[0006] Furthermore, the edge computing layer also includes a second training module for training the BP network model, wherein training the BP network model includes: Calculate the output value and linear combination result of each hidden layer neuron in the hidden layer, as well as the output value and linear combination result of each hidden layer neuron in the output layer during forward propagation. Calculate the value of the cross-loss function; During backpropagation, the error terms of the output layer and the hidden layer are calculated separately; Update the weights and biases from the hidden layer to the output layer.

[0007] Secondly, the present invention provides an intelligent control method for a sealing component production line, comprising the following steps: Step 1: Collect production line data and preprocess the collected production line data to obtain cross-modal time series feature sequences; Step 2: Remove noise and outliers from the cross-modal time series feature sequence using an extended Kalman filter to obtain the cross-modal time series feature sequence X; Step 3: Input the cross-modal temporal feature sequence X into the encoder-decoder-based neural network module and the BP neural network module to obtain the first fault prediction result sequence and the second fault prediction result sequence; Step 4: Sum the first fault prediction result sequence and the second fault prediction result sequence according to preset weights to obtain the final prediction result sequence; Step 5: Visualize the production line's operating status through a cloud-based monitoring platform and issue control commands to the edge computing layer.

[0008] Furthermore, the process between steps two and three also includes: The preprocessed production line data is divided into a training set and a dataset. Training the encoder-decoder-based neural network module; Training the BP God Network model.

[0009] Furthermore, the training of the codec-based neural network module includes: Calculate the output value and linear combination result of each hidden layer neuron in the hidden layer, as well as the output value and linear combination result of each hidden layer neuron in the output layer during forward propagation. Calculate the value of the cross-loss function; During backpropagation, the error terms of the output layer and the hidden layer are calculated separately; Update the weights and biases from the hidden layer to the output layer.

[0010] Furthermore, the training of the BP God Network model includes: The cross-modal temporal feature sequence X is input into the encoder; The BiLSTM module in the encoder performs mapping analysis and feature extraction on the cross-modal temporal feature sequence X and obtains the hidden state after recursive update. And output sequence Y, and obtain the hidden state Provided to the BiLSTM module in the decoder; The decoder obtains the hidden state Then, hide the state. Transform into the decoder's initial state And based on the encoder's output sequence Y, the prediction result sequence for the next time step is obtained; The predicted sequence for the next time step is recursively fed back into the encoder-decoder-based neural network module. This process is repeated multiple times to obtain an output sequence of length T, which is the first fault prediction result sequence.

[0011] Thirdly, the present invention provides a computer-readable storage medium including a stored program that, when the program is running, controls the electrical equipment where the computer-readable storage medium is located to execute the intelligent control system for the sealing component production line described above.

[0012] The beneficial effects of this invention are as follows: This invention uses a codec-based neural network module for prediction. This module captures richer temporal information, context, and forward / reverse sequence features. Through continuous iteration and unit state calculation, BiLSTM uses historical data at each time step and the inflow / outflow information from the forget gate, input gate, and output gate to concatenate the prediction results of two LSTM models in the structure to obtain a first fault prediction result sequence. This first sequence is then predicted using a BP neural network module to obtain a second fault prediction result sequence. Finally, the first and second prediction result sequences are weighted and summed, significantly improving the ability to capture data information and thus enhancing prediction accuracy. Attached Figure Description

[0013] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0014] Figure 1 This is a system framework diagram of the intelligent control system for a sealing component production line according to the present invention; Figure 2 This is a schematic diagram of the fault prediction model. Detailed Implementation

[0015] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0016] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, each technical and scientific term used in these embodiments has the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0017] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0018] In this invention, terms such as "upper," "lower," "left," "right," "front," "back," "vertical," "horizontal," "side," and "bottom" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used only to facilitate the description of the structural relationships of the various components or elements of this invention and do not specifically refer to any component or element in this invention. They should not be construed as limiting the invention.

[0019] In this invention, terms such as "fixed connection," "connected," and "linked" should be interpreted broadly, indicating a fixed connection, an integral connection, or a detachable connection; a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can determine the specific meaning of these terms in this invention based on the specific circumstances, and they should not be construed as limitations on the invention.

