Linear motor demagnetization fault diagnosis system and method based on flux density signal

Abstract

The invention relates to a linear motor demagnetization fault diagnosis system and method based on a flux density signal, and the system comprises the steps: collecting an air gap flux density signal in the operation process of a motor through an acquisition module, taking the air gap flux density signal as an effective demagnetization fault signal, carrying out the data preprocessing through a preprocessing mode based on wavelet transformation, so as to obtain air gap flux density data, and carrying out the fault diagnosis of the demagnetization fault. And noise interference can be eliminated. A prediction module is arranged to synchronously extract and fuse local features and global features of air gap flux density feature signals, and finally real-time monitoring and demagnetization fault diagnosis of the running state of the linear motor are achieved, so that the problems that in a traditional diagnosis method, demagnetization fault positioning is low in accuracy and poor in stability are solved. And a foundation is laid for predictive maintenance, regular overhaul and fault-tolerant compensation of the linear motor, so that efficient and reliable operation of a fault diagnosis system is ensured.

Description

Technical Field

[0001] This invention relates to the field of motor fault diagnosis technology, and more specifically, to a linear motor demagnetization fault diagnosis system and method based on magnetic flux density signals. Background Technology

[0002] Given the high demands of modern industrial production on motor performance and reliability, and the potentially serious consequences of failing to diagnose and repair motor faults in a timely manner, motor demagnetization fault diagnosis is imperative. Existing technologies for permanent magnet motor demagnetization fault diagnosis include model-driven diagnostic methods, high-frequency signal injection diagnostic methods, and data-driven diagnostic methods.

[0003] Model-driven diagnostic methods construct mathematical analytical models based on the electromagnetic constraints of the motor and use classical state estimation or state parameter identification methods to diagnose motor demagnetization faults. However, model-driven diagnostic methods require the motor to be symmetrical, involve large computational loads in the model, and the measurement results are easily affected by noise.

[0004] High-frequency signal injection diagnostic methods involve injecting high-frequency pulse signals into the motor control system via a controller, and then analyzing changes in key motor system parameters or high-frequency response characteristics to accurately diagnose demagnetization faults. However, high-frequency signal injection diagnostic methods cannot achieve real-time motor fault diagnosis, and the injection of high-frequency signals increases harmonic currents in the motor, which not only affects the overall performance of the motor but also introduces deviations into the fault diagnosis results.

[0005] Data-driven diagnostic methods acquire output signals such as motor current, voltage, vibration, noise, and magnetic field, and use signal processing techniques to extract features from the fault signals. By comparing these features with those under normal conditions, demagnetization faults in the motor can be diagnosed. Effective demagnetization fault signals in data-driven diagnostic methods mainly include current, voltage, and magnetic flux density signals. Commonly used signal processing techniques include Fourier transform, wavelet transform, and Hilbert transform. Data-driven diagnostic methods introduce deep learning mechanisms to achieve automatic extraction of data features, significantly improving fault diagnosis efficiency. However, commonly used signal processing methods still have shortcomings in extracting motor demagnetization fault features; how to further improve the accuracy of fault diagnosis remains a key issue that urgently needs to be addressed. Summary of the Invention

[0006] The technical problem to be solved by this invention is how to overcome the technical defects of existing permanent magnet motor demagnetization fault diagnosis methods, such as large model calculation volume, introduction of deviation in fault diagnosis results, and difficulty in extracting motor demagnetization fault characteristics. To overcome these technical defects, this invention provides a linear motor demagnetization fault diagnosis system and method based on magnetic flux density signals, specifically including a linear motor demagnetization fault diagnosis system based on magnetic flux density signals and a linear motor demagnetization fault diagnosis method based on magnetic flux density signals.

[0007] This invention provides a linear motor demagnetization fault diagnosis system based on magnetic flux density signals, comprising:

[0008] The acquisition module is used to monitor the air gap magnetic flux density signal during the operation of the linear motor, and to perform data preprocessing on the air gap magnetic flux density signal based on wavelet transform to obtain air gap magnetic flux density data.

[0009] The prediction module, electrically connected to the acquisition module, is used to sequentially perform local feature extraction based on convolution operation and global feature extraction based on converter model algorithm on the air gap magnetic flux density data to obtain demagnetization fault diagnosis results.

