Magnetic-thermal-acoustic combined detection method and device for operation state of electromagnetic pump
By combining Hall effect sensors, infrared thermal imaging, and acoustic sensor arrays with deep learning algorithms, the accuracy problem of electromagnetic pump operation status detection was solved, realizing intelligent operation and maintenance of electromagnetic pumps and improving detection accuracy and generalization.
Patent Information
- Application Number
- CN202511631006.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-10
- Publication Date
- 2026-03-03
AI Technical Summary
Existing technologies cannot comprehensively and accurately detect the operating status of electromagnetic pumps, especially winding coil faults, resulting in limitations in monitoring methods.
A Hall sensor array is used to obtain the magnetic field distribution, an infrared thermal imaging module is used to obtain the temperature field information, and an acoustic sensor array is used to obtain the sound field distribution. By combining background denoising algorithms and deep learning algorithms, information fusion and analysis are performed to build a knowledge base for the operating status of electromagnetic pumps.
It enables accurate assessment of the operating status and defect diagnosis of electromagnetic pumps, improves the level of operation and maintenance, provides non-contact non-destructive testing and background noise suppression, and improves the accuracy and generalization of testing.
Smart Images
Figure CN121594959A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of power equipment technology, specifically relating to a method and device for combined magnetothermal-acoustic detection of the operating status of an electromagnetic pump. Background Technology
[0002] An electromagnetic pump is a type of pump that operates based on the principle of electromagnetic induction. It uses electromagnetic force to propel or extract liquid metal, achieving transport and circulation. Electromagnetic pumps offer advantages such as energy saving, low pollution, and high controllability in the transfer of liquid metal. Based on the conductive properties of liquid metal, electromagnetic pumps utilize the interaction of mutually perpendicular magnetic and electric fields within the pump channel to generate a Lorentz force that drives the directional movement of the liquid metal. Therefore, electromagnetic pumps have no moving parts, eliminating wear, and the Lorentz force generated by electromagnetic coupling provides a stable wave peak for the directional movement of the liquid metal, resulting in lower costs. However, because the winding coils of electromagnetic pumps are prone to failure during long-term operation, real-time monitoring and detection of the pump's operating status become crucial.
[0003] Currently, methods for detecting the operating status of electromagnetic pumps mainly focus on temperature rise characteristics, vibration characteristics, or current, which are closely related to the overall design of the equipment. However, due to the complexity of electromagnetic pump operation, the unclear propagation patterns of temperature rise and vibration, and the fact that monitoring methods typically use a single detection technique, existing methods have certain limitations and cannot comprehensively and accurately determine the operating status of electromagnetic pumps. Summary of the Invention
[0004] The purpose of this invention is to provide a method and apparatus for combined magnetothermal-acoustic detection of the operating status of an electromagnetic pump, which can accurately assess the operating status of the electromagnetic pump and diagnose defects, thereby improving the operation and maintenance level of the electromagnetic pump.
[0005] The technical solution of the present invention is as follows: A method for combined magnetothermal-acoustic detection of the operating status of an electromagnetic pump, comprising the following steps:
[0006] S01: The spatial magnetic field distribution of the electromagnetic pump stator is obtained through a Hall sensor array, the temperature field information of the electromagnetic pump shell is obtained through an infrared thermal imaging module, and the spatial sound field distribution of the electromagnetic pump is obtained through an acoustic sensor array.
[0007] S02: Remove ambient heat sources and eddy current heat sources from the original temperature measurement image using a background denoising algorithm;
[0008] S03: By visually fusing the preprocessed multimodal detection information, high-dimensional characterization parameters of the electromagnetic pump's operating status are obtained;
[0009] S04: By visualizing and fusing data and combining deep learning algorithms, the operating status of the electromagnetic pump is detected and analyzed.
[0010] Step S01 includes installing Hall sensors at certain intervals along the axis of the electromagnetic pump to obtain information on changes in the electromagnetic pump's magnetic field; obtaining information on the original heating of the electromagnetic pump using an infrared thermal imaging lens for detecting the temperature rise characteristics of the electromagnetic pump, and adjusting the measurement angle according to the electromagnetic pump's installation position during the measurement process; and installing acoustic sensors at certain intervals along the axis of the electromagnetic pump to capture the sound field distribution.
[0011] Step S02 includes using an infrared image denoising algorithm to remove environmental heat sources and eddy current heat sources from the original temperature measurement image, thereby improving the quality of the infrared image and fully extracting the effective information in the image to achieve accurate measurement of heat information.
[0012] The specific algorithm in step S02 is as follows:
[0013]
[0014] In the formula, f in Given an infrared image with background noise as input, NN1() is the feature encoding network, L is the high-dimensional representation information of the encoded infrared image, and NN2() is the feature decoding network. out This is the output of a noise-removed infrared image.
[0015] Step S03 includes using a fusion algorithm for heterogeneous data, such as feature extraction and fusion, to achieve the visualization fusion of magnetic-thermal-acoustic detection information of electromagnetic pump operating status, obtain high-dimensional characterization parameters of electromagnetic pump operating status, and establish an electromagnetic pump operating status knowledge base.
[0016] The specific algorithm for step S03 is as follows:
[0017] M=λ1*PCA(M1)+λ2*PCA(M2)+λ3*PCA(M3) (2)
[0018] In the formula: PCA() is the principal component analysis dimensionality reduction operation, M1, M2 and M3 are the electromagnetic pump magnetic field, thermal field and acoustic field signals respectively, λ1, λ2 and λ3 are the weight coefficients of different detection variables, and the initial value is set to 1.
[0019] Step S04 includes setting three defects in the electromagnetic pump: inter-turn short circuit, loose bolts, and insulation degradation, and labeling the fault types. It also involves acquiring electromagnetic pump magnetic-thermal-acoustic data under multiple operating conditions under these defects and performing visualization and feature fusion based on S03. A dataset including fused feature vectors and fault labels is constructed. Based on the labeled electromagnetic pump fault dataset, a convolutional neural network deep learning neural network model is constructed for feature extraction. Finally, a fully connected neural network is used for state classification to achieve the detection and analysis of the electromagnetic pump's operating status.
[0020] A combined magneto-thermal-acoustic detection device for the operating status of an electromagnetic pump includes a Hall sensor array, an infrared thermal imaging module, an acoustic sensor array, a background denoising module, a visualization information fusion module, and an electromagnetic pump operating status detection and analysis module. The infrared thermal imaging module acquires temperature field information of the electromagnetic pump casing. The background denoising module removes environmental heat sources and eddy current heat sources from the original temperature measurement image using an infrared image denoising algorithm. The background denoising module is connected to the infrared thermal imaging module to remove noise signals and improve the acquired infrared temperature rise image. The visualization information fusion module is connected to the background denoising module, the Hall sensor array, and the acoustic sensor array. It performs visualization fusion on the temperature rise information, magnetic field information, and acoustic field information processed by the background denoising module using PCA dimensionality reduction and weighted averaging. The electromagnetic pump operating status detection and analysis module is connected to the visualization information fusion module and uses a multilayer perceptron to detect and identify the operating status of the electromagnetic pump.
[0021] The beneficial effects of this invention are as follows: The method of this invention can acquire electromagnetic pump operating status information more comprehensively. Through a deep learning neural network model, it realizes feature extraction, feature fusion, and status classification of different state quantities, achieving nonlinear mapping between internal defects of the electromagnetic pump and external detection information, thus improving the accuracy and generalization of electromagnetic pump status detection. The device of this invention can realize non-contact, non-destructive testing, background noise suppression, and feature fusion of the magnetic field, temperature field, and sound field of an operating electromagnetic pump. By integrating with the electromagnetic pump magnetic-thermal-acoustic detection method and model, it provides technical support for electromagnetic pump operating status maintenance, which is of great significance for improving the intelligent operation and maintenance level of electromagnetic pumps and ensuring the safe and stable operation of equipment. Attached Figure Description
[0022] Figure 1 A flowchart of a method for combined magnetothermal-acoustic detection of the operating status of an electromagnetic pump provided by the present invention;
[0023] Figure 2 This is a schematic diagram of a magnetothermal-acoustic combined detection device for the operating status of an electromagnetic pump provided by the present invention.
[0024] In the figure: 1 Hall sensor array, 2 infrared thermal imaging module, 3 acoustic sensor array, 4 background noise reduction module, 5 visualization information fusion module, 6 electromagnetic pump operation status detection and analysis module, 7 electromagnetic pump under test. Detailed Implementation
[0025] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0026] like Figure 1 As shown, a method for combined magnetothermal-acoustic detection of the operating status of an electromagnetic pump includes the following steps:
[0027] S01: The spatial magnetic field distribution of the electromagnetic pump stator is obtained through a Hall sensor array, the temperature field information of the electromagnetic pump shell is obtained through an infrared thermal imaging module, and the spatial sound field distribution of the electromagnetic pump is obtained through an acoustic sensor array.
[0028] Specifically, Hall sensors are installed at a certain interval about 70mm along the axis of the electromagnetic pump to obtain information on changes in the electromagnetic pump's magnetic field; an infrared thermal imaging lens is used to detect the temperature rise characteristics of the electromagnetic pump to obtain information on the original heating of the electromagnetic pump, and the measurement angle can be adjusted according to the installation position of the electromagnetic pump during the measurement process; acoustic sensors are installed at a certain interval between 10mm and 60mm along the axis of the electromagnetic pump to capture the sound field distribution.
[0029] S02: Remove ambient heat sources and eddy current heat sources from the original temperature measurement image using a background denoising algorithm;
[0030] Specifically, by using an infrared image denoising algorithm, the environmental heat source and eddy current heat source in the original temperature measurement image are removed, thereby improving the quality of the infrared image and fully extracting the effective information in the image, so as to achieve accurate measurement of heat information.
[0031] The specific formula is as follows:
[0032]
[0033] In the formula, f in Given an infrared image with background noise as input, NN1() is the feature encoding network, L is the high-dimensional representation information of the encoded infrared image, and NN2() is the feature decoding network. out This is the output of a noise-removed infrared image.
[0034] S03: By visually fusing the preprocessed multimodal detection information, high-dimensional characterization parameters of the electromagnetic pump's operating status are obtained;
[0035] Specifically, by using fusion algorithms for heterogeneous data such as feature extraction and fusion, the visualization fusion of magnetic-thermal-acoustic detection information of electromagnetic pump operation status is realized, high-dimensional characterization parameters of electromagnetic pump operation status are obtained, and a knowledge base of electromagnetic pump operation status is established.
[0036] The specific formula is as follows:
[0037] M=λ1*PCA(M1)+λ2*PCA(M2)+λ3*PCA(M3) (2)
[0038] In the formula: PCA() is the principal component analysis dimensionality reduction operation, M1, M2 and M3 are the electromagnetic pump magnetic field, thermal field and acoustic field signals respectively, λ1, λ2 and λ3 are the weight coefficients of different detection variables, and the initial value is set to 1.
[0039] S04: By visualizing and fusing data and combining deep learning algorithms, the operating status of the electromagnetic pump is detected and analyzed.
[0040] Specifically, the process involves setting up three defects in the electromagnetic pump: inter-turn short circuit, loose bolts, and insulation degradation, and labeling the fault types. Magnetism-thermal-acoustic data of the electromagnetic pump under multiple operating conditions under these defects are acquired and visualized using S03. This results in the construction of a dataset including fused feature vectors and fault labels. Based on the labeled electromagnetic pump fault dataset, a convolutional neural network deep learning neural network model is constructed for feature extraction. Finally, a fully connected neural network is used for state classification, enabling the detection and analysis of the electromagnetic pump's operating status.
[0041] like Figure 2 As shown, an electromagnetic pump operating status joint magnetothermal-acoustic detection device includes a Hall sensor array 1, an infrared thermal imaging module 2, an acoustic sensor array 3, a background denoising module 4, a visualization information fusion module 5, and an electromagnetic pump operating status detection and analysis module 6. The infrared thermal imaging module 2 collects temperature field information of the electromagnetic pump casing. The background denoising module 4 removes environmental heat sources and eddy current heat sources from the original temperature measurement image using an infrared image denoising algorithm. The background denoising module 4 is connected to the infrared thermal imaging module 2 to remove noise signals and improve the acquired infrared temperature rise image. The visualization information fusion module 5 is connected to the background denoising module 4, the Hall sensor array 1, and the acoustic sensor array 3. It performs visualization fusion on the temperature rise information, magnetic field information, and acoustic field information processed by the background denoising module 4 using PCA dimensionality reduction and weighted averaging. The electromagnetic pump operating status detection and analysis module 6 is connected to the visualization information fusion module 5 and uses a multilayer perceptron to detect and identify the electromagnetic pump operating status.
Claims
1. A method for combined magnetothermal-acoustic detection of the operating status of an electromagnetic pump, characterized in that, Includes the following steps: S01: The spatial magnetic field distribution of the electromagnetic pump stator is obtained through a Hall sensor array, the temperature field information of the electromagnetic pump shell is obtained through an infrared thermal imaging module, and the spatial sound field distribution of the electromagnetic pump is obtained through an acoustic sensor array. S02: Remove ambient heat sources and eddy current heat sources from the original temperature measurement image using a background denoising algorithm; S03: By visually fusing the preprocessed multimodal detection information, high-dimensional characterization parameters of the electromagnetic pump's operating status are obtained; S04: By visualizing and fusing data and combining deep learning algorithms, the operating status of the electromagnetic pump is detected and analyzed.