[0020] Example 1: like Figure 1 As shown, this embodiment provides an intelligent control system for a sealing component production line, including a terminal equipment layer, an edge computing layer, and a cloud monitoring platform; The terminal equipment layer consists of environmental data acquisition equipment, image data acquisition equipment, and noise acquisition equipment, and is responsible for collecting the operating status data, operating image data, and noise data of the sealing production line. The edge computing layer is used to predict production line faults and formulate corresponding control strategies based on the production line operation status data, operation image data and noise data collected by the terminal device layer, so as to prevent equipment failure, misoperation and other problems.

[0021] The cloud-based monitoring platform is used to visualize the production line's operating status and issue control commands to the edge computing layer to achieve centralized management across regions. The layers interact with each other through corresponding communication protocols to form an efficient and complete closed-loop control system.

[0022] The edge computing layer includes: a fault prediction model, a first training module, a second training module, and a control strategy formulation module; like Figure 2 As shown, the fault prediction model includes: a data preprocessing module, an extended Kalman filter, a codec-based neural network module and a BP neural network module, and a fully connected module; The data preprocessing module is used to process the collected production line operating status data, operating image data, and noise data to generate cross-modal time series feature sequences. Extended Kalman filter is used to remove noise and outliers from cross-modal time series feature sequences; The encoder-decoder-based neural network module includes an encoder and a decoder, which are connected through a recursive structure. Both the encoder and decoder use a BiLSTM module as the core to capture the temporal information, context and forward and reverse sequence features in the cross-modal temporal feature sequence to obtain the first fault prediction result sequence. The BP neural network module is used to obtain the second fault prediction result sequence based on the cross-modal temporal feature sequence; The fully connected module is used to sum the first fault prediction result sequence and the second fault prediction result sequence according to preset weights to obtain the final prediction result sequence. The first training module, used to train the encoder-decoder-based neural network module, includes the following steps: C1: Forward Propagation Calculate the output value of each hidden layer neuron and the result of linear combination: ; ; in, For the first The output values ​​of each hidden layer neuron; In the first layer (from input layer to hidden layer), the... The input neuron to the first The weights of each hidden layer neuron, with superscript... Used to represent the first layer; For the first The values ​​of each input neuron; For the first The bias values ​​of each hidden layer neuron. For the first The result of a linear combination of hidden layer neurons, i.e., weighted input plus bias.

[0023] The formula for calculating the output value of each hidden layer neuron in the output layer and the linear combination result is as follows: ; ; in, For the first The linear combination of the output layer neurons is the weighted input plus the bias. For the second layer (from hidden layer to output layer), the first... The hidden layer neurons to the first The weights of each output layer neuron, with superscript... Indicates the second layer; For the first The bias values ​​of each output layer neuron. For the first The output value of each hidden layer neuron, that is, the output of the activation function of the previous layer (hidden layer); For the first The output value of each output layer neuron This is the activation function.

[0024] C2: Calculate the value of the cross-loss function, the formula of which is shown below: ; in, The value of the cross-entropy loss function. This represents the total number of output categories. As an indicator variable, it represents the first... Real labels for each category For the first The probability of the category.

[0025] C3: Backpropagation: The formula for calculating the output layer error term is as follows: ; in, This is the error term for the k-th output neuron; The formula for calculating the hidden layer error term is as follows: ; in, For the first Error terms for each hidden layer neuron; Activation function exist The derivative at point .

[0026] C4: Updates the weights and biases from the hidden layer to the output layer, using the following formula: , ; The formula for updating the weights and biases from the input layer to the hidden layer is as follows: , ; in, For the first The hidden layer neurons to the first The weights of each output layer neuron. This is the learning rate.