[0010] An alarm device, electrically connected to the prediction module, is used to receive the demagnetization fault diagnosis results and issue a warning signal when the linear motor is determined to have a demagnetization fault.

[0011] The demagnetization fault diagnosis system for linear motors based on magnetic flux density signals disclosed in this invention acquires air gap magnetic flux density signals during motor operation using a data acquisition module. These signals serve as valid demagnetization fault signals. Preprocessing based on wavelet transform eliminates noise interference by obtaining the air gap magnetic flux density data. A prediction module simultaneously extracts and fuses local and global features of the air gap magnetic flux density characteristic signal, ultimately enabling real-time monitoring of the linear motor's operating status and demagnetization fault diagnosis. This addresses the low accuracy and poor stability of traditional diagnostic methods for demagnetization fault location, laying the foundation for predictive maintenance, periodic overhauls, and fault-tolerant compensation for linear motors, thereby ensuring the efficient and reliable operation of the fault diagnosis system.

[0012] In one possible implementation, the acquisition module includes:

[0013] A sensor assembly for monitoring the air gap magnetic flux density signal during the operation of the linear motor;

[0014] The preprocessing device, electrically connected to the sensor assembly, is configured to perform data preprocessing on the air gap magnetic flux density signal using a wavelet transform-based preprocessing method to obtain the air gap magnetic flux density data.

[0015] The acquisition module with the above structure can monitor the air gap magnetic flux density signal during the operation of the linear motor in real time. Then, the acquired air gap magnetic flux density data is preprocessed by the preprocessing device, which not only ensures high-efficiency data acquisition, but also overcomes noise interference.

[0016] In one possible implementation, the sensor assembly includes a base and multiple fiber optic magnetic field sensors mounted on the base in a number adapted to the number of permanent magnets in the linear motor. All of the fiber optic magnetic field sensors are electrically connected to the preprocessing device. This approach can acquire magnetic field information from multiple nodes in real time, and the acquired magnetic field information reflects the magnetic state of the permanent magnets at that node, further improving the efficiency and accuracy of linear motor demagnetization fault diagnosis.

[0017] In one possible implementation, the preprocessing device is configured to perform the following steps:

[0018] A1: Receive all the air gap magnetic flux density signals obtained by the fiber optic magnetic field sensor and integrate them into a data vector;

[0019] A2: Perform data cleaning on the data vector to eliminate or correct inaccurate, incomplete, duplicate and / or invalid records, and obtain cleaning results;

[0020] A3: Wavelet transform technology is used to suppress sensor noise and remove abnormal data points in the cleaning results to obtain the transform results;

[0021] A4: Standardize the transformation result to obtain the air gap magnetic flux density data.

[0022] The preprocessing device that performs the above steps cleans the received data. Wavelet transform technology can suppress sensor noise and remove abnormal data points. Then, standardization of the air gap magnetic flux density data can ensure the convergence and stability of the model.

[0023] In one possible implementation, the prediction module includes:

[0024] The convolutional unit, electrically connected to the preprocessing device, is configured to extract local features of the time-series data from the air gap magnetic flux density data through multi-layer convolution and pooling operations to obtain local features.

[0025] The converter, which is electrically connected to both the convolutional unit and the alarm device, is configured to receive the local features, capture long-distance dependencies in time-series data through a multi-head self-attention mechanism, obtain the correlation of air gap magnetic flux density at each node, and obtain the demagnetization fault diagnosis result based on a multi-level feature extraction and fusion mechanism of fault features.

[0026] The prediction module with the above structure automatically extracts local features of the air gap magnetic flux density data from the time series data through convolutional units, and obtains diagnostic results by using a converter to capture long-distance dependencies in the time series data using a multi-head self-attention mechanism, thereby further improving the diagnostic accuracy.

[0027] In one possible implementation, the converter includes:

[0028] The encoder, electrically connected to the convolutional unit, is configured to encode the local features through a fusion of multi-head attention mechanism, residual connection and feedforward neural network mapping to obtain an encoded sequence;

[0029] The decoder, which is electrically connected to both the convolutional unit and the encoder, is configured to extract long-range dependencies from the encoded sequence through a fusion of multi-head attention mechanism, residual connections, and feedforward neural network mapping, and obtain the demagnetization fault diagnosis result based on the long-range dependencies through a fully connected layer.

