A method and system for motor condition monitoring based on FBG
By using an FBG-based sensor network and a Transformer neural network model, multi-parameter monitoring of motor status was achieved, solving the problem of poor anti-interference capability of traditional sensors and improving the accuracy of motor status monitoring.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- WUXI UNIV
- Filing Date
- 2026-05-06
- Publication Date
- 2026-07-31
AI Technical Summary
Existing motor condition monitoring technologies are limited by the detection modes of traditional sensors, have poor resistance to external interference, are difficult to achieve multi-parameter distributed monitoring, and are easily affected by the superposition of multiple fields inside the motor, making it difficult to accurately reflect the motor condition.
An FBG-based sensor network is used to collect real-time data on strain, magnetic field, and temperature inside the motor. Through signal preprocessing and feature extraction, combined with a Transformer neural network model, diagnostics are performed to monitor the motor's condition.
It enables multi-parameter monitoring of motor status, reduces the impact on the internal structure of the motor, improves the accuracy and anti-interference ability of monitoring, and can more accurately reflect the normal and fault states of the motor.
Smart Images
Figure CN122487900A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of motor condition monitoring technology, and specifically to a motor condition monitoring method and system based on FBG. Background Technology
[0002] The stator windings, stator core, rotor, and bearings of an electric motor are prone to failure, often accompanied by overheating, torque reduction, excessive losses, pulsation, and noise. Many electrical and mechanical faults have characteristic features that can be detected to determine the abnormal operating state and cause of the motor. However, characteristic signals related to motor condition are easily obscured by noise, and the symptoms of multiple faults overlap, making single-parameter detection insufficient for motor condition monitoring. Comprehensive motor condition monitoring requires a greater variety and number of sensors. Current motor condition monitoring technologies are limited by traditional sensor detection modes, exhibiting poor resistance to external interference and susceptibility to the superposition of multiple fields within the motor, hindering the implementation of multi-parameter distributed monitoring. Summary of the Invention
[0003] To address the problems existing in the prior art, the present invention provides a motor condition monitoring method and system based on FBG.
[0004] Other features and advantages of this application will become apparent from the following detailed description, or may be learned in part from practice of this application.
[0005] According to a first aspect of this application, a motor condition monitoring method based on FBG is provided, comprising: Based on the FBG sensor network installed and fixed inside the motor, the center wavelength data of the FBG sensor network is collected in real time. The center wavelength data of the FBG sensor network are preprocessed to obtain the high-frequency center wavelength change related to strain and magnetic field, and the low-frequency center wavelength change related to temperature. Feature extraction is performed on the high-frequency center wavelength changes related to strain and magnetic field, and the low-frequency center wavelength changes related to temperature. MTF and trend term visualization are used respectively to obtain time-domain feature images related to strain and magnetic field, and time-domain feature images related to temperature. Multi-feature fusion is performed on time-domain feature images related to strain, magnetic field, and temperature by stacking them by channel to obtain fused feature images; A pre-established Transformer neural network model is used to process the fused feature images to obtain the motor state results.
[0006] In some embodiments of this application, based on the foregoing scheme, obtaining the high-frequency center wavelength change related to strain and magnetic field, and obtaining the low-frequency center wavelength change related to temperature, based on the FBG sensing network installed and fixed inside the motor, includes: Multiple bare fiber Bragg gratings are circumferentially pasted on the inner surface of the motor housing near the stator silicon steel sheet at one end of the motor, and multiple nickel-plated FBGs are installed and fixed at the stator winding of the motor. Based on the bare fiber grating, a first center wavelength change is obtained, and based on the nickel-plated FBG, a second center wavelength change is obtained. The first center wavelength change was filtered and corrected by using the least squares method combined with a Butterworth high-pass filter to obtain the high-frequency center wavelength change related to strain. The second center wavelength change was filtered and corrected by using the least squares method combined with a Butterworth high-pass filter to obtain the high-frequency center wavelength change related to the magnetic field. Adaptive mode decomposition is used to extract the trend term from the first and second center wavelength changes to obtain the temperature-related low-frequency center wavelength changes.