[0027] The second training module is used to train the BP God Network model, and it includes the following steps: B1: Input the cross-modal temporal feature sequence X, after removing noise and outliers with an extended Kalman filter, into the encoder; B2: The BiLSTM module in the encoder performs mapping analysis and feature extraction on the cross-modal temporal feature sequence X and obtains the hidden state after recursive update. And output sequence Y, and obtain the hidden state Provided to the BiLSTM module in the decoder; B3: The decoder obtains the hidden state. Then, hide the state. Transform into the decoder's initial state And based on the encoder's output sequence Y, the prediction result sequence for the next time step is obtained; B4: The prediction result sequence for the next time step is recursively fed back into the encoder-decoder-based neural network module. This process is repeated multiple times to obtain an output sequence of length T, which is the first fault prediction result sequence.

[0028] Example 2: This embodiment provides an intelligent control method for a sealing component production line, including the following steps: S1: Collect production line data and preprocess the collected production line data to obtain cross-modal time series feature sequences.

[0029] The production line data includes: production line operation status time sequence data, infrared thermal images, acoustic data, and vibration data.

[0030] S1-1: To comprehensively perceive the operating status of the production line, a multimodal data acquisition system is constructed to collect production line data; In this embodiment, the production line data collected by the multimodal data acquisition system mainly includes four categories: ① production line operating status time-series data (e.g., current, voltage, and power of the production line); ② infrared thermal images (monitoring the temperature field of key parts); ③ acoustic fingerprint data; ④ vibration data, etc.

[0031] S1-2: Perform deep feature fusion on the collected production line data to obtain cross-modal time series feature sequences; Specifically, outlier removal and other processing are performed on the time-series data of the generating line's running status; wavelet packet decomposition (WPD) is used to extract features such as energy entropy and singular values ​​from the voiceprint signal; and convolutional neural network (CNN) is used to process infrared images to extract deep spatial features. S2: Divide the preprocessed production line data into a training set and a dataset; S3: Construct a fault prediction model and train the fault prediction model using a training set; The fault prediction model includes: an extended Kalman filter, a codec-based neural network module, a backpropagation neural network module, and a fully connected module.

[0032] A: Extended Kalman filter is used to remove noise and outliers in cross-modal time series feature sequences, extract the accurate operating status of the production line, and improve data quality and stability.

[0033] B: The encoder-decoder-based neural network module includes an encoder and a decoder, which are connected through a recursive structure. Both the encoder and decoder use BiLSTM modules as the core to capture richer temporal information, context, and forward and reverse sequence features. Through continuous iteration and unit state calculation, BiLSTM uses historical data at each time step and the inflow and outflow information of the control forget gate, input gate, and output gate to concatenate the prediction results of the two LSTM models in the structure into vectors to obtain the first fault prediction result sequence.

[0034] The training of the encoder-decoder-based neural network module includes the following steps: B1: Input the cross-modal temporal feature sequence X, after removing noise and outliers with an extended Kalman filter, into the encoder; B2: The BiLSTM module in the encoder performs mapping analysis and feature extraction on the cross-modal temporal feature sequence X and obtains the hidden state after recursive update. And output sequence Y, and obtain the hidden state Provided to the BiLSTM module in the decoder; B3: The decoder obtains the hidden state. Then, hide the state. Transform into the decoder's initial state And based on the encoder's output sequence Y, the prediction result sequence for the next time step is obtained; B4: The prediction result sequence for the next time step is recursively fed back into the encoder-decoder-based neural network module. This process is repeated multiple times to obtain an output sequence of length T, which is the first fault prediction result sequence.

[0035] C: BP neural network module, used to obtain a second fault prediction result sequence based on a data sequence X that has been filtered by an extended Kalman filter to remove noise and outliers.

[0036] Training the BP God Network model includes the following steps: Specifically, the following steps are included: C1: Forward Propagation Calculate the output value of each hidden layer neuron and the result of linear combination: ; ; in, For the first The output values ​​of each hidden layer neuron; In the first layer (from input layer to hidden layer), the... The input neuron to the first The weights of each hidden layer neuron, with superscript... Used to represent the first layer; For the first The values ​​of each input neuron; For the first The bias values ​​of each hidden layer neuron. For the first The result of a linear combination of hidden layer neurons, i.e., weighted input plus bias.