[0030] Another technical solution of the present invention is to provide a method for diagnosing demagnetization faults of a linear motor based on magnetic flux density signals, comprising the following steps:

[0031] S1: Establish communication between the output of the convolutional network and the input of the converter network to obtain a convolution-converter model. Optimize the parameters of the convolution-converter model using simulated motor demagnetization fault data to obtain a prediction module.

[0032] S2: The air gap magnetic flux density signal during the operation of the linear motor is monitored by the acquisition module, and the air gap magnetic flux density signal is preprocessed based on the wavelet transform preprocessing method to obtain air gap magnetic flux density data.

[0033] S3: The prediction module sequentially performs local feature extraction based on convolution operation and global feature extraction based on converter model algorithm on the air gap magnetic flux density data to obtain demagnetization fault diagnosis results.

[0034] S4: Receive the demagnetization fault diagnosis result through the alarm device, and issue a warning signal when it is determined that the linear motor has a demagnetization fault.

[0035] The demagnetization fault diagnosis method for linear motors based on magnetic flux density signals disclosed in this invention first optimizes the parameters of the convolution-converter model using simulated demagnetization fault data to obtain a prediction module. Then, an acquisition module collects the air gap magnetic flux density signal during motor operation, using this as the effective demagnetization fault signal. Data preprocessing based on wavelet transform is then performed to obtain air gap magnetic flux density data, eliminating noise interference. Subsequently, the prediction module simultaneously extracts and fuses the local and global features of the air gap magnetic flux density characteristic signal, ultimately achieving real-time monitoring of the linear motor's operating status and demagnetization fault diagnosis. This solves the problems of low accuracy and poor stability in demagnetization fault location in traditional diagnostic methods, laying the foundation for predictive maintenance, periodic overhaul, and fault-tolerant compensation of linear motors, thereby ensuring the efficient and reliable execution of the fault diagnosis process.

[0036] In one possible implementation, step S1 includes the following steps:

[0037] S11: Establish communication between the output of the convolutional network and the input of the converter network to obtain the convolution-converter model;

[0038] S12: Collect the simulated data of the motor demagnetization fault, and perform data preprocessing on the demagnetization fault data based on the wavelet transform preprocessing method to obtain the simulated data preprocessing result;

[0039] S13: Input the simulation data preprocessing results into the convolution-converter model to obtain fault category information using the forward propagation of the convolution-converter model;

[0040] S14: Calculate the model error gradient based on the fault category information using the model error gradient calculation formula, and update the parameters of the convolution-transformer model through the backpropagation update method of the model error gradient;

[0041] S15: Determine whether the current convolution-transformer model meets the accuracy index requirements of the confusion matrix.

[0042] If so, the current convolution-converter model is embedded into the chip to obtain the prediction module;

[0043] If not, then return to step S11.

[0044] The above method uses existing demagnetization fault data to train the convolutional-converter model, enabling it to determine whether the linear motor is in normal operation or demagnetization fault condition from data features. After training, the accuracy of the confusion matrix is ​​used as an evaluation metric to measure the model's performance in demagnetization fault diagnosis, further improving the model's generalization ability. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of a linear motor demagnetization fault diagnosis system based on magnetic flux density signal disclosed in an embodiment of the present invention.

[0046] Figure 2 This is a schematic diagram of motor air gap magnetic flux density monitoring disclosed in an embodiment of the present invention;

[0047] Figure 3 This is a schematic diagram illustrating the operating principle of the converter model disclosed in this embodiment of the invention.

[0048] Figure 4 This is a flowchart of the method disclosed in the embodiments of the present invention.

[0049] Explanation of reference numerals in the attached figures:

[0050] 1. Base, 2. Permanent magnet, 3. Fiber optic magnetic field sensor, 4. Mover core, 5. Coil, 6. Stator. Detailed Implementation

[0051] First, those skilled in the art should understand that these embodiments are merely used to explain the technical principles of the embodiments of this application and are not intended to limit the scope of protection of the embodiments of this application. Those skilled in the art can make adjustments as needed to adapt to specific application scenarios.

[0052] In the description of the embodiments of this application, it should be noted that, unless otherwise explicitly specified and limited, the technical terms "electrical connection" and "establishing an electrical connection relationship" should be interpreted broadly, that is, it should be understood that both or more have an electrical relationship, which can be achieved through a wire, a radio connection, or a combination of both; it can be a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms in the embodiments of this application according to the specific circumstances.

