High-precision bolt looseness detection device and method based on TMR sensor space array

Abstract

The invention provides a high-precision bolt looseness detection device based on a TMR sensor space array, and belongs to the field of bolt looseness detection.The device comprises a first permanent magnet and TMR sensors, the multiple TMR sensors are evenly distributed in the axial direction and the outer circumferential direction of a bolt, each TMR sensor comprises a second permanent magnet and a TMR sensor chip, and the second permanent magnet is arranged on the first permanent magnet. When the first permanent magnet moves relative to the second permanent magnet, the magnetic field where the TMR sensor chip is located changes. The invention further provides a high-precision bolt looseness detection method. The plurality of TMR sensors jointly form a sensor array to detect the loosening condition of the nut in the horizontal direction and the vertical direction, and when the nut is loosened, the plurality of TMR sensors jointly form the sensor array to monitor the loosening condition of the nut, so that the nut loosening detection precision is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of bolt loosening detection, and particularly relates to a high-precision bolt loosening detection device and method based on a TMR sensor spatial array. BACKGROUND

[0002] Bolt connection is the main way of angle steel connection of a power transmission tower, but due to the complex field environment, long-term influence of wind load vibration, conductor dancing and temperature change and other factors, the bolt is prone to loosening, which reduces the tower stiffness and even leads to tower instability. At present, regular inspection and tightening of loose bolts every year is an important supervision work, but the labor cost is high and the labor intensity is large, and a new solution needs to be sought.

[0003] The adoption of a sensing node to effectively monitor bolt loosening becomes an important means to solve the above problems. Among them, (1) a contact type measurement scheme, such as placing a pressure, a grating sensing unit on a nut, although it can realize multi-point real-time monitoring, but during installation, the nut, bolt punching and surface welding need to be disassembled, which will damage the bolt structure, and the installation is time-consuming and laborious, and the sensor has large power consumption, a complex power supply system, large equipment volume and inconvenient installation; (2) the adoption of ultrasonic detection and other technical means has the advantage of high detection precision, but due to the high cost, the equipment cannot realize multi-point real-time monitoring, which leads to the inability to realize large-scale detection; (3) the bolt monitoring sensing device adopting computer vision and image processing technology adopts non-contact monitoring in the use process, but the installation is complex and is easily affected by light environment, and cannot be monitored all day round, and the supporting equipment is complex.

[0004] In recent years, with the increasing requirement for magnetic field detection, tunneling magnetoresistance (TMR) sensors are more and more widely used. The TMR sensor is based on the spin-dependent tunneling effect, and the magnetic tunnel junction is a sandwich structure of ferromagnetic layer / non-magnetic insulating layer / ferromagnetic layer. When saturated magnetization, the magnetization directions of the two ferromagnetic layers are parallel, but the coercivities of the two are different. When reverse magnetization, the magnetization vector of the ferromagnetic layer with low coercivity first flips, so that the magnetization direction changes from parallel to antiparallel. The tunneling probability is closely related to the magnetization direction, and when parallel, most of the spin sub-electrons and few of the spin sub-electrons enter the corresponding empty state, and the tunneling current is large; when antiparallel, the situation is opposite, and the tunneling current is small. The tunneling conductance changes with the change of the magnetization direction, and the conductance value in the parallel state is higher than that in the antiparallel state. The external magnetic field can adjust the magnetization direction, cause the tunneling resistance to change, and produce the TMR effect. Compared with the Hall effect or anisotropic magnetoresistance (AMR) sensor, the TMR sensor has a high tunneling structure magnetic resistance change rate, a high output resolution under a weak magnetic field change, is suitable for scenes where the structure micro displacement causes the magnetic field disturbance, and provides a technical basis for a small volume and low power consumption sensor module.

[0005] As a high-performance magnetic sensor, the TMR sensor has many significant advantages in magnetic field detection, such as high sensitivity, high signal-to-noise ratio, high reliability, small size, low power consumption, and low cost. Therefore, it is feasible to apply the TMR sensor to bolt loosening detection. For example, in the Chinese utility model patent "A Bolt Loosening Monitoring Device and System" with announcement number CN221706424U, the magnet and the sensor body are fixed to the end face of the screw and the nut, respectively. However, the following problems still exist when applying a single TMR sensor to a bolt detection device: (1) In the horizontal direction, only the specific rotation angle of the bolt can be detected. If loosening occurs at other angles, it may not be detected, and the detection accuracy still needs to be improved; (2) In the vertical direction, due to the limited detection range of a single TMR sensor, it is difficult to accurately detect the loosening of the bolt in the vertical direction. Summary of the Invention

[0006] The technical problem to be solved by this invention is: how to solve the problem of low detection accuracy when a single TMR sensor detects loose bolts.