[0037] The formula for calculating the output value of each hidden layer neuron in the output layer and the linear combination result is as follows: ; ; in, For the first The linear combination of the output layer neurons is the weighted input plus the bias. For the second layer (from hidden layer to output layer), the first... The hidden layer neurons to the first The weights of each output layer neuron, with superscript... Indicates the second layer; For the first The bias values ​​of each output layer neuron. For the first The output value of each hidden layer neuron, that is, the output of the activation function of the previous layer (hidden layer); For the first The output value of each output layer neuron This is the activation function.

[0038] C2: Calculate the value of the cross-loss function, the formula of which is shown below: ; in, The value of the cross-entropy loss function. This represents the total number of output categories. As an indicator variable, it represents the first... Real labels for each category For the first The probability of the category.

[0039] C3: Backpropagation: The formula for calculating the output layer error term is as follows: ; in, This is the error term for the k-th output neuron; The formula for calculating the hidden layer error term is as follows: ; in, For the first Error terms for each hidden layer neuron; Activation function exist The derivative at point .

[0040] C4: Updates the weights and biases from the hidden layer to the output layer, using the following formula: , ; The formula for updating the weights and biases from the input layer to the hidden layer is as follows: , ; in, For the first The hidden layer neurons to the first The weights of each output layer neuron. This is the learning rate.

[0041] D: Fully connected module, used to sum the first fault prediction result sequence and the second fault prediction result sequence according to preset weights to obtain the final prediction result sequence.

[0042] S4: Input the test set into the trained fault prediction model to obtain the final prediction result sequence; Specifically, it includes the following steps: S4-1: Remove noise and outliers generated during data acquisition and uploading by using an extended Kalman filter to obtain the cross-modal time series feature sequence X; S4-2: Input the cross-modal temporal feature sequence X into the encoder-decoder-based neural network module and the BP neural network module to obtain the first fault prediction result sequence and the second fault prediction result sequence; S4-3: Sum the first fault prediction result sequence and the second fault prediction result sequence according to preset weights to obtain the final prediction result sequence.

[0043] Example 3: This embodiment provides a computer-readable storage medium including a stored program, wherein, when the program is executed, the computer-readable storage medium controls the execution of the operation method of the compressed air energy storage system in a medium-deep saline aquifer described in Embodiment 2.

[0044] The same or similar parts between the various embodiments in this specification can be referred to mutually. In particular, the terminal embodiments are basically similar to the method embodiments, so the description is relatively simple, and the relevant parts can be referred to the description in the method embodiments.

[0045] In the embodiments provided by this invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between systems or units may be electrical, mechanical, or other forms.

[0046] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0047] Additionally, it should be noted that the flowcharts in the accompanying drawings illustrate methods according to embodiments of this disclosure. In the descriptions corresponding to the flowcharts or block diagrams in the drawings, the operations or steps corresponding to different blocks may occur in a different order than disclosed in the description; sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps may actually be executed substantially in parallel, or sometimes in reverse order, depending on the function involved. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions.

[0048] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An intelligent control system for a sealing component production line, characterized in that, include: Terminal equipment layer: used to collect operating status data, operating image data, and noise data of the sealing production line; The edge computing layer is used to predict production line failures and formulate corresponding control strategies based on the production line operation status data, operation image data and noise data collected by the terminal device layer. The cloud-based monitoring platform is used to visualize the production line's operating status and issue control commands to the edge computing layer. The edge computing layer includes: a fault prediction model and a control strategy formulation module; The fault prediction model includes: a data preprocessing module, an extended Kalman filter, a codec-based neural network module, a backpropagation neural network module, a fully connected module, a first training module, and a second training module; The data preprocessing module is used to process the collected production line operating status data, operating image data, and noise data to generate cross-modal time series feature sequences. Extended Kalman filter is used to remove noise and outliers from cross-modal time series feature sequences; The encoder-decoder-based neural network module includes an encoder and a decoder, which are connected through a recursive structure. Both the encoder and decoder use a BiLSTM module as the core to capture the temporal information, context and forward and reverse sequence features in the cross-modal temporal feature sequence to obtain the first fault prediction result sequence. The BP neural network module is used to obtain the second fault prediction result sequence based on the cross-modal temporal feature sequence; The fully connected module is used to sum the first fault prediction result sequence and the second fault prediction result sequence according to preset weights to obtain the final prediction result sequence. The control strategy formulation module is used to formulate control strategies based on the final prediction result sequence.