[0053] In the description of the embodiments of this application, it should be noted that, unless otherwise explicitly specified and limited, the term "forming a communication link structure" means that the multiple communication elements or modules involved form a network structure or network link structure through communication connection. Communication or communication connection means that there is information transmission between the first feature and the second feature. This information transmission can be unidirectional or bidirectional. The communication connection can be realized by electrical connection of wires, radio connection, electrical connection of electromagnetic media (such as optical fiber, semiconductor), communication realized by channel, etc.

[0054] In the embodiments of this application, unless otherwise expressly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "on top of," and "over" the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0055] The present application will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0056] See Figures 1-4 This application discloses a linear motor demagnetization fault diagnosis system based on magnetic flux density signals. Figure 1 This is a schematic diagram of the demagnetization fault diagnosis system. The demagnetization fault diagnosis system includes an acquisition module, a prediction module, and an alarm device. The prediction module is electrically connected to the acquisition module, and the alarm device is electrically connected to the prediction module.

[0057] See Figure 1 and Figure 2 In this demagnetization fault diagnosis system, the acquisition module monitors the air gap magnetic flux density signal during the operation of the linear motor and performs data preprocessing on the air gap magnetic flux density signal based on wavelet transform to obtain air gap magnetic flux density data. In this embodiment, the acquisition module includes a sensor assembly and a preprocessing device, which is electrically connected to the sensor assembly.

[0058] In the acquisition module, the sensor assembly is used to monitor the air gap magnetic flux density signal during the operation of the linear motor. See also... Figure 2 The sensor assembly is located below the permanent magnet 2. The sensor assembly includes a base 1 and multiple fiber optic magnetic field sensors 3 mounted on the base 1 in an number equal to the number of permanent magnets 2 of the linear motor. All fiber optic magnetic field sensors 3 are electrically connected to the pretreatment device.

[0059] In the acquisition module, the preprocessing device is configured to perform data preprocessing on the air gap magnetic flux density signal using a wavelet transform-based preprocessing method to obtain air gap magnetic flux density data. Specifically, in this embodiment, the preprocessing device is configured to perform the following steps: A1: Receive all air gap magnetic flux density signals acquired by the fiber optic magnetic field sensor 3 and integrate them into a data vector; A2: Clean the data vector to eliminate or correct inaccurate, incomplete, duplicate, and / or invalid records to obtain a cleaning result; A3: Use wavelet transform technology to suppress sensor noise and remove abnormal data points in the cleaning result to obtain a transform result; A4: Standardize the transform result to obtain air gap magnetic flux density data.

[0060] See Figure 1 and Figure 3 In this demagnetization fault diagnosis system, the prediction module is used to sequentially perform local feature extraction based on convolution operations and global feature extraction based on a converter model algorithm on the air gap magnetic flux density data to obtain the demagnetization fault diagnosis results. For example... Figure 1 As shown, in this embodiment, the prediction module includes a convolutional unit (CNN) and a transformer. The CNN is electrically connected to the preprocessing device, and the transformer is electrically connected to both the CNN and the alarm device.

[0061] In the prediction module, the convolutional unit is configured to extract local features from the time series data of the air gap magnetic flux density data through multi-layer convolution and pooling operations to obtain local features; the converter is configured to receive the local features, capture the long-distance dependencies in the time series data through a multi-head self-attention mechanism, obtain the correlation of the air gap magnetic flux density of each node, and obtain the demagnetization fault diagnosis result based on the multi-level feature extraction and fusion mechanism of fault features.

[0062] The converter includes an encoder and a decoder. The encoder is electrically connected to a convolutional unit, while the decoder is electrically connected to both the convolutional unit and the encoder. The encoder is configured to encode local features using a fusion of multi-head attention, residual connections, and feedforward neural network mapping to obtain an encoded sequence. The decoder is configured to extract long-range dependencies from the encoded sequence using a fusion of multi-head attention, residual connections, and feedforward neural network mapping, and then use a fully connected layer to obtain demagnetization fault diagnosis results based on these long-range dependencies. Figure 3 In the illustrated workflow, the encoder executes the process on the left, while the decoder executes the process on the right. Finally, the decoder obtains the prediction result, which is presented in the form of probability.