[0007] The present invention solves the above-mentioned technical problems through the following technical solution: a high-precision bolt loosening detection device based on a TMR sensor spatial array. The device includes a first permanent magnet and a TMR sensor. Multiple TMR sensors are evenly arranged along the axial direction and the outer circumference of the bolt. The TMR sensor includes a second permanent magnet and a TMR sensor chip. When the first permanent magnet moves relative to the second permanent magnet, the magnetic field of the TMR sensor chip changes.

[0008] In this invention, multiple TMR sensors together form a sensor array to detect the loosening of the nut in the horizontal and vertical directions. When the nut becomes loose, the sensor array composed of multiple TMR sensors can jointly monitor the loosening of the nut, thereby improving the accuracy of nut loosening detection.

[0009] Preferably, the rotation angle of the bolt in the horizontal direction is detected by multiple TMR sensors evenly arranged in the outer circumference direction of the bolt, and the displacement of the bolt in the vertical direction is detected by multiple TMR sensors evenly arranged in the axial direction of the bolt.

[0010] Preferably, the second permanent magnet is fixed to the housing by a cantilever beam, which can be a single cantilever beam, a double cantilever beam, or a ring-shaped cantilever beam.

[0011] Compared with the single cantilever beam structure, the double cantilever beam structure has higher frequency stability and vibration suppression capability in dynamic response, which helps to improve the identification accuracy of magnetic field disturbance, and is especially suitable for complex outdoor application scenarios such as high wind speed and strong electromagnetic interference; the ring-shaped cantilever structure has multi-directional response capability, which can capture magnetic field disturbances in different directions, thereby improving the detection sensitivity and direction resolution. The structure is particularly suitable for complex structures with asymmetric vibration sources or uneven spatial magnetic field distribution, effectively enhancing the response capability and discrimination accuracy of the sensor to the slight loosening state.

[0012] Preferably, the device further comprises a first metal plug and a solder pad, the TMR sensor chip is fixed at one end of the first metal plug through the solder pad, and the other end of the first metal plug is fixed on the shell through the solder pad after penetrating through the shell.

[0013] Preferably, the device further comprises a protective shell and a second metal plug, one end of the second metal plug is fixed on the solder pad outside the shell, and the other end of the second metal plug is fixed on the protective shell through the solder pad after penetrating through the protective shell.

[0014] Preferably, the device further comprises a detection circuit and a signal processing circuit, the input end of the detection circuit is connected to the TMR sensor chip for detecting the output voltage of the TMR sensor chip, and the input end of the signal processing circuit is connected to the output end of the detection circuit for filtering, amplifying and compressing the output voltage to obtain a processed signal.

[0015] The application also provides a high-precision bolt loosening detection method based on a TMR sensor spatial array, which adopts the detection device, and the method comprises the following steps: placing a first permanent magnet on the nut of a bolt to be detected, determining the installation positions of a plurality of TMR sensors according to the distance between the nut and the screw cap, performing feature processing on the output voltages of the plurality of TMR sensors to obtain physical features, inputting the output voltages of the plurality of TMR sensors and the physical features into a trained deep learning model, and predicting the rotation angle of the bolt to be detected in the horizontal direction and the displacement of the bolt to be detected in the vertical direction.

[0016] Preferably, the physical features comprise: the resistance change amount of each TMR sensor, the estimated value of the horizontal rotation angle of the bolt calculated independently based on the output voltage of each TMR sensor in the outer circumferential direction of the nut, the estimated value of the vertical displacement of the bolt calculated independently based on the output voltage of each TMR sensor in the axial direction of the nut, the spatial variance of the output voltages of all TMR sensors in the outer circumferential direction of the nut, and the difference value of the output voltages of any two TMR sensors in the axial direction of the nut.