2. The intelligent control system for a sealing component production line according to claim 1, characterized in that, The edge computing layer also includes a first training module for training a codec-based neural network module; The training of the encoder-decoder-based neural network module includes: The cross-modal temporal feature sequence X is input into the encoder; The BiLSTM module in the encoder performs mapping analysis and feature extraction on the cross-modal temporal feature sequence X and obtains the hidden state after recursive update. And output sequence Y, and obtain the hidden state Provided to the BiLSTM module in the decoder; The decoder obtains the hidden state Then, hide the state. Transform into the decoder's initial state And based on the encoder's output sequence Y, the prediction result sequence for the next time step is obtained; The predicted sequence for the next time step is recursively fed back into the encoder-decoder-based neural network module. This process is repeated multiple times to obtain an output sequence of length T, which is the first fault prediction result sequence.

3. The intelligent control system for a sealing component production line according to claim 1, characterized in that, The edge computing layer also includes a second training module for training the BP God Network model, wherein training the BP God Network model includes: Calculate the output value and linear combination result of each hidden layer neuron in the hidden layer, as well as the output value and linear combination result of each hidden layer neuron in the output layer during forward propagation. Calculate the value of the cross-loss function; During backpropagation, the error terms of the output layer and the hidden layer are calculated separately; Update the weights and biases from the hidden layer to the output layer.

4. An intelligent control method for a sealing component production line, characterized in that, Includes the following steps: Step 1: Collect production line data and preprocess the collected production line data to obtain cross-modal time series feature sequences; Step 2: Remove noise and outliers from the cross-modal time series feature sequence using an extended Kalman filter to obtain the cross-modal time series feature sequence X; Step 3: Input the cross-modal temporal feature sequence X into the encoder-decoder-based neural network module and the BP neural network module to obtain the first fault prediction result sequence and the second fault prediction result sequence; Step 4: Sum the first fault prediction result sequence and the second fault prediction result sequence according to preset weights to obtain the final prediction result sequence; Step 5: Visualize the production line's operating status through a cloud-based monitoring platform and issue control commands to the edge computing layer.

5. The intelligent control method for a sealing component production line according to claim 4, characterized in that, Between steps two and three, the following is also included: The preprocessed production line data is divided into a training set and a dataset. Training the encoder-decoder-based neural network module; Training the BP God Network model.

6. The intelligent control method for a sealing component production line according to claim 5, characterized in that, The training of the encoder-decoder-based neural network module includes: Calculate the output value and linear combination result of each hidden layer neuron in the hidden layer, as well as the output value and linear combination result of each hidden layer neuron in the output layer during forward propagation. Calculate the value of the cross-loss function; During backpropagation, the error terms of the output layer and the hidden layer are calculated separately; Update the weights and biases from the hidden layer to the output layer.

7. The intelligent control method for a sealing component production line according to claim 5, characterized in that, The training of the BP God Network model includes: The cross-modal temporal feature sequence X is input into the encoder; The BiLSTM module in the encoder performs mapping analysis and feature extraction on the cross-modal temporal feature sequence X and obtains the hidden state after recursive update. And output sequence Y, and obtain the hidden state Provided to the BiLSTM module in the decoder; The decoder obtains the hidden state Then, hide the state. Transform into the decoder's initial state And based on the encoder's output sequence Y, the prediction result sequence for the next time step is obtained; The predicted sequence for the next time step is recursively fed back into the encoder-decoder-based neural network module. This process is repeated multiple times to obtain an output sequence of length T, which is the first fault prediction result sequence.

8. A computer-readable storage medium comprising a stored program, characterized in that, The program controls the electrical equipment containing the computer-readable storage medium to perform the intelligent control method for a sealing component production line as described in any one of claims 4 to 7 during program execution.

Images