Method, device and system for detecting demagnetization of permanent magnet material

Abstract

The application relates to a method, device and system for detecting demagnetization of a permanent magnet material, wherein the method comprises: obtaining a force change signal of the target permanent magnet material from a device for detecting demagnetization of the permanent magnet material; and determining a demagnetization state of the target permanent magnet material according to the force change signal. Compared with a method for detecting demagnetization of a permanent magnet material by using a Hall probe to detect the magnetic field strength on the surface of the permanent magnet material, the method in the embodiment is not affected by manual operation. Compared with a method for evaluating the health state of a permanent magnet material by using a camera to capture cracks, depressions and the like on the surface of the material, since the relationship between the force change signal and the demagnetization degree is relatively clear, different demagnetization degrees correspond to different force change signals, and therefore the method has higher detection accuracy and can determine the specific demagnetization degree. In summary, the method for detecting demagnetization of a permanent magnet material in the embodiment solves the problem that the detection accuracy of the current method for detecting demagnetization of a permanent magnet material is low.

Description

Technical Field

[0001] This application relates to the field of permanent magnet material testing, and in particular to a method, apparatus and system for detecting demagnetization of permanent magnet materials. Background Technology

[0002] Permanent magnet materials, with their excellent high remanence and coercivity, combined with their outstanding resistance to demagnetization and wide temperature range adaptability, exhibit superior magnetic performance retention characteristics under complex working conditions. This material advantage not only effectively overcomes the interference of ambient temperature fluctuations on magnetic properties but also maintains stable magnetic flux output under long-term alternating magnetic fields. They have become core materials in high-tech and advanced manufacturing fields, particularly in robotics, new energy vehicles, aerospace, intelligent sensing, and communication base stations, playing an indispensable role in applications such as magnetic encoders, permanent magnet synchronous motors, satellite attitude control components, magnetic sensors, and filters. However, due to the inherent aging properties of permanent magnet materials and the complexity of the actual operating conditions of the equipment they are used in, as well as the high-temperature and humid working environment, irreversible demagnetization of permanent magnet materials is prone to occur. After demagnetization, it may lead to decreased performance of related equipment, energy waste, production downtime, increased maintenance costs, and potential safety hazards. These problems and losses have a serious impact on enterprise operations and production efficiency. Therefore, laying the foundation for regular maintenance and fault-tolerant compensation of permanent magnet materials to achieve high-performance and stable operation of equipment is of great significance to promoting the development of the manufacturing industry.

[0003] Currently, there are two main methods for detecting demagnetization of permanent magnet materials. One method involves using a Hall effect probe to detect the magnetic field strength on the surface of the permanent magnet material. This method is affected by factors such as the thickness of the Hall effect probe element and the skill level of the user, leading to significant variations in measurement data and consequently low accuracy and efficiency. The second method involves using a camera to photograph cracks and dents on the material surface to assess the health status of the permanent magnet material. However, this method relies heavily on surface texture information, resulting in significant randomness, large errors, and an inability to determine the specific degree of demagnetization.

[0004] There is currently no effective solution to the problem of low detection accuracy in existing methods for detecting demagnetization of permanent magnet materials. Summary of the Invention

[0005] This invention provides a method, apparatus, and system for detecting demagnetization of permanent magnet materials, in order to solve the problem of low detection accuracy in current methods for detecting demagnetization of permanent magnet materials.

[0006] In a first aspect, the present invention provides a method for detecting demagnetization of permanent magnet materials, comprising:

[0007] Obtain force change signals about the target permanent magnet material from equipment for detecting demagnetization of permanent magnet materials;

[0008] The demagnetization state of the target permanent magnet material is determined based on the force change signal.

[0009] The device for detecting demagnetization of permanent magnet materials includes a transmission device, a fixing frame, and a magnetic signal acquisition device. The transmission device is used to move the target permanent magnet material. The fixing frame is fixedly installed directly above the transmission path of the transmission device. The magnetic signal acquisition device includes a rectangular permanent magnet material, a first thin-film pressure sensor, and a second thin-film pressure sensor. The first thin-film pressure sensor and the second thin-film pressure sensor are respectively attached to the upper and lower surfaces of the rectangular permanent magnet material and their vertical projections overlap. The fixing frame fixes the magnetic signal acquisition device by contacting the upper surface of the first thin-film pressure sensor and the lower surface of the second thin-film pressure sensor.

[0010] During the process of the transmission device moving the target permanent magnet material, the first thin-film pressure sensor and the second thin-film pressure sensor provide the force change signal.

[0011] Secondly, this invention provides an apparatus for detecting demagnetization of permanent magnet materials, comprising:

[0012] The signal acquisition module is used to acquire force change signals about the target permanent magnet material from the equipment for detecting the demagnetization of permanent magnet materials;

[0013] The demagnetization detection module is used to determine the demagnetization state of the target permanent magnet material based on the force change signal.

[0014] The device for detecting demagnetization of permanent magnet materials includes a transmission device, a fixing frame, and a magnetic signal acquisition device. The transmission device is used to move the target permanent magnet material. The fixing frame is fixedly installed directly above the transmission path of the transmission device. The magnetic signal acquisition device includes a rectangular permanent magnet material, a first thin-film pressure sensor, and a second thin-film pressure sensor. The first thin-film pressure sensor and the second thin-film pressure sensor are respectively attached to the upper and lower surfaces of the rectangular permanent magnet material and their vertical projections overlap. The fixing frame fixes the magnetic signal acquisition device by contacting the upper surface of the first thin-film pressure sensor and the lower surface of the second thin-film pressure sensor.

[0015] During the process of the transmission device moving the target permanent magnet material, the first thin-film pressure sensor and the second thin-film pressure sensor provide the force change signal.

[0016] Thirdly, the present invention provides a computer including a memory and a processor, the memory storing a computer program, the processor being configured to run the computer program to perform the method for detecting demagnetization of permanent magnet materials as described in the first aspect.