[0007] In some embodiments of this application, based on the foregoing scheme, obtaining the first center wavelength change based on the bare fiber grating includes: Based on the temperature change at the location of the bare grating inside the motor And the circumferential strain change on the inner surface of the outer casing near the stator silicon steel sheet at one end of the motor. The change in the first center wavelength caused by thermal expansion and strain effects The expression is: ; in: For temperature coefficient, For strain coefficient, , The effective elastic coefficient of the fiber grating. This is the initial value of the first center wavelength.
[0008] In some embodiments of this application, based on the foregoing scheme, the second center wavelength change is obtained by coating the FBG with the nickel plating: Based on the temperature change of the end winding and the change in magnetic flux density The change in the second center wavelength caused by thermal expansion and magnetostriction effects The expression is: ; in: It is the temperature coefficient. It is the magnetic flux density coefficient. This is the initial value for the second center wavelength.
[0009] In some embodiments of this application, based on the foregoing scheme, feature extraction is performed on the high-frequency center wavelength changes related to strain and magnetic field, and the low-frequency center wavelength changes related to temperature. These features are then visualized using MTF and trend terms, respectively, to obtain time-domain feature images related to strain and magnetic field, and time-domain feature images related to temperature. Two-dimensional images of the strain-related high-frequency center wavelength changes were generated using MTF to obtain strain-related time-domain feature images. Two-dimensional images of the high-frequency center wavelength changes related to the magnetic field are generated using MTF to obtain time-domain feature images related to the magnetic field. A time-domain image of the temperature-related low-frequency center wavelength variation was generated using trend term visualization to obtain a temperature-related time-domain feature image.
[0010] In some embodiments of this application, based on the foregoing scheme, the step of performing multi-feature fusion on feature images related to strain, magnetic field, and temperature to obtain fused feature images includes: Multiple frames of time-domain feature images related to strain, magnetic field, and temperature are stacked by channel to obtain a fused feature image.
[0011] In some embodiments of this application, based on the foregoing scheme, the method for establishing the Transformer neural network model is as follows: The Transformer neural network model is trained by fusing a set of feature images, which includes multi-feature fusion images related to strain, magnetic field, and temperature under four states: normal motor operation, static air gap eccentricity, dynamic air gap eccentricity, and insulation layer damage.
[0012] In some embodiments of this application, based on the aforementioned scheme, the bare fiber Bragg grating includes a first bare fiber Bragg grating, a second bare fiber Bragg grating, a third bare fiber Bragg grating, and a fourth bare fiber Bragg grating. The first, second, third, and fourth bare fiber Bragg gratings are circumferentially attached to the inner surface of the motor housing near the stator silicon steel sheet at one end using adhesive. The first, second, third, and fourth bare fiber Bragg gratings are evenly distributed circumferentially on the inner surface of the motor housing, and are spaced apart by 90 degrees. The first, second, third, and fourth bare fiber Bragg gratings are all connected end-to-end in series on the same optical fiber, with the first bare fiber Bragg grating at the beginning and the fourth bare fiber Bragg grating at the end. The nickel-plated FBG includes a first nickel-plated FBG, a second nickel-plated FBG, and a third nickel-plated FBG. One end of each of the first, second, and third nickel-plated FBGs is fixed to the gap in the end winding of the motor stator using adhesive. The first, second, and third nickel-plated FBGs are evenly distributed circumferentially around the end winding of the motor stator, and are spaced 120 degrees apart from each other.
[0013] According to a second aspect of this application, an FBG-based motor condition monitoring system is provided, comprising: Bare fiber Bragg gratings, multiple bare fiber Bragg gratings are circumferentially attached to the inner surface of the housing near the stator silicon steel sheet at one end of the motor using adhesive at both ends; A nickel-plated FBG is coated, and multiple nickel-plated FBGs are installed and fixed at the end windings of the motor stator. The high-speed demodulator is connected to the bare fiber grating and the nickel-plated FBG via optical fiber, and is used to acquire the center wavelength of the FBG sensor network in real time. The host computer is used to perform signal preprocessing, feature extraction, and multi-feature fusion on the center wavelength data of the FBG sensor network to obtain a multi-feature fused feature image related to strain, magnetic field, and temperature. The fused feature image is then processed by a pre-established Transformer neural network to obtain the motor state result. A display is used to show the bare fiber Bragg grating, the nickel-plated FBG, and the motor status results.
[0014] According to a third aspect of this application, a computer-readable storage medium is provided that stores a computer program thereon, the computer program including executable instructions that, when executed by a processor, implement the method described above.
[0015] The beneficial effects of this application are as follows: This application provides a motor condition monitoring method and system based on FBG (Fiber Bragg Grating). A sensor network is constructed using a bare fiber optic grating and a nickel-plated FBG to monitor internal motor strain, magnetic field, and temperature. The FBG center wavelength data undergoes signal preprocessing to resolve the coupling issues between strain and temperature, and between magnetic field and temperature. Feature extraction is performed on the decoupled data to obtain feature images related to the motor condition. A Transformer neural network model is then used to diagnose the current motor condition, and the results are displayed in real-time via a human-machine interface. Current motor condition monitoring systems rely on limited detection parameters and signal processing methods, failing to provide rich information for motor condition diagnosis. In contrast, the sensor structure used in this application is small in size, and the implantation of multiple sensors does not alter the internal structure of the motor. It can simultaneously monitor multiple internal parameters, sensitively capturing various features related to the motor condition, and is less susceptible to the influence of multiple superimposed fields within the motor, thus more accurately reflecting the normal and various fault states of the motor.
[0016] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and do not limit this application. Attached Figure Description
[0017] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and are intended to explain the invention, but do not constitute an undue limitation thereof. In the drawings: Figure 1 This is a flowchart of a motor condition monitoring method based on FBG according to the present invention; Figure 2 This is a schematic diagram of the diagnostic results of the Transformer test set in a specific embodiment of the present invention; Figure 3 This is a schematic diagram of the human-computer interaction interface of the present invention; Figure 4 This is a schematic diagram of a motor condition monitoring system based on FBG according to the present invention; Figure 5 This is a simplified schematic diagram showing the distribution of the FBG inside the motor according to the present invention.
[0018] Among them, 1 is a bare fiber Bragg grating; 11 is a first bare fiber Bragg grating; 12 is a second bare fiber Bragg grating; 13 is a third bare fiber Bragg grating; and 14 is a fourth bare fiber Bragg grating. 2. FBG coated with nickel plating; 21. FBG coated with first nickel plating; 22. FBG coated with second nickel plating; 23. FBG coated with third nickel plating; 3. Transmission optical fiber; 4. High-speed demodulator; 5. First cable; 6. Host computer; 7. Second cable; 8. Display. Detailed Implementation
[0019] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.
[0020] It should be understood that the terms "comprising" and other similar expressions in the specification, claims, and accompanying drawings of this invention are intended to cover a non-exclusive inclusion, such as a process, method, system, or apparatus that includes a series of steps or units and is not limited to the listed steps or units. Furthermore, "first" and "second" are used to distinguish different objects and are not intended to describe a specific order.
[0021] According to the first aspect of this application, Figure 1 As shown, this embodiment provides a motor condition monitoring method based on FBG, including: Step S1: Based on the FBG (Fiber Bragg Grating) sensor network installed and fixed inside the motor, collect the center wavelength data of the FBG sensor network in real time.
[0022] In some embodiments of this example, multiple bare fiber optic gratings 1 are circumferentially pasted on the inner surface of the outer casing near the stator silicon steel sheet at one end of the motor. Multiple nickel-plated FBGs 2 are installed and fixed at the stator winding end of the motor. In this way, a sensor network is formed to monitor strain, magnetic field, and temperature in real time. Based on the analysis of multiple characteristic parameters such as strain, magnetic field, and temperature, four motor states are obtained: normal motor, static air gap eccentricity, dynamic air gap eccentricity, and insulation layer damage. The motor status is displayed by the human-machine interface to realize status monitoring.
[0023] Step S2: Perform signal preprocessing on the center wavelength data of the FBG sensor network to obtain the high-frequency center wavelength change related to strain and magnetic field and the low-frequency center wavelength change related to temperature.
[0024] In some embodiments of this example, obtaining the first center wavelength change based on the bare fiber grating 1 includes: Based on the temperature change at the location of the bare grating inside the motor And the circumferential strain change on the inner surface of the outer casing near the stator silicon steel sheet at one end of the motor. The change in the first center wavelength caused by thermal expansion and strain effects The expression is: ; in: For temperature coefficient, For strain coefficient, , The effective elastic coefficient of the fiber grating. This is the initial value of the first center wavelength.
[0025] In some embodiments of this example, obtaining the second center wavelength change based on the nickel plating coating of FBG2 includes: Based on the temperature change of the end winding and the change in magnetic flux density The change in the second center wavelength caused by thermal expansion and magnetostriction effects The expression is: ; in: It is the temperature coefficient. It is the magnetic flux density coefficient. This is the initial value for the second center wavelength.
[0026] In this embodiment, a first center wavelength change is obtained based on the bare fiber grating 1, and a second center wavelength change is obtained based on the nickel plating coating FBG2. The first center wavelength change was filtered and corrected by using the least squares method combined with a Butterworth high-pass filter to obtain the high-frequency center wavelength change related to strain.
[0027] Specifically, a first fourth-order Butterworth high-pass filter is set, and the filter coefficients of the first fourth-order Butterworth high-pass filter are corrected based on the least squares method. The first center wavelength change is filtered and corrected by the corrected first fourth-order Butterworth high-pass filter to obtain the high-frequency center wavelength change related to strain.
[0028] The second center wavelength change is filtered and corrected by using the least squares method combined with a Butterworth high-pass filter to obtain the high-frequency center wavelength change related to the magnetic field.
[0029] Specifically, a second-order fourth-order Butterworth high-pass filter is set, and the filter coefficients of the second-order fourth-order Butterworth high-pass filter are corrected based on the least squares method. The second center wavelength change is filtered and corrected by the corrected second-order fourth-order Butterworth high-pass filter to obtain the high-frequency center wavelength change related to the magnetic field.
[0030] Adaptive mode decomposition is used to extract the trend term from the first and second center wavelength changes to obtain the temperature-related low-frequency center wavelength changes.
[0031] Thus, signal preprocessing is performed on the center wavelength signals of each FBG in the FBG sensor network to obtain center wavelength change data related to strain, magnetic field, and temperature information. Feature extraction and state diagnosis methods are then used to process the center wavelength change data to obtain the motor state.
[0032] Step S3: Extract features from the high-frequency center wavelength changes related to strain and magnetic field and the low-frequency center wavelength changes related to temperature to obtain time-domain feature images related to strain and magnetic field, and time-domain feature images related to temperature.
[0033] In some embodiments of this example, the feature extraction includes: Using a Markov Transition Field (MTF), the high-frequency center wavelength changes related to strain and magnetic field are converted into time-domain feature images related to strain and time-domain feature images related to magnetic field, respectively, with the first time period as the segment length. Trend visualization is used to generate a time-domain image of the temperature-related low-frequency center wavelength change. The temperature-related low-frequency center wavelength change is converted into a temperature-related time-domain feature image with a segment length of the second time period.
[0034] Specifically, the high-frequency center wavelength change related to strain is obtained, a first Markov transfer matrix is constructed, and the strain-related temporal feature image is converted into a segment with a first time period as the segment length through the first Markov transfer matrix.
[0035] The high-frequency center wavelength variation related to the magnetic field is obtained, and a second Markov transfer matrix is constructed. The second Markov transfer matrix is used to convert the image into a time-domain feature image related to the magnetic field with the first time period as the segment length.
[0036] The low-frequency trend term and trend term amplitude of the low-frequency center wavelength change related to temperature are extracted. The trend term visualization is used to generate time-domain images of the low-frequency trend term and trend term amplitude. The second time segment is converted into a time-domain feature map related to temperature.
[0037] Step S4: Feature layer fusion is used to perform multi-feature fusion on the time-domain feature images related to strain, the time-domain feature images related to magnetic field, and the time-domain feature images related to magnetic field to obtain fused feature images.
[0038] In some embodiments of this example, the feature fusion includes: Multiple frames of time-domain feature images related to strain, magnetic field, and temperature are stacked by channel to obtain a fused feature image.
[0039] Step S5: Process the fused feature images using a pre-established Transformer neural network model to obtain diagnostic classification results.
[0040] In this embodiment, center wavelength data of the FBG sensor network under four states—normal motor operation, static air gap eccentricity, dynamic air gap eccentricity, and insulation layer damage—are collected. After signal preprocessing, feature extraction, and feature fusion, the center wavelength data related to strain, magnetic field, and temperature are converted into fused feature maps. A Transformer neural network model is constructed, including multilayer perceptrons, patch layers, and patch coding layers. A batch of fused feature images and their corresponding labels are obtained by calling functions. A batch loading function and specific preprocessing in the Transformer framework are defined. Some hyperparameters of the model are trained, and the model is saved after training is completed.
[0041] In one specific embodiment, such as Figure 2 As shown, the accuracy of the diagnostic classification result obtained by using a pre-established Transformer neural network model to process the fused feature image is 85.8333%.
[0042] like Figure 3 The diagram shows a schematic of the human-computer interaction interface of the present invention. FBG1, FBG2, FBG3 and FBG4 are four bare fiber optic gratings circumferentially pasted on the inner surface of the outer shell near the stator silicon steel sheet at one end of the motor. MFBG1, MFBG2 and MFBG3 are three nickel-plated FBGs installed and fixed at the winding end of the motor stator. They can display the motor normal, air gap static eccentricity, air gap dynamic eccentricity and insulation layer status.
[0043] According to the second aspect of this application, such as Figure 4 and Figure 5 As shown, this embodiment provides a motor condition monitoring system based on FBG, comprising: The host computer is used to perform signal preprocessing, feature extraction, and multi-feature fusion on the center wavelength data of the FBG sensor network to obtain a multi-feature fused feature image related to strain, magnetic field, and temperature. The fused feature image is then processed using a pre-established Transformer neural network to obtain the motor state result.
[0044] A display is used to show the bare fiber Bragg grating, the nickel-plated FBG, and the motor status results.
[0045] A bare fiber Bragg grating 1, and multiple bare fiber Bragg gratings 1 are respectively attached circumferentially to the inner surface of the outer casing of the motor near the stator silicon steel sheet at both ends using adhesive; A nickel-plated FBG2 is coated, and multiple nickel-plated FBG2s are installed and fixed at the end windings of the motor stator. The high-speed demodulator 4 is connected to the bare fiber grating 1 and the nickel-plated FBG 2 via the transmission fiber 3, and is used to acquire the center wavelength data of the FBG sensor network in real time. The host computer 6 is electrically connected to the high-speed demodulator 4 via the first cable 5. It includes an interactive real-time monitoring system, comprising a signal preprocessing module, a feature extraction module, a feature fusion module, and a state diagnosis module. The signal preprocessing module preprocesses the center wavelength data of the FBG sensor network to obtain high-frequency center wavelength changes related to strain and magnetic field, and low-frequency center wavelength changes related to temperature. The feature extraction module uses MTF and trend term visualization to process the high-frequency and low-frequency center wavelength changes, thereby obtaining time-domain feature images related to strain and magnetic field, and time-domain feature images related to temperature. The feature fusion module uses feature layer fusion to perform multi-feature fusion on the time-domain feature images related to strain, magnetic field, and magnetic field to obtain fused feature images. The state diagnosis module uses a pre-established Transformer neural network to process the fused feature images to obtain the motor state results. The display 8 is electrically connected to the host computer 6 via the second cable 7, and is used to display the status results of the bare fiber grating 1, the nickel-plated FBG2, and the motor.
[0046] In some embodiments of this example, the bare fiber Bragg grating 1 includes a first bare fiber Bragg grating 11, a second bare fiber Bragg grating 12, a third bare fiber Bragg grating 13, and a fourth bare fiber Bragg grating 14. The first bare fiber Bragg grating 11, the second bare fiber Bragg grating 12, the third bare fiber Bragg grating 13, and the fourth bare fiber Bragg grating 14 are circumferentially attached to the inner surface of the motor housing near the stator silicon steel sheet at one end using adhesive. The first bare fiber Bragg grating 11, the second bare fiber Bragg grating 12, the third bare fiber Bragg grating 13, and the fourth bare fiber Bragg grating 14 are evenly distributed circumferentially on the inner surface of the motor housing, and are spaced apart by 90 degrees. The first bare fiber Bragg grating 11, the second bare fiber Bragg grating 12, the third bare fiber Bragg grating 13, and the fourth bare fiber Bragg grating 14 are all connected end to end on the same optical fiber, with the first bare fiber Bragg grating 11 at the beginning and the fourth bare fiber Bragg grating 14 at the end. The nickel-plated fiber Bragg grating 2 includes a first nickel-plated FBG21, a second nickel-plated FBG22, and a third nickel-plated FBG23. One end of each of these FBGs is adhesively fixed to the gap in the stator winding of the motor. These FBGs are spatially distributed circumferentially around the stator winding, spaced 120 degrees apart. The display 8 can show the first bare fiber Bragg grating 11, the second bare fiber Bragg grating 12, the first nickel-plated FBG21, the second nickel-plated FBG22, the third nickel-plated FBG23, and the motor status results.
[0047] Based on the above system, this embodiment analyzes and processes the signals collected by the FBG sensor network, detects the parameter values of each measuring point in real time, and outputs the detection values and status diagnosis results to the display 8, providing a basis for the staff to carry out the next operation.
[0048] Thus, this embodiment uses a small, compact FBG sensing element to construct a sensing network, which can simultaneously monitor multiple physical quantities inside the motor and extract rich features related to the motor's state. The sensing network and the signal processing and analysis methods used in this system eliminate the influence of multi-field coupling effects inside the motor, and can more accurately reflect the normal operation of the motor and various fault states.
[0049] Specifically, this embodiment corresponds one-to-one with the above method embodiments. The functions of each module have been described in detail in the corresponding method embodiments, so they will not be repeated here.
[0050] According to a third aspect of this application, this embodiment also provides a computer-readable storage medium having a computer program stored thereon, the computer program including executable instructions that, when executed by a processor, implement the above-described method.
[0051] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.
[0052] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.
[0053] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A motor condition monitoring method based on FBG, characterized in that, include: Based on the FBG sensor network installed and fixed inside the motor, the center wavelength data of the FBG sensor network is collected in real time. The center wavelength data of the FBG sensor network are preprocessed to obtain the high-frequency center wavelength change related to strain and magnetic field, and the low-frequency center wavelength change related to temperature. Feature extraction is performed on the high-frequency center wavelength changes related to strain and magnetic field, and the low-frequency center wavelength changes related to temperature. MTF and trend term visualization are used respectively to obtain time-domain feature images related to strain and magnetic field, and time-domain feature images related to temperature. Multi-feature fusion is performed on time-domain feature images related to strain, magnetic field, and temperature by stacking them by channel to obtain fused feature images; A pre-established Transformer neural network model is used to process the fused feature images to obtain the motor state results.
2. The method according to claim 1, characterized in that, The FBG sensing network, which is fixedly installed inside the motor, obtains high-frequency center wavelength changes related to strain and magnetic field, and low-frequency center wavelength changes related to temperature, including: Multiple bare fiber Bragg gratings are circumferentially pasted on the inner surface of the motor housing near the stator silicon steel sheet at one end of the motor, and multiple nickel-plated FBGs are installed and fixed at the stator winding of the motor. Based on the bare fiber grating, a first center wavelength change is obtained, and based on the nickel-plated FBG, a second center wavelength change is obtained. The first center wavelength change was filtered and corrected by using the least squares method combined with a Butterworth high-pass filter to obtain the high-frequency center wavelength change related to strain. The second center wavelength change was filtered and corrected by using the least squares method combined with a Butterworth high-pass filter to obtain the high-frequency center wavelength change related to the magnetic field. Adaptive mode decomposition is used to extract the trend term from the first and second center wavelength changes to obtain the temperature-related low-frequency center wavelength changes.
3. The method according to claim 2, characterized in that, The process of obtaining the first center wavelength change based on the bare fiber grating includes: Based on the temperature change at the location of the bare grating inside the motor And the circumferential strain change on the inner surface of the outer casing near the stator silicon steel sheet at one end of the motor. The change in the first center wavelength caused by thermal expansion and strain effects The expression is: ； in: For temperature coefficient, The strain coefficient is... , The effective elastic coefficient of the fiber grating. This is the initial value of the first center wavelength.
4. The method according to claim 2, characterized in that, The second center wavelength change is obtained by coating the FBG with the nickel plating: Based on the temperature change of the end winding and the change in magnetic flux density The change in the second center wavelength caused by thermal expansion and magnetostriction effects The expression is: ； in: It is the temperature coefficient. It is the magnetic flux density coefficient. This is the initial value for the second center wavelength.
5. The method according to claim 1, characterized in that, The high-frequency center wavelength changes related to strain and magnetic field, and the low-frequency center wavelength changes related to temperature are feature extracted and visualized using MTF and trend terms, respectively, to obtain time-domain feature images related to strain and magnetic field, and time-domain feature images related to temperature, including: Two-dimensional images of the strain-related high-frequency center wavelength changes are generated using MTF to obtain strain-related time-domain feature images; Two-dimensional images of the high-frequency center wavelength changes related to the magnetic field are generated using MTF to obtain time-domain feature images related to the magnetic field. A time-domain image of the temperature-related low-frequency center wavelength variation was generated using trend term visualization to obtain a temperature-related time-domain feature image.
6. The method according to claim 1, characterized in that, The process of fusing multiple features from feature images related to strain, magnetic field, and temperature to obtain a fused feature image includes: Multiple frames of time-domain feature images related to strain, magnetic field, and temperature are stacked by channel to obtain a fused feature image.
7. The method according to claim 1, characterized in that, The method for establishing the Transformer neural network model is as follows: The Transformer neural network model is trained by fusing a set of feature images, which includes multi-feature fusion images related to strain, magnetic field, and temperature under four states: normal motor operation, static air gap eccentricity, dynamic air gap eccentricity, and insulation layer damage.
8. The method according to claim 1, characterized in that: The bare fiber Bragg grating includes a first bare fiber Bragg grating, a second bare fiber Bragg grating, a third bare fiber Bragg grating, and a fourth bare fiber Bragg grating. The first, second, third, and fourth bare fiber Bragg gratings are circumferentially attached to the inner surface of the motor housing near the stator silicon steel sheet at one end using adhesive. The first, second, third, and fourth bare fiber Bragg gratings are evenly distributed circumferentially on the inner surface of the motor housing, with each grating spaced 90 degrees apart. The first, second, third, and fourth bare fiber Bragg gratings are all connected end-to-end in series on the same optical fiber, with the first bare fiber Bragg grating at the beginning and the fourth bare fiber Bragg grating at the end. The nickel-plated FBG includes a first nickel-plated FBG, a second nickel-plated FBG, and a third nickel-plated FBG. One end of each of the first, second, and third nickel-plated FBGs is fixed to the gap in the end winding of the motor stator using adhesive. The first, second, and third nickel-plated FBGs are evenly distributed circumferentially around the end winding of the motor stator, and are spaced 120 degrees apart from each other.
9. A motor condition monitoring system based on FBG, characterized in that, include: Bare fiber Bragg gratings, multiple bare fiber Bragg gratings are circumferentially attached to the inner surface of the housing near the stator silicon steel sheet at one end of the motor using adhesive at both ends; A nickel-plated FBG is coated, and multiple nickel-plated FBGs are installed and fixed at the end windings of the motor stator. The high-speed demodulator is connected to the bare fiber grating and the nickel-plated FBG via optical fiber, and is used to acquire the center wavelength of the FBG sensor network in real time. The host computer is used to perform signal preprocessing, feature extraction, and multi-feature fusion on the center wavelength data of the FBG sensor network to obtain a multi-feature fused feature image related to strain, magnetic field, and temperature. The fused feature image is then processed by a pre-established Transformer neural network to obtain the motor state result. A display is used to show the bare fiber Bragg grating, the nickel-plated FBG, and the motor status results.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, The computer program includes executable instructions that, when executed by a processor, implement the method described in any one of claims 1-8.