[0063] In this demagnetization fault diagnosis system, an alarm device receives the demagnetization fault diagnosis results and issues a warning signal when a demagnetization fault is determined to have occurred in the linear motor. When an abnormal signal pattern is detected, the alarm device triggers an alarm mechanism and performs corresponding checks and repairs based on the fault type output by the model.

[0064] The following will further disclose the operation method of the linear motor demagnetization fault diagnosis system based on magnetic flux density signal in this embodiment. The overall flowchart of the method is as follows: Figure 4 As shown, the method includes the following steps:

[0065] S1: Establish communication between the output of the convolutional network and the input of the converter network to obtain the convolution-converter model (i.e., the CNN-Transformer motor demagnetization fault diagnosis model in the figure). Use the simulated data of motor demagnetization fault to optimize the parameters of the convolution-converter model to obtain the prediction module.

[0066] See Figure 4 In this embodiment, step S1 includes the following steps:

[0067] S11: Establish communication between the output of the convolutional network and the input of the converter network to obtain the convolution-converter model.

[0068] S12: Collect simulated data of motor demagnetization fault, and perform data preprocessing on the demagnetization fault data based on wavelet transform to obtain the simulation data preprocessing results.

[0069] S13: Input the preprocessed simulation data into the convolution-converter model to obtain fault category information using the forward propagation of the convolution-converter model.

[0070] S14: Calculate the model error gradient based on the fault category information using the model error gradient calculation formula, and update the parameters of the convolution-transformer model through the backpropagation update method of the model error gradient.

[0071] S15: Determine whether the current convolution-transformer model meets the accuracy requirements of the confusion matrix.

[0072] If so, the current convolution-transformer model is embedded into the chip to obtain the prediction module;

[0073] If not, then return to step S11.

[0074] S2: The air gap magnetic flux density signal during the operation of the linear motor is monitored by the acquisition module, and the air gap magnetic flux density signal is preprocessed based on the wavelet transform preprocessing method to obtain air gap magnetic flux density data.

[0075] S3: The prediction module sequentially performs local feature extraction based on convolution operation and global feature extraction based on converter model algorithm on the air gap magnetic flux density data to obtain demagnetization fault diagnosis results.

[0076] S4: Receives the demagnetization fault diagnosis results through the alarm device and issues a warning signal when it is determined that the linear motor has a demagnetization fault.

[0077] The linear motor demagnetization fault diagnosis system based on magnetic flux density signals disclosed in this embodiment acquires air gap magnetic flux density signals during motor operation using a data acquisition module. These signals serve as valid demagnetization fault signals. Preprocessing based on wavelet transform eliminates noise interference by obtaining air gap magnetic flux density data. A prediction module simultaneously extracts and fuses local and global features of the air gap magnetic flux density characteristic signal, ultimately enabling real-time monitoring of the linear motor's operating status and demagnetization fault diagnosis. This addresses the low accuracy and poor stability of demagnetization fault location in traditional diagnostic methods, laying the foundation for predictive maintenance, periodic overhauls, and fault-tolerant compensation of linear motors, thereby ensuring the efficient and reliable operation of the fault diagnosis system.

[0078] In the description of the embodiments of this application, it should be noted that the terms "inner" and "outer" and other terms indicating direction or positional relationship are based on the direction or positional relationship shown in the drawings. This is only for the convenience of description and does not indicate or imply that the device or component must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this application.

[0079] In the description of this application, the references to terms such as "an embodiment," "some embodiments," "in this embodiment," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0080] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A linear motor demagnetization fault diagnosis system based on magnetic flux density signals, characterized in that, include: The acquisition module is used to monitor the air gap magnetic flux density signal during the operation of the linear motor, and to perform data preprocessing on the air gap magnetic flux density signal based on wavelet transform to obtain air gap magnetic flux density data. The prediction module, electrically connected to the acquisition module, is used to sequentially perform local feature extraction based on convolution operation and global feature extraction based on converter model algorithm on the air gap magnetic flux density data to obtain demagnetization fault diagnosis results. An alarm device, electrically connected to the prediction module, is used to receive the demagnetization fault diagnosis results and issue a warning signal when the linear motor is determined to have a demagnetization fault.

2. The linear motor demagnetization fault diagnosis system based on magnetic flux density signal according to claim 1, characterized in that, The acquisition module includes: A sensor assembly for monitoring the air gap magnetic flux density signal during the operation of the linear motor; The preprocessing device, electrically connected to the sensor assembly, is configured to perform data preprocessing on the air gap magnetic flux density signal using a wavelet transform-based preprocessing method to obtain the air gap magnetic flux density data.

3. The linear motor demagnetization fault diagnosis system based on magnetic flux density signal according to claim 2, characterized in that, The sensor assembly includes a base (1) and a plurality of fiber optic magnetic field sensors (3) disposed on the base (1) in a number adapted to the number of permanent magnets (2) of the linear motor. All of the fiber optic magnetic field sensors (3) are electrically connected to the preprocessing device.

4. The linear motor demagnetization fault diagnosis system based on magnetic flux density signal according to claim 3, characterized in that, The preprocessing device is configured to perform the following steps: A1: Receive all the air gap magnetic flux density signals obtained by the fiber optic magnetic field sensor (3) and integrate them into a data vector; A2: Perform data cleaning on the data vector to eliminate or correct inaccurate, incomplete, duplicate and / or invalid records, and obtain cleaning results; A3: Wavelet transform technology is used to suppress sensor noise and remove abnormal data points in the cleaning results to obtain the transform results; A4: Standardize the transformation result to obtain the air gap magnetic flux density data.

5. The linear motor demagnetization fault diagnosis system based on magnetic flux density signal according to any one of claims 2-4, characterized in that, The prediction module includes: The convolutional unit, electrically connected to the preprocessing device, is configured to extract local features of the time-series data from the air gap magnetic flux density data through multi-layer convolution and pooling operations to obtain local features. The converter, which is electrically connected to both the convolutional unit and the alarm device, is configured to receive the local features, capture long-distance dependencies in time-series data through a multi-head self-attention mechanism, obtain the correlation of air gap magnetic flux density at each node, and obtain the demagnetization fault diagnosis result based on a multi-level feature extraction and fusion mechanism of fault features.

6. The linear motor demagnetization fault diagnosis system based on magnetic flux density signal according to claim 5, characterized in that, The converter includes: The encoder, electrically connected to the convolutional unit, is configured to encode the local features through a fusion of multi-head attention mechanism, residual connection and feedforward neural network mapping to obtain an encoded sequence; The decoder, which is electrically connected to both the convolutional unit and the encoder, is configured to extract long-range dependencies from the encoded sequence through a fusion of multi-head attention mechanism, residual connections, and feedforward neural network mapping, and obtain the demagnetization fault diagnosis result based on the long-range dependencies through a fully connected layer.

7. A method for diagnosing demagnetization faults in linear motors based on magnetic flux density signals, characterized in that, The linear motor demagnetization fault diagnosis system based on magnetic flux density signals according to any one of claims 1-6 includes the following steps: S1: Establish communication between the output of the convolutional network and the input of the converter network to obtain a convolution-converter model. Optimize the parameters of the convolution-converter model using simulated motor demagnetization fault data to obtain a prediction module. S2: The air gap magnetic flux density signal during the operation of the linear motor is monitored by the acquisition module, and the air gap magnetic flux density signal is preprocessed based on the wavelet transform preprocessing method to obtain air gap magnetic flux density data. S3: The prediction module sequentially performs local feature extraction based on convolution operation and global feature extraction based on converter model algorithm on the air gap magnetic flux density data to obtain demagnetization fault diagnosis results. S4: Receive the demagnetization fault diagnosis result through the alarm device, and issue a warning signal when it is determined that the linear motor has a demagnetization fault.

8. The method for diagnosing demagnetization faults of a linear motor based on magnetic flux density signals according to claim 7, characterized in that, Step S1 includes the following steps: S11: Establish communication between the output of the convolutional network and the input of the converter network to obtain the convolution-converter model; S12: Collect the simulated data of the motor demagnetization fault, and perform data preprocessing on the demagnetization fault data based on the wavelet transform preprocessing method to obtain the simulated data preprocessing result; S13: Input the simulation data preprocessing results into the convolution-converter model to obtain fault category information using the forward propagation of the convolution-converter model; S14: Calculate the model error gradient based on the fault category information using the model error gradient calculation formula, and update the parameters of the convolution-transformer model through the backpropagation update method of the model error gradient; S15: Determine whether the current convolution-transformer model meets the accuracy index requirements of the confusion matrix. If so, the current convolution-converter model is embedded into the chip to obtain the prediction module; If not, then return to step S11.

Images