[0017] Fourthly, this invention provides a system for detecting demagnetization of permanent magnet materials, comprising:

[0018] Equipment for detecting demagnetization of permanent magnet materials and the computer described in the first aspect;

[0019] The device for detecting demagnetization of permanent magnet materials includes a transmission device, a fixing frame, and a magnetic signal acquisition device. The transmission device is used to move the target permanent magnet material. The fixing frame is fixedly installed directly above the transmission path of the transmission device. The magnetic signal acquisition device includes a rectangular permanent magnet material, a first thin-film pressure sensor, and a second thin-film pressure sensor. The first thin-film pressure sensor and the second thin-film pressure sensor are respectively attached to the upper and lower surfaces of the rectangular permanent magnet material and their vertical projections overlap. The fixing frame fixes the magnetic signal acquisition device by contacting the upper surface of the first thin-film pressure sensor and the lower surface of the second thin-film pressure sensor.

[0020] During the process of the transmission device moving the target permanent magnet material, the first thin-film pressure sensor and the second thin-film pressure sensor provide force change signals to the computer.

[0021] Compared to related technologies, this method for detecting demagnetization of permanent magnet materials first acquires force change signals of the target permanent magnet material using a demagnetization detection device, and then determines the demagnetization state of the target permanent magnet material based on these force change signals. Compared to detecting demagnetization of permanent magnet materials by using a Hall effect probe to detect the magnetic field strength on the surface of the permanent magnet material, the method in this embodiment is not affected by manual operation. Compared to assessing the health status of permanent magnet materials by using a camera to photograph cracks, dents, etc., on the material surface, the relationship between the force change signal and the degree of demagnetization is relatively clear; different degrees of demagnetization correspond to different force change signals. Therefore, the method in this embodiment has higher detection accuracy and can determine the specific degree of demagnetization. In summary, the method for detecting demagnetization of permanent magnet materials in this embodiment solves the problem of low detection accuracy in current methods for detecting demagnetization of permanent magnet materials.

[0022] Details of one or more embodiments of this application are set forth in the following drawings and description to make other features, objects and advantages of this application more readily apparent. Attached Figure Description

[0023] Figure 1This is a flowchart of the method for detecting demagnetization of permanent magnet materials provided in this embodiment;

[0024] Figure 2 This is a structural diagram of the permanent magnet material demagnetization detection system provided in this embodiment;

[0025] Figure 3 This is a schematic diagram showing the cooperation between the magnetic signal acquisition device and the mounting bracket provided in this embodiment;

[0026] Figure 4 This is a force analysis diagram of the rectangular magnet material and the target magnet material repelling each other in this embodiment;

[0027] Figure 5 This is a force change signal diagram when the rectangular magnet material and the target magnet material repel each other in this embodiment;

[0028] Figure 6 This is a force analysis diagram of the rectangular magnet material and the target magnet material attracting each other in this embodiment;

[0029] Figure 7 This is a force change signal diagram when the rectangular magnet material and the target magnet material attract each other in this embodiment;

[0030] Figure 8 This is a force variation signal diagram of the rectangular magnet material repelling the target magnet material with different degrees of demagnetization in this embodiment;

[0031] Figure 9 This is a force variation signal diagram when a rectangular magnet material is attracted to a target magnet material with different degrees of demagnetization in this embodiment;

[0032] Figure 10 This is a schematic diagram illustrating the transformation of the force change signal from Cartesian coordinates to polar coordinates in this embodiment;

[0033] Figure 11 This is a detailed comparison of rainbow diagrams derived from the force change signals of the target permanent magnet material under five different demagnetization states in this embodiment;

[0034] Figure 12 This is an architecture diagram of the DMA-RpViT demagnetization detection model provided in this embodiment;

[0035] Figure 13 This is an architecture diagram of the downsampling layer in the DMA-RpViT demagnetization detection model provided in this embodiment;

[0036] Figure 14 This is an architecture diagram of the feature extraction stage in the DMA-RpViT demagnetization detection model provided in this embodiment;

[0037] Figure 15This is an architecture diagram of the DMA module in the DMA-RpViT demagnetization detection model provided in this embodiment. Detailed Implementation

[0038] To better understand the purpose, technical solution, and advantages of this application, the application is described and illustrated below in conjunction with the accompanying drawings and embodiments.

[0039] Unless otherwise defined, the technical or scientific terms used in this application shall have the general meaning as understood by one of ordinary skill in the art to which this application pertains. Words such as “a,” “an,” “an,” “the,” “the,” and “these,” used in this application, do not indicate quantitative limitation and may be singular or plural. The terms “comprising,” “including,” “having,” and any variations thereof used in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that comprises a series of steps or modules (units) is not limited to the listed steps or modules (units) but may include steps or modules (units) not listed, or may include other steps or modules (units) inherent to such processes, methods, products, or devices. The terms “connected,” “linked,” and “coupled,” used in this application, are not limited to physical or mechanical connections but may include electrical connections, whether direct or indirect. The term “multiple” used in this application refers to two or more. The "and / or" operator describes the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: A alone, A and B simultaneously, and B alone. Typically, the character " / " indicates that the objects before and after it are in an "or" relationship. The terms "first," "second," and "third," etc., used in this application are merely for distinguishing similar objects and do not represent a specific ordering of the objects.

[0040] This embodiment provides a method for detecting demagnetization of permanent magnet materials, which is executed by a computer. Figure 1 This is a flowchart of the method for detecting demagnetization of permanent magnet materials provided in this embodiment, as follows: Figure 1 As shown, the process includes steps S110 and S120.

[0041] Step S110: Obtain the force change signal of the target permanent magnet material from the device for detecting demagnetization of permanent magnet material provided in this embodiment.

[0042] S120 determines the demagnetization state of the target permanent magnet material based on the force change signal.

[0043] Reference Figure 2The device for detecting demagnetization of permanent magnet materials provided in this embodiment includes a transmission device 10, a fixing frame 20, and a magnetic signal acquisition device 30; the transmission device 10 is used to move the target permanent magnet material 00; the fixing frame 20 is fixedly installed directly above the transmission path of the transmission device 10. (Refer to...) Figure 3 The magnetic signal acquisition device 30 includes a rectangular permanent magnet material 33, a first thin-film pressure sensor 31, and a second thin-film pressure sensor 32. The first thin-film pressure sensor 31 and the second thin-film pressure sensor 32 are respectively attached to the upper and lower surfaces of the rectangular permanent magnet material 33 and their vertical projections overlap. The fixing frame 20 fixes the magnetic signal acquisition device 30 by contacting the upper surface of the first thin-film pressure sensor 31 and the lower surface of the second thin-film pressure sensor 32. During the process of the transmission device 10 moving the target permanent magnet material 00, the first thin-film pressure sensor 31 and the second thin-film pressure sensor 32 provide force change signals.

[0044] For example, the permanent magnet material can be a neodymium iron boron permanent magnet. The target permanent magnet material is the permanent magnet to be tested. The thin-film pressure sensor is a one-dimensional single-point pressure sensor.

[0045] Continue to refer to Figure 2 In this embodiment, the transmission device 10 is a transmission belt 11, which forms a transmission path as it moves. To ensure that the transmission belt 11 can stably move the target permanent magnet material 00, i.e., to prevent relative sliding between the transmission belt 11 and the target permanent magnet material 00, a clamp can be fixedly installed on the surface of the transmission belt 11 to hold the target permanent magnet material 00. To ensure that the detection results are not affected by the equipment for detecting the demagnetization of permanent magnet materials, the transmission belt 11, the fixing frame 20, and the clamp should be non-magnetic. For example, the fixing frame 20 and the clamp are made of plastic, and the transmission belt 11 is made of rubber. Simultaneously, the fixing frame 20 and the clamp can be telescopic, used to hold rectangular permanent magnet materials 33 and the target permanent magnet material 00 of different sizes, respectively. Preferably, the fixing frame 20 is fixedly installed directly above the center of the transmission path of the transmission device 10.

[0046] The principle behind demagnetization detection using the aforementioned device is as follows: As the target permanent magnet material 00 approaches, passes by, and moves away from directly beneath the rectangular permanent magnet material 33, the force between them changes. This causes a change in the force exerted by the rectangular permanent magnet material 33 on the two thin-film pressure sensors. Therefore, the force change signals from the two thin-film pressure sensors reflect the change in the force between the rectangular permanent magnet material 33 and the target permanent magnet material 00. For a specific specification of the target permanent magnet material 00, when it is in an undemagnetized state, the force change signals from the two thin-film pressure sensors can be measured. When the target permanent magnet material 00 demagnetizes, the force between it and the rectangular permanent magnet material 33 decreases, resulting in different force change signals from the two thin-film pressure sensors. Therefore, based on the force change signals of the target permanent magnet material 00, it is possible to determine whether demagnetization has occurred and the degree of demagnetization. The rectangular permanent magnet material 33 can be a healthy, undemagnetized permanent magnet material.

[0047] Furthermore, the method for detecting demagnetization of permanent magnet materials provided in this embodiment can simultaneously detect multiple target permanent magnet materials 00. Specifically, multiple target permanent magnet materials 00 can be placed at equal intervals on the conveyor belt 11, at which point the magnetic signal acquisition device 30 can sequentially acquire force change signals related to each target permanent magnet material 00. It should be noted that, in order to ensure that the target permanent magnet materials 00 do not interfere with each other, a sufficiently large distance should be maintained between two adjacent target permanent magnet materials 00 so that they do not simultaneously exert force on the rectangular permanent magnet material 33. This distance needs to be determined according to the specific specifications and models of the target permanent magnet materials 00 and the rectangular permanent magnet material 33.

[0048] Accordingly, to determine the relative positions of each target permanent magnet material 00 in the transmission path, a photoelectric sensor 40 can be fixedly installed on one side of the transmission path. The photoelectric sensor 40 is fixed in position within the transmission device 10 and does not move within the transmission path; for example, the photoelectric sensor 40 can be installed on the side plate of the transmission belt 11. Since the relative positions of the photoelectric sensor 40 in the transmission path are known, when a target permanent magnet material 00 passes directly in front of the photoelectric sensor 40, it can be detected by the photoelectric sensor 40, thus determining the relative position of the target permanent magnet material 00 in the transmission path at that moment. Since the spacing between adjacent target permanent magnet materials 00 is known, for each target permanent magnet material 00, the relationship between the force change signal and the relative position of the target permanent magnet material 00 can be determined, i.e., a force signal change curve can be established with the relative position of the target permanent magnet material 00 on the horizontal axis and the magnitude of the force signal on the vertical axis.

[0049] Furthermore, in this embodiment, to ensure that the measurement results of the first thin-film pressure sensor 31 and the second thin-film pressure sensor 32 do not interfere with each other, they should be installed parallel to the healthy rectangular permanent magnet material 33, and the contact surfaces on the upper and lower surfaces of the rectangular permanent magnet material 33 should correspond to each other (i.e., vertical projection overlap). To ensure the accuracy of the measurement results, the target permanent magnet material 00 to be detected must pass directly below the healthy rectangular permanent magnet material 33, so that the direction of the instantaneous attraction and repulsion is perpendicular.

[0050] The following is an example of how to use a device for detecting the demagnetization of permanent magnet materials.

[0051] 1. Place the target permanent magnet material to be tested on the retractable plastic clamp on the conveyor belt. The retractable plastic clamp is suitable for target permanent magnet materials of different sizes. The target permanent magnet material to be tested moves at a constant speed with the conveyor belt and is located at the center line of the conveyor belt so that it can pass directly under the rectangular permanent magnet material.

[0052] 2. When the photoelectric sensor detects the target permanent magnet material, it feeds back the position signal to the computer. The computer begins reading data from the upper and lower thin-film pressure sensors and can obtain the distance change between the target permanent magnet material and the rectangular permanent magnet material. Finally, it calculates the pressure values ​​of the first and second thin-film pressure sensors as a function of the relative position of the target permanent magnet material, i.e., the force change signal. Specifically, when the target permanent magnet material and the rectangular permanent magnet material repel each other, the rectangular permanent magnet material experiences an upward repulsive force, the pressure value of the first thin-film pressure sensor increases, and the pressure value of the second thin-film pressure sensor decreases. When the target permanent magnet material and the rectangular permanent magnet material attract each other, the rectangular permanent magnet material experiences a downward attractive force, the pressure value of the first thin-film pressure sensor remains unchanged, and the pressure value of the second thin-film pressure sensor increases.

[0053] In this step, the attractive and repulsive forces between the two NdFeB permanent magnets can be approximated using a magnetic dipole model. A magnetic dipole is a system composed of two magnetic charges of equal magnitude, opposite direction, and close proximity. The general expression for the interaction force between two magnetic dipoles is:

[0054]

[0055] in, and These are the magnetic moments of two magnetic dipoles. It is a vector pointing from one magnetic dipole to another, and r is... The modulus, μ0, is the permeability of free space, with a precise value of 4π × 10⁻⁶. -7 T·m / A.

[0056] (1) When the force between the target permanent magnet material and the rectangular permanent magnet material is repulsive:

[0057] Referring to Figure 4 , F is the repulsive force between the two. Since a one-dimensional single-point pressure sensor is used in this embodiment, only the vertical component force F1 = Fsinθ is considered. G is the gravity existing in the rectangular permanent magnet material 33 itself. F 上 is the pressure measured by the first thin-film pressure sensor 31, and F 下 is the pressure measured by the second thin-film pressure sensor 32.

[0058] From Figure 5 it can be seen that when the target permanent magnet material (the permanent magnet to be measured) is at point A, F 下 = G, and F1 gradually increases from 0 N. When the target permanent magnet material gradually approaches the rectangular permanent magnet material along the conveyor belt, the repulsive force between the two gradually increases. When F1 < G, F 上 is 0 N. As the distance continues to decrease, F 下 = G - F1, and its change is a gradually decreasing curve. When the target permanent magnet material is at point B, F1 = G, and at this time F 上 is 0 N, and F 下 is also 0 N. As the distance gradually decreases, F1 gradually increases. When F1 > G, F 下 is 0 N. As the distance continues to decrease, F 上 = F1 - G, and its change is a gradually increasing curve. When the target permanent magnet material is at point C, when the target permanent magnet material is exactly below the rectangular permanent magnet material, the curve rises to the highest point. As the position continuously moves away from directly below the rectangular permanent magnet material, F1 gradually decreases, and F 上 = F1 - G, and its change is a gradually decreasing curve. When the target permanent magnet material is at point D, F1 = G, and at this time F 上 is 0 N, and F 下 is also 0 N. Subsequently, since F1 < G, F 上 is always 0 N. As the distance continuously increases, F [[ID=�7]] 下 = G - F1, and its change is a gradually increasing curve. Until the target permanent magnet material is at point E, F1 decreases to 0 N, and F 下 = G.

[0059] Let the period from F1 > 0 to F1 = 0 be a complete cycle. Since there is a situation where when F 上 = 0 N, F 下 is not 0. When F 下 = 0 N, F 上 is not 0. For the data collected by the two thin-film pressure sensors, by splicing, F 上 and F 下The change in distance forms a continuous curve, which serves as a signal of force change.

[0060] (2) When there is an attractive force between the target permanent magnet material and the rectangular permanent magnet material:

[0061] Reference Figure 6 F is the repulsive force between the two forces. Since this embodiment uses a one-dimensional single-point pressure sensor, only the vertical component F1 = Fsinθ is considered. G is the gravity inherent in the rectangular permanent magnet material 00 itself. 上 The pressure F is measured by the first thin-film pressure sensor 31. 下 It is the pressure measured by the second thin-film pressure sensor 32.

[0062] Depend on Figure 7 It can be seen that F 上 The value is 0N throughout. When the target permanent magnet material (the permanent magnet under test) reaches point A, F... 下 =G, F1 gradually increases from 0N. Then, F 下 =G+F1, as it gets closer to directly below the rectangular permanent magnet material, F 下 The change is a gradually rising curve. When the target permanent magnet material reaches point B, directly below the rectangular permanent magnet material, F... 下 The maximum value is obtained. Then F... 下 Gradually decrease until the target permanent magnet material is located at point C, F 下 =G.

[0063] Due to F 上 If it is always 0N, then only F needs to be considered. 下 A curve that varies with distance. Let the period from F1>0 to F1=0 be considered a complete cycle, and then define this segment F... 下 Data that varies with distance is used as a force change signal.

[0064] Figure 8 It is the curve showing the change in pressure values ​​of the upper and lower thin-film pressure sensors when the force between the target permanent magnet material and the fixed rectangular permanent magnet material is a repulsive force. Figure 9 This is a curve showing the change in pressure values ​​from two thin-film pressure sensors when the force between the target permanent magnet material and the fixed rectangular permanent magnet material is attractive. The horizontal axis represents the distance between the target permanent magnet material and the rectangular permanent magnet material directly below it. Both figures show the force change curves for the target permanent magnet material at its healthy state, demagnetization level of 20%, 40%, 60%, and 80%. As can be seen from the figures, the force change curves for the target permanent magnet material differ depending on whether there is attraction or repulsion, providing a basis for distinguishing different degrees of demagnetization.

[0065] 3. The computer receives the force change signal and determines the demagnetization state of the target permanent magnet material based on the force change signal.

[0066] In this embodiment, S120 specifically includes: converting the force change signal into a two-dimensional rainbow image through rainbow view conversion; inputting the two-dimensional rainbow image into the trained DMA-RpViT demagnetization detection model; and predicting the demagnetization state of the target permanent magnet material through the DMA-RpViT demagnetization detection model.

[0067] The force change signal of the target permanent magnet material is obtained and treated as a one-dimensional signal to be processed. Then, the one-dimensional signal is converted into a two-dimensional image using a rainbow view conversion signal processing method. The specific conversion process is as follows:

[0068] First, refer to Figure 10 The force change signal x of the target permanent magnet is transformed from Cartesian coordinates to polar coordinates, and the frequency and amplitude of the original signal are represented by a 2D polar symmetric pattern:

[0069]

[0070]

[0071] Where, x i It is the amplitude of x at the i-th sampling point, which is the F value corresponding to that point. 上 or F 下 The value of x, where τ is the time delay parameter, 1≤τ≤10, i+τ This represents the amplitude of the τ sampling points following the i-th sampling point; x max and x min Let x represent the maximum and minimum amplitudes of the force change signal x, respectively; v is the gain angle of each part of the image, v < θ. l θ l θ is the reference starting angle of the l-th mirror symmetry plane; r(i) represents the polar radius, and θ(i) and φ(i) represent the rotation angles of the mirror symmetry plane in the clockwise and counterclockwise directions, respectively.

[0072] θ l The calculation formula is as follows:

[0073]

[0074] In the formula, m = 0, 1, 2, ..., n-1; n is the number of mirror-symmetric planes, which is set to 6 in this embodiment. Therefore, the two parameters τ and v are very important and need to be set to appropriate values. This embodiment introduces a normalized cross-correlation method to select appropriate parameters. Through the above steps, the signal can be better visualized, as shown below:

[0075]

[0076] In the above formula, A and B are 2D matrices corresponding to two different images, with a size of m×n. A mn and B mn Represents the grayscale value at the corresponding position. and τ represents the average gray value, and r(A,B) represents the prior similarity, ranging from 0 to 1. A larger value for r indicates a higher similarity between images A and B. In this embodiment, by comparing rainbow images under different demagnetization states, the parameters τ and v, where r is minimized (i.e., the similarity is minimized), are selected as the optimal parameters.

[0077] In this embodiment, the signal processing method for rainbow view transformation incorporates different color information during image mapping to enhance local features. The detailed principle of the rainbow view transformation method is as follows:

[0078] 1. First, it can be based on The mapped coordinates (α(i), β(i)) of the mirror-symmetric plane in the clockwise or counterclockwise direction are calculated. The calculation method is as follows:

[0079]

[0080] In the formula, Round() represents the rounding function, and Ang represents θ(i) or The value of R is 100. When Ang represents θ(i), (α(i),β(i)) represents the mapped coordinates of the mirror-symmetric plane in the clockwise direction; conversely, it represents the mapped coordinates in the counterclockwise direction.

[0081] 2. Calculate the amplitude Amp(i) in the clockwise or counterclockwise direction based on the mapped coordinates (α(i), β(i)), as shown below:

[0082]

[0083] In the formula, (α(i),β(i)) represents the mapped coordinates of the mirror-symmetric plane in the clockwise direction.

[0084] 3. To visualize signal amplitude changes, a rainbow color mapping function is used. Specific color values ​​are determined based on the normalized amplitude values. The color mapping process is as follows:

[0085]

[0086] Where Rainbow() represents the color mapping function, and Amp max and Amp minThis refers to the maximum and minimum amplitude of the color mapping function. Rainbow color mapping follows its preset color gradation rules, assigning different colors to different input values. Therefore, a rainbow image can produce different color changes corresponding to variations in amplitude.

[0087] Reference Figure 11 The image shows a detailed comparison of rainbow patterns generated from the force change signals of two thin-film pressure sensors under five different demagnetization states (undemagnetized, 20% demagnetized, 40% demagnetized, 60% demagnetized, and 80% demagnetized) of the target permanent magnet material. The image clearly shows significant differences in the rainbow patterns corresponding to different demagnetization states, thus providing a basis for identifying different demagnetization states.

[0088] The above is a detailed explanation of the rainbow view conversion. The converted two-dimensional image is then input into the trained DMA-RpViT demagnetization detection model, which can then predict the demagnetization state of the target permanent magnet material.

[0089] The DMA-RpViT demagnetization detection model requires pre-training. Specifically, two-dimensional rainbow images derived from force change signals corresponding to the same demagnetization state are used as similar samples. A sample library of rainbow images corresponding to permanent magnet materials with demagnetization degrees ranging from 1% to 99% is used to train the DMA-RpViT demagnetization detection model.

[0090] It should be noted that, since permanent magnet materials with demagnetization faults are difficult to customize, permanent magnet materials in good condition can be heated to different temperatures using temperature-controlled ceramic plates to obtain permanent magnet materials with different degrees of demagnetization. The temperature of the permanent magnet materials is measured using an infrared thermometer, and the magnetic flux density is quantitatively analyzed using a gaussmeter to determine the degree of demagnetization. Then, the obtained permanent magnet materials with demagnetization degrees ranging from 1% to 100% are arranged with the N pole facing upwards and the S pole facing upwards, respectively. Using the permanent magnet material demagnetization detection device in this embodiment, a rich sample library of samples with different demagnetization degrees can be obtained, which is then used to train the DMA-RpViT demagnetization detection algorithm.

[0091] It should also be noted that converting the force change signal into a two-dimensional rainbow image as the DMA-RpViT demagnetization detection model is a preferred method. As mentioned earlier, the force change signals corresponding to permanent magnet materials with different degrees of demagnetization are also different. Therefore, in other embodiments, the force change signal can also be directly used as the input to the DMA-RpViT demagnetization detection model. In the corresponding embodiment, S120 specifically includes: inputting the force change signal into the trained DMA-RpViT demagnetization detection model, and predicting the demagnetization state of the target permanent magnet material through the DMA-RpViT demagnetization detection model. Accordingly, the training samples of the DMA-RpViT demagnetization detection model are the force change signals corresponding to permanent magnet materials with a demagnetization degree of 1-99%.

[0092] Reference Figure 12 In this embodiment, the DMA-RpViT demagnetization detection model includes a sequentially connected input layer, intermediate layers (composed of feature extraction stages one through four), and an output layer (composed of pooling layers and linear layers). Sequential connection means that the output of one network layer is the input of the next. Specifically, the input of the DMA-RpViT demagnetization detection model is the input of the input layer, the output of the input layer is the input of the intermediate layer, the output of the intermediate layer is the input of the output layer, and the output of the output layer is the output of the DMA-RpViT demagnetization detection model.

[0093] Reference Figure 13 The input layer performs convolution and pooling on the input of the DMA-RpViT demagnetization detection model. Specifically, it may include sequentially connected 3×3 convolutions, activation layers, and 3×3 convolutions. The intermediate layer extracts features from the output of the input layer, and the output layer pools and maps the output of the intermediate layer to obtain the demagnetization state of the target permanent magnet material. The intermediate layer consists of four sequentially connected feature extraction stages, connected by downsampling layers. Each feature extraction stage is composed of at least one SEVT module and at least one OEVT module alternately connected. Specifically, the downsampling layer may include sequentially connected OEVT modules, 3×3 depthwise convolutions, 1×1 convolutions, and a feedforward neural network (FNN) layer. In the first, second, and fourth feature extraction stages, there is one SEVT module and one OEVT module; in the third feature extraction stage, there are seven SEVT modules and seven OEVT modules.

[0094] Reference Figure 14 The SEVT module includes a first 3×3 depthwise convolution, a DMA module, and a first feedforward neural network connected in sequence, with input and output residual connections for the first 3×3 depthwise convolution. The OEVT module includes a second 3×3 depthwise convolution, an SE (squeeze-excite) module, and a second feedforward neural network connected in sequence, with input and output residual connections for the second 3×3 depthwise convolution.

[0095] The overall data flow of the DMA-RpViT demagnetization detection model is as follows:

[0096] The input layer to the DMA-RpViT demagnetization detection model is X. m (In this embodiment, a two-dimensional rainbow image is used.) Convolution and pooling operations are performed to extract basic features and obtain the intermediate feature map Z1.

[0097] The first feature extraction stage performs a convolution operation on Z1 to further extract features and obtain the intermediate feature map Z2.

[0098] The first downsampling layer downsamples Z2 to reduce the size of the feature map while increasing the number of channels to obtain the intermediate feature map Z3.

[0099] The second feature extraction stage performs a convolution operation on Z3 to further extract features, resulting in the intermediate feature map Z4.

[0100] The second downsampling layer downsamples Z4 to reduce the size of the feature map while increasing the number of channels to obtain the intermediate feature map Z5.

[0101] The third feature extraction stage performs a convolution operation on Z5 to further extract features, resulting in the intermediate feature map Z6.

[0102] The third downsampling layer downsamples Z6 to reduce the size of the feature map while increasing the number of channels to obtain the intermediate feature map Z7.

[0103] In the fourth feature extraction stage, convolution operation is performed on Z7 to further extract features and obtain the intermediate feature map Z8.

[0104] The pooling layer in the output layer performs a global pooling operation on the feature map using Z8, converting the feature map into a fixed-length vector Z9.

[0105] The linear layer in the output layer further maps Z9 to the final regression target value, outputting a continuous value Z10 to represent the regression prediction result, i.e. the degree of demagnetization of the target permanent magnet material.

[0106] Among them, Z1 to Z2 have a specification of (H / 4, W / 4, 48); Z3 to Z4 have a specification of (H / 8, W / 8, 96); Z5 to Z6 have a specification of (H / 16, W / 16, 192); Z7 to Z8 have a specification of (H / 32, W / 32, 384); Z9 is a feature vector with a dimension of 1152; and Z10 is a vector with a dimension of 1.

[0107] Furthermore, this embodiment provides a new DMA module. (Refer to...) Figure 15The feature processing steps of the DMA module include: dividing the input feature maps of the DMA module into three groups; performing a 1×1 convolution on the first group of input feature maps to obtain the first feature map; performing average pooling on the first feature map along the X and Y directions respectively; concatenating the two average pooling results along the Z direction and performing a 1×1 convolution to obtain the second feature map; weighting the three groups of input feature maps based on the activation results of the second feature map; performing a 3×3 depthwise separable convolution on the second group of input feature maps to obtain the third feature map; performing a 5×5 dilated convolution on the second group of input feature maps to obtain the fourth feature map; and weighting and summing the third feature map... The activation results of the first and fourth feature maps are used to obtain the fifth feature map. Group normalization and average pooling are then performed on the three weighted input feature maps sequentially to obtain the sixth feature map. Average pooling is then performed on the fifth feature map to obtain the seventh feature map. The activation result of the sixth feature map is then multiplied by the fifth feature map to obtain the eighth feature map. The activation result of the seventh feature map is then multiplied by the group normalization results of the three input feature maps to obtain the ninth feature map. The activation result of the sum of the eighth and ninth feature maps is then used to weight the three input feature maps to obtain the output feature map of the DMA module.

[0108] The following example illustrates in detail the feature processing steps of the DMA module described above.

[0109] 1. Dynamic Multi-Scale Fusion: The input feature map of size C×H×W is divided into G groups, each with shape C / G×H×W. Different operations are performed on each group of feature maps.

[0110] (1) For one set of input feature maps, local features are captured by a 1×1 regular convolution, outputting a feature map of size C / G×H×W (first feature map). Then, average pooling (XAvg Pool) is performed along the X direction to obtain a feature map of shape C / G×1×W, and average pooling (YAvg Pool) is performed along the Y direction to obtain a feature map of shape C / G×H×1. The results of pooling along the X and Y directions are then concatenated along the image height direction and convolved by Conv(1×1) to obtain a feature map of shape C / G×1×1 (second feature map). Finally, two non-linear sigmoid functions are used to fit the 2D binary distribution of the linear convolution, and each set of original input feature maps is re-weighted to achieve cross-channel feature interaction.

[0111] The global average pooling formulas along the horizontal and vertical directions are as follows:

[0112]

[0113] In the formula, x c(H,i) refers to the input feature at the c-th channel with height h and width i. c (j,W) refers to the input feature at the c-th channel with height j and width w.

[0114] 2. Perform a 3×3 depthwise separable convolution on one set of input feature maps to capture deep-level features, outputting a feature map of size C / G×H×W (the third feature map). Then, perform a 5×5 dilated convolution on the other set of input feature maps to expand the receptive field, thereby obtaining more contextual information, outputting a feature map of size C / G×H×W (the fourth feature map). Normalize the weights using the Softmax activation function, and then fuse the feature maps through weighted summation to obtain the fifth feature map, thus achieving feature fusion at different scales.

[0115] Among them, deep features F are extracted. local The calculation formula is:

[0116] F local =DWConv 3×3 (X)

[0117] In the formula, DWConv 3×3 It is a depthwise separable convolution, which is decomposed into a 3×3 channel-wise convolution and a 1×1 point-wise convolution.

[0118] Among them, extracting contextual features F mid The calculation formula is:

[0119] F mid =DilatedConv 5×5 (X)

[0120] In the formula, DilatedConv 5×5 It is a dilated convolution with a dilation rate of 2 and an equivalent receptive field of 9×9.

[0121] 2. Cross-spatial learning component: This component consists of two parallel branches. Each reweighted set of input feature maps undergoes group normalization (GroupNorm). Then, both branches perform global average pooling (Avg Pool) to obtain feature maps of shape C / G×1×1 (the sixth and seventh feature maps, respectively). Next, both branches use the natural nonlinear function Softmax of a 2D Gaussian mapping to fit linear transformations (obtaining the activation results for the sixth and seventh feature maps, respectively). Finally, cross-spatial learning is achieved by multiplying the output of the previous step with matrix dot products (i.e., performing a matrix dot product between the activation result of the sixth feature map and the fifth feature map to obtain the eighth feature map, and performing a matrix dot product between the activation result of the seventh feature map and the group normalization results of the three sets of input feature maps to obtain the ninth feature map).

[0122] The formula for global average pooling is:

[0123]

[0124] In the formula, x c (i,j) refers to the input feature at the c-th channel with height i and width j.

[0125] 3. The learning results of the two parallel branches across space are added together (the eighth and ninth feature maps are added together), and then sigmoid activation and re-weighting are performed. The final output feature map has a shape of C×H×W.

[0126] As described in the above embodiments, the method for detecting demagnetization of permanent magnet materials first acquires force change signals of the target permanent magnet material using a device for detecting demagnetization of permanent magnet materials, and then determines the demagnetization state of the target permanent magnet material based on the force change signals. Compared to detecting demagnetization of permanent magnet materials by using a Hall probe to detect the magnetic field strength on the surface of the permanent magnet material, the method in this embodiment is not affected by manual operation. Compared to assessing the health status of permanent magnet materials by using a camera to photograph cracks, dents, etc. on the material surface, the method in this embodiment has higher detection accuracy and can determine the specific degree of demagnetization because the relationship between the force change signal and the degree of demagnetization is relatively clear, with different degrees of demagnetization corresponding to different force change signals. In summary, the method for detecting demagnetization of permanent magnet materials in this embodiment solves the problem of low detection accuracy in current methods for detecting demagnetization of permanent magnet materials.

[0127] Furthermore, the method for detecting demagnetization of permanent magnet materials employs a novel DMA-RpViT demagnetization detection model to identify force change signals. Compared to conventional deep learning models, this novel DMA-RpViT demagnetization detection model is more suitable for identifying force change signals and has higher recognition accuracy, thereby further improving the accuracy of demagnetization detection of permanent magnet materials.

[0128] This embodiment also provides a device for detecting the demagnetization of permanent magnet materials. This device is used to implement the method for detecting the demagnetization of permanent magnet materials provided in this embodiment, and will not be repeated hereafter. The terms "module," "unit," "subunit," etc., used below refer to combinations of software and / or hardware that achieve a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0129] The device for detecting demagnetization of permanent magnet materials provided in this embodiment includes a signal acquisition module and a demagnetization detection module.

[0130] The signal acquisition module is used to acquire force change signals about the target permanent magnet material from the equipment for detecting the demagnetization of permanent magnet materials.

[0131] The demagnetization detection module is used to determine the demagnetization state of the target permanent magnet material based on force change signals.

[0132] The equipment for detecting demagnetization of permanent magnet materials includes a transmission device, a fixing frame, and a magnetic signal acquisition device. The transmission device is used to move the target permanent magnet material. The fixing frame is fixedly installed directly above the transmission path of the transmission device. The magnetic signal acquisition device includes a rectangular permanent magnet material, a first thin-film pressure sensor, and a second thin-film pressure sensor. The first and second thin-film pressure sensors are respectively attached to the upper and lower surfaces of the rectangular permanent magnet material and their vertical projections overlap. The fixing frame fixes the magnetic signal acquisition device by contacting the upper surface of the first thin-film pressure sensor and the lower surface of the second thin-film pressure sensor. During the process of the transmission device moving the target permanent magnet material, the first and second thin-film pressure sensors provide force change signals.

[0133] It should be noted that the above modules can be functional modules or program modules, and can be implemented through software or hardware. For modules implemented through hardware, the above modules can reside in the same processor; or the above modules can be located in different processors in any combination.

[0134] This embodiment also provides a computer, including a memory and a processor. The memory stores a computer program, and the processor is configured to run the computer program to execute the method for detecting demagnetization of permanent magnet materials provided in this embodiment.

[0135] This embodiment also provides a system for detecting the demagnetization of permanent magnet materials, which includes: a device for detecting the demagnetization of permanent magnet materials and a computer provided in this embodiment.

[0136] The equipment for detecting demagnetization of permanent magnet materials includes a transmission device, a fixing frame, and a magnetic signal acquisition device. The transmission device is used to move the target permanent magnet material. The fixing frame is fixedly installed directly above the transmission path of the transmission device. The magnetic signal acquisition device includes a rectangular permanent magnet material, a first thin-film pressure sensor, and a second thin-film pressure sensor. The first and second thin-film pressure sensors are respectively attached to the upper and lower surfaces of the rectangular permanent magnet material and their vertical projections overlap. The fixing frame fixes the magnetic signal acquisition device by contacting the upper surface of the first thin-film pressure sensor and the lower surface of the second thin-film pressure sensor. During the process of the transmission device moving the target permanent magnet material, the first and second thin-film pressure sensors provide force change signals to the computer.

[0137] It should be understood that the specific embodiments described herein are merely illustrative of the application and not intended to limit it. All other embodiments derived by those skilled in the art based on the embodiments provided in this application without inventive effort are within the scope of protection of this application.

[0138] Obviously, the accompanying drawings are merely some examples or embodiments of this application. Those skilled in the art can apply this application to other similar situations based on these drawings without any creative effort. Furthermore, it is understood that although the work done in this development process may be complex and lengthy, for those skilled in the art, certain design, manufacturing, or production modifications made based on the technical content disclosed in this application are merely conventional technical means and should not be considered as insufficient disclosure of this application.

Claims

1. A method for detecting demagnetization of a permanent magnet material, characterized by, The method comprises: obtaining a force change signal of a target permanent magnet material from a device for detecting demagnetization of a permanent magnet material; converting the force change signal into a two-dimensional rainbow map through rainbow view conversion; inputting the two-dimensional rainbow map into a trained DMA-RpViT demagnetization detection model to predict the demagnetization state of the target permanent magnet material through the DMA-RpViT demagnetization detection model; wherein the device for detecting demagnetization of a permanent magnet material comprises a transmission device, a fixed frame, and a magnetic force signal acquisition device; the transmission device is used to move the target permanent magnet material; the fixed frame is fixedly installed above the transmission path of the transmission device; the magnetic force signal acquisition device comprises a rectangular permanent magnet material, a first thin film pressure sensor, and a second thin film pressure sensor; the first thin film pressure sensor and the second thin film pressure sensor are respectively attached to the upper and lower surfaces of the rectangular permanent magnet material and vertically overlap; the fixed frame fixes the magnetic force signal acquisition device by contacting the upper surface of the first thin film pressure sensor and the lower surface of the second thin film pressure sensor; in the process of moving the target permanent magnet material by the transmission device, the first thin film pressure sensor and the second thin film pressure sensor provide the force change signal.

2. The method of claim 1, wherein, The parameters in the rainbow view conversion are determined by a normalized cross-correlation method.

3. The method of claim 1, wherein, The DMA-RpViT demagnetization detection model comprises sequentially connected input layers, intermediate layers, and output layers; the input layers are used to perform convolution and pooling on the input of the DMA-RpViT demagnetization detection model; the intermediate layers perform feature extraction on the output of the input layers; and the output layers perform pooling and mapping on the output of the intermediate layers to obtain the demagnetization state of the target permanent magnet material.

4. The method of claim 3, wherein the magnetic field is applied to the permanent magnet material in a direction that is substantially perpendicular to the direction of the magnetic field of the permanent magnet material. The intermediate layers are divided into four feature extraction stages connected in sequence, and each two adjacent feature extraction stages are connected through a down-sampling layer; each feature extraction stage is composed of at least one SEVT module and at least one OEVT module connected alternately; the SEVT module comprises sequentially connected first 3×3 deep convolution, DMA module, and first feedforward neural network, and the input and output of the first 3×3 deep convolution are connected in residual; the OEVT module comprises sequentially connected second 3×3 deep convolution, SE module, and second feedforward neural network, and the input and output of the second 3×3 deep convolution are connected in residual.

5. The method of claim 4, wherein the step of detecting the demagnetization of the permanent magnet material is performed by a method comprising: The feature processing steps of the DMA module comprise: dividing the input feature map of the DMA module into three groups; performing 1×1 convolution on the first group of input feature maps to obtain a first feature map, performing average pooling on the first feature map along the X direction and the Y direction respectively, concatenating the two average pooling results along the Z direction, and then performing 1×1 convolution to obtain a second feature map, and weighting the three groups of input feature maps based on the activation result of the second feature map; performing 3×3 deep separable convolution on the second group of input feature maps to obtain a third feature map, performing 5×5 atrous convolution on the second group of input feature maps to obtain a fourth feature map, and weighting and summing the activation results of the third feature map and the fourth feature map to obtain a fifth feature map; perform group normalization and average pooling on the three groups of weighted input feature maps in sequence to obtain a sixth feature map, and perform average pooling on the fifth feature map to obtain a seventh feature map; perform matrix point multiplication operation on the activation result of the sixth feature map and the fifth feature map to obtain an eighth feature map, and perform matrix point multiplication operation on the activation result of the seventh feature map and the group normalization result of the three groups of input feature maps to obtain a ninth feature map; weight the three groups of input feature maps through the activation result of the addition result of the eighth feature map and the ninth feature map to obtain the output feature map of the DMA module.

6. The method of claim 4, wherein the step of detecting the demagnetization of the permanent magnet material is performed by a method comprising: In the first, second and fourth feature extraction stages, the SEVT module and the OEVT module are each one; in the third feature extraction stage, the SEVT module and the OEVT module are each seven.

7. A device for detecting demagnetization of a permanent magnet material, characterized by Comprise: a signal acquisition module for acquiring a force change signal about a target permanent magnet material from a device for detecting demagnetization of permanent magnet materials; a demagnetization detection module for converting the force change signal into a two-dimensional rainbow map through rainbow view conversion and inputting the two-dimensional rainbow map into a trained DMA-RpViT demagnetization detection model, and predicting the demagnetization state of the target permanent magnet material through the DMA-RpViT demagnetization detection model; wherein the device for detecting demagnetization of permanent magnet materials comprises a transmission device, a fixed frame, and a magnetic force signal acquisition device; the transmission device is used to move the target permanent magnet material; the fixed frame is fixedly installed directly above the transmission path of the transmission device; the magnetic force signal acquisition device comprises a rectangular permanent magnet material, a first thin film pressure sensor and a second thin film pressure sensor, the first thin film pressure sensor and the second thin film pressure sensor are respectively attached to the upper and lower surfaces of the rectangular permanent magnet material and vertically overlap, and the fixed frame fixes the magnetic force signal acquisition device by contacting the upper surface of the first thin film pressure sensor and the lower surface of the second thin film pressure sensor; During the movement of the target permanent magnet material by the transmission device, the first thin film pressure sensor and the second thin film pressure sensor provide the force change signal.

8. A computer comprising a memory and a processor, characterized in that, The memory stores a computer program, and the processor is configured to run the computer program to execute the method for detecting demagnetization of permanent magnet materials according to any one of claims 1 to 6.

9. A system for detecting demagnetization of a permanent magnetic material, characterized by Comprise: a device for detecting demagnetization of permanent magnet materials and a computer according to claim 8; the device for detecting demagnetization of permanent magnet materials comprises a transmission device, a fixed frame, and a magnetic force signal acquisition device; The transmission device is used for moving the target permanent magnetic material; the fixing frame is fixedly installed above the transmission path of the transmission device; the magnetic force signal acquisition device comprises a rectangular permanent magnetic material, a first thin film pressure sensor and a second thin film pressure sensor, the first thin film pressure sensor and the second thin film pressure sensor are respectively attached to the upper and lower surfaces of the rectangular permanent magnetic material and vertically overlap, and the fixing frame fixes the magnetic force signal acquisition device by contacting the upper surface of the first thin film pressure sensor and the lower surface of the second thin film pressure sensor. In the process that the transmission device moves the target permanent magnetic material, the first thin film pressure sensor and the second thin film pressure sensor provide force change signals to the computer.

Images