2. The method for combined magnetothermal-acoustic detection of the operating status of an electromagnetic pump as described in claim 1, characterized in that: Step S01 includes installing Hall sensors at certain intervals along the axis of the electromagnetic pump to obtain information on changes in the electromagnetic pump's magnetic field; obtaining information on the original heating of the electromagnetic pump using an infrared thermal imaging lens for detecting the temperature rise characteristics of the electromagnetic pump, and adjusting the measurement angle according to the electromagnetic pump's installation position during the measurement process; and installing acoustic sensors at certain intervals along the axis of the electromagnetic pump to capture the sound field distribution.
3. The method for combined magnetothermal-acoustic detection of the operating status of an electromagnetic pump as described in claim 1, characterized in that: Step S02 includes using an infrared image denoising algorithm to remove environmental heat sources and eddy current heat sources from the original temperature measurement image, thereby improving the quality of the infrared image and fully extracting the effective information in the image to achieve accurate measurement of heat information.
4. The method for combined magnetothermal-acoustic detection of the operating status of an electromagnetic pump as described in claim 3, characterized in that: The specific algorithm in step S02 is as follows: In the formula, f in Given an infrared image with background noise as input, NN1() is the feature encoding network, L is the high-dimensional representation information of the encoded infrared image, and NN2() is the feature decoding network. out This is the output of a noise-removed infrared image.
5. The method for combined magnetothermal-acoustic detection of the operating status of an electromagnetic pump as described in claim 1, characterized in that: Step S03 includes using a fusion algorithm for heterogeneous data, such as feature extraction and fusion, to achieve the visualization fusion of magnetic-thermal-acoustic detection information of electromagnetic pump operating status, obtain high-dimensional characterization parameters of electromagnetic pump operating status, and establish an electromagnetic pump operating status knowledge base.
6. The method for combined magnetothermal-acoustic detection of the operating status of an electromagnetic pump as described in claim 5, characterized in that: The specific algorithm for step S03 is as follows: M=λ1*PCA(M1)+λ2*PCA(M2)+λ3*PCA(M3) (2) In the formula: PCA() is the principal component analysis dimensionality reduction operation, M1, M2 and M3 are the electromagnetic pump magnetic field, thermal field and acoustic field signals respectively, λ1, λ2 and λ3 are the weight coefficients of different detection variables, and the initial value is set to 1.
7. The method for combined magnetothermal-acoustic detection of the operating status of an electromagnetic pump as described in claim 1, characterized in that: Step S04 includes setting three defects in the electromagnetic pump: inter-turn short circuit, loose bolts, and insulation deterioration, and labeling the fault types. It also involves acquiring electromagnetic pump magnetic-thermal-acoustic data under multiple operating conditions under these defects and performing visualization and feature fusion based on S03. A dataset including fused feature vectors and fault labels is constructed. Based on the labeled electromagnetic pump fault dataset, a convolutional neural network deep learning neural network model is constructed for feature extraction. Finally, a fully connected neural network is used for state classification to achieve the detection and analysis of the electromagnetic pump's operating status.
8. A combined magnetothermal-acoustic detection device for the operating status of an electromagnetic pump, characterized in that: The system includes a Hall sensor array, an infrared thermal imaging module, an acoustic sensor array, a background denoising module, a visualization information fusion module, and an electromagnetic pump operation status detection and analysis module. The infrared thermal imaging module acquires the temperature field information of the electromagnetic pump casing. The background denoising module uses an infrared image denoising algorithm to remove environmental heat sources and eddy current heat sources from the original temperature measurement image. Connected to the infrared thermal imaging module, the background denoising module removes noise signals and improves the acquired infrared temperature rise image. The visualization information fusion module is connected to the background denoising module, the Hall sensor array, and the acoustic sensor array. It performs visualization fusion on the temperature rise information, magnetic field information, and acoustic field information processed by the background denoising module using PCA dimensionality reduction and weighted averaging. The electromagnetic pump operation status detection and analysis module is connected to the visualization information fusion module and uses a multilayer perceptron to detect and identify the electromagnetic pump's operation status.