[0017] Preferably, the first TMR sensor measures the rotation angle of the nut in the horizontal direction i as follows: ​

[0018] a sensitivity constant of the i-th TMR sensor in the outer circumferential direction of the bolt, i a sensitivity coefficient of the i-th TMR sensor in the outer circumferential direction of the bolt, i a rotation angle of the nut in the horizontal direction measured by the i-th TMR sensor, i a reference value of the output voltage of the i-th TMR sensor in the bolt fastening state; i a displacement of the nut in the vertical direction measured by the i-th TMR sensor j

[0019] wherein, j an output voltage of the i-th TMR sensor in the axial direction of the nut, j a sensitivity constant of the i-th TMR sensor in the axial direction of the bolt, j a sensitivity coefficient of the i-th TMR sensor in the axial direction of the bolt, j a displacement of the nut in the vertical direction measured by the i-th TMR sensor, j a reference value of the output voltage of the i-th TMR sensor in the bolt fastening state; a spatial variance of the output voltage of all TMR sensors in the outer circumferential direction of the nut

[0020] wherein, is a total number of TMR sensors in the outer circumferential direction of the nut; is a difference between the output voltages of any two TMR sensors in the axial direction of the nut

[0021] wherein, respectively represent the output voltages of any two TMR sensors in the axial direction of the nut, and it is assumed that the TMR sensors are uniformly arranged in the axial direction of the nut​​​​​​​​​​​​​​​​J One TMR sensor, , , , .

[0022] Preferably, the training process of a deep learning model includes: On a calibration platform, by moving the nut of the bolt to be tested, the rotation angle of the bolt in the horizontal direction and the displacement in the vertical direction are read, and the output voltage of multiple TMR sensors is recorded. The output voltage is processed to obtain physical features, and training samples are constructed. The training samples include the output voltage, physical features, the rotation angle of the bolt to be detected in the horizontal direction and the displacement in the vertical direction. Using output voltage and physical characteristics as inputs, and the rotation angle of the bolt to be detected in the horizontal direction and the displacement in the vertical direction as output labels, a deep learning model is trained to obtain a well-trained deep learning model. Attached Figure Description

[0023] Figure 1 This is a front view of the high-precision bolt loosening detection device based on a TMR sensor spatial array provided in Embodiment 1 of the present invention; Figure 2 This is a side view of the high-precision bolt loosening detection device based on a TMR sensor spatial array provided in Embodiment 1 of the present invention; Figure 3 This is a side view from another perspective of the high-precision bolt loosening detection device based on a TMR sensor spatial array provided in Embodiment 1 of the present invention; Figure 4 This is a schematic diagram showing the positions of multiple TMR sensors in the high-precision bolt loosening detection device based on a TMR sensor spatial array provided in Embodiment 1 of the present invention. Figure 5 This is a schematic diagram of the installation of the TMR sensor in the high-precision bolt loosening detection device based on a TMR sensor spatial array provided in Embodiment 1 of the present invention. Figure 6 This is a schematic diagram of the high-precision bolt loosening detection method based on a TMR sensor spatial array provided in Embodiment 2 of the present invention; Figure 7 This is a schematic diagram of the TMR sensor in the high-precision bolt loosening detection method based on a TMR sensor spatial array provided in Embodiment 2 of the present invention. Figure 8 This is a schematic diagram of the cantilever beam in the high-precision bolt loosening detection device based on a TMR sensor spatial array provided in Embodiment 1 of the present invention; Figure 9This is a schematic diagram of another structure of the cantilever beam in the high-precision bolt loosening detection device based on a TMR sensor spatial array provided in Embodiment 1 of the present invention; Figure 10 This is a schematic diagram of another structure of the cantilever beam in the high-precision bolt loosening detection device based on a TMR sensor spatial array provided in Embodiment 1 of the present invention; In the diagram: 10 First permanent magnet, 20 TMR sensor, 21 Second permanent magnet, 22 Cantilever beam, 23 TMR sensor chip, 24 Housing, 31 First metal plug, 32 Solder pad, 33 Second metal plug, 40 Protective shell, 100 Bolt to be tested, 101 Nut, 102 Screw, 1021 Nut, 1022 Screw rod, 50 Detection circuit, 60 Signal processing circuit. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments and with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0025] Example 1 like Figure 1 As shown, this embodiment provides a high-precision bolt loosening detection device based on a TMR sensor spatial array. By arranging multiple TMR sensors 20 around the bolt 100 to be detected, the accuracy of bolt loosening detection is improved. See also... Figure 2 and Figure 3 The bolt 100 to be tested includes a nut 101 and a screw 102. The nut 101 is threaded onto the threaded shank 1022 of the screw 102. When the bolt is used to connect the angle steel of the transmission tower, the end of the threaded shank 1022 away from the nut 1021 passes through the angle steel, and then the nut 101 is threaded onto the threaded shank 1022. Tightening the nut 101 achieves a fixed connection to the angle steel. The bolt 100 to be tested may include, but is not limited to, M10, M12, M16, M18, and M20.

[0026] The device includes a first permanent magnet 10, a TMR sensor 20, a first metal plug 31, a solder pad 32, a second metal plug 33, and a protective shell 40. The first permanent magnet 10 is fixed to the nut 101 of the bolt 100 to be tested by magnetic force or structural adhesive. Multiple TMR sensors 20 are evenly arranged along the axial direction and the outer circumference of the bolt, with the axial direction of the bolt being... Figure 2 The Z-direction shown is the outer circumferential direction of the bolt. Figure 1The R direction shown in the figure is perpendicular to the Z direction. The TMR sensor 20 comprises a second permanent magnet 21 and a TMR sensor chip 23, and the model of the TMR sensor chip 23 can be MagnTek MT6835. Referring to Figure 5 , Figure 5 In the figure, only one TMR sensor is installed, the second permanent magnet 21 is fixed on the shell 24 through the cantilever beam 22, the second permanent magnet 21 is bonded on the cantilever beam 22 through glue, the two ends of the cantilever beam 22 are respectively fixed on the two inner side walls of the shell 24, the TMR sensor chip 23 is fixed on one end of the first metal plug 31 through the soldering pad 32, the other end of the first metal plug 31 is fixed on the shell 24 through the soldering pad 32 after passing through the shell 24; the shell 24 is fixed and connected on the inner side wall of the protective shell 40 through the second metal plug 33 and the soldering pad 32, one end of the second metal plug 33 is fixed on the soldering pad 32 outside the shell 24, and the other end of the second metal plug 33 is fixed on the protective shell 40 through the soldering pad 32 after passing through the protective shell 40. When the first permanent magnet 10 moves relative to the second permanent magnet 21, the magnetic field in which the TMR sensor chip 23 is located changes.

[0027] In the present application, a plurality of TMR sensors jointly constitute a sensor array to detect the loosening of the nut in the horizontal direction and the vertical direction. The plurality of TMR sensors 20 arranged uniformly in the circumferential direction of the bolt are used to detect the rotation angle of the bolt in the horizontal direction, and the plurality of TMR sensors 20 arranged uniformly in the axial direction of the bolt are used to detect the displacement of the bolt in the vertical direction, so as to cover the displacement range and rotation range of the nut as much as possible when the nut is loosened. When the nut is loosened, the plurality of TMR sensors jointly constitute a sensor array to jointly monitor the loosening of the nut, thereby improving the accuracy of the nut loosening detection.

[0028] Figure 4 As shown in the figure, eight TMR sensors are taken as an example, the eight TMR sensors are arranged uniformly in two layers, one in each of the four faces of the shell 24, and the sensors in each layer are arranged in a central symmetric manner. In actual work, different shell sizes, face numbers and sensor installation numbers can be designed according to different sizes of the measured nut, for example, for a large size of the measured nut, a shell with six faces can be designed, and three layers of a total of 18 TMR sensors can be installed.

[0029] The selection of the first permanent magnet 10 and the second permanent magnet 21 needs to consider the matching of the magnetic field strength and the size, and is preferably made of a rare earth high magnetic energy level material (such as neodymium iron boron), so as to ensure that a sufficient magnetic field strength is provided in a smaller volume to trigger the response of the TMR sensor chip, and has good temperature resistance and corrosion resistance, and is suitable for outdoor long-time use requirements.

[0030] The first metal plug 31 and the solder pad 32, and the second metal plug 33 and the solder pad 32, are reinforced with conductive adhesive to improve overall signal stability and anti-interference capability, making them suitable for long-cycle, low-maintenance remote monitoring applications. To ensure the reliability and repeatability of the sensor response, the relative position between the cantilever beam 22 and the housing 24 must be kept stable during installation to avoid error drift caused by stress concentration or micro-deformation of the housing. In addition, a flexible buffer layer can be added at the installation location to provide both vibration isolation and temperature compensation, adapting to the typical working environment of large day-night temperature differences and frequent micro-vibrations of transmission towers. In actual assembly, the protective shell 40 can be made of insulating material with dustproof and corrosion-resistant properties to adapt to outdoor high humidity and high temperature difference environments.

[0031] The second permanent magnet 21, the cantilever beam 22, and the TMR sensor chip 23 constitute a magnetic sensing element. During the installation of the TMR sensor 20, the second permanent magnet 21 is aligned with the first permanent magnet 10 to ensure that the magnetic sensing element and the end face of the nut are in the same axial plane, avoiding magnetic deviation caused by installation tilt. Simultaneously, the solder pad connection area requires the use of highly reliable solder and a protective coating to enhance oxidation and electrochemical corrosion resistance, ensuring the stability of signal output during long-term operation.

[0032] Cantilever beam 22 can be a single cantilever beam, a double cantilever beam, or a cantilever beam with a ring, such as... Figure 8 As shown, the cantilever beam 22 is a single cantilever beam, with the second permanent magnet 21 bonded to the middle position of the cantilever beam 22, and the two ends of the cantilever beam 22 fixed to the two inner sidewalls of the housing 24, respectively. Figure 9 As shown, the cantilever beam 22 is a double cantilever beam, which can be considered as two parallel single cantilever beams. The second permanent magnet 21 is bonded to the middle position of the cantilever beam 22. Compared with the single cantilever beam structure, the double cantilever beam structure exhibits higher frequency stability and vibration suppression capability in dynamic response, which helps to improve the accuracy of magnetic field disturbance recognition. It is particularly suitable for complex outdoor application scenarios such as high wind speed and strong electromagnetic interference. This structure can also achieve an optimized balance between sensitivity and anti-interference capability by adjusting the distance between the two arms.

[0033] like Figure 10 As shown, the cantilever beam 22 is a ring-shaped cantilever beam with a ring structure in the middle, and the second permanent magnet 21 is bonded to the ring structure. The ring-shaped cantilever structure has multi-directional response capability, which can capture magnetic field disturbances in different directions, thereby improving detection sensitivity and directional resolution. This structure is particularly suitable for complex structures with asymmetric vibration sources or uneven spatial magnetic field distribution, effectively enhancing the sensor's response capability and discrimination accuracy to minor loosening conditions.

[0034] See Figure 6 and Figure 7When the bolt loosens, the position of the nut 101 on the screw 1022 changes, causing the first permanent magnet 10 to move. The first permanent magnet 10 moves relative to the second permanent magnet 21, and the change in the magnetic field of the second permanent magnet 21 causes a change in the vibration frequency of the cantilever beam 22. Since the cantilever beam 22 is connected to the housing 24, this causes a change in the vibration frequency of the housing 24, which in turn causes a change in the magnetic field of the TMR sensor chip 23 in each TMR sensor 20. This changes the magnetic sensing resistor in the TMR sensor chip 23. The resistance of the external resistor changes due to the influence of the changing magnetic field, causing a change in the current in the circuit, which in turn affects the resistance of the external resistor. The voltage across the two ends changes, and the external resistor The voltage across the two ends is the output voltage of the TMR sensor chip 23. The input terminal of the detection circuit 50 is connected to the TMR sensor chip 23 to detect the output voltage of the TMR sensor chip 23. The input terminal of the signal processing circuit 60 is connected to the output terminal of the detection circuit 50 to filter, amplify and compress the output voltage to obtain the processed signal, which is the output voltage of the TMR sensor.

[0035] The signal processing circuit 60 includes a filtering unit, an amplification unit, and a compression unit. TMR signals are often affected by noise, especially high-frequency noise. Therefore, a low-pass filter can be used to smooth the signal and remove this noise. The filter's transfer function... for: ,in This is the time constant of the filter. Filtering reduces noise components in the signal, thereby improving signal stability and accuracy. Secondly, since the output signal of a TMR sensor may be weak, it typically needs to be amplified by a gain amplifier to increase its amplitude, making it easier for the system to detect. Gain factor G The processed signal for:

[0036] Dynamic range compression (DLL) technology can be used to limit the amplitude of signal variations within a reasonable range, avoiding misinterpretations caused by excessive signal amplification. The formula for DLL is: ,in, The compression factor controls the intensity of compression. To remove noise from the signal, Fourier transform denoising can be used, which transforms the signal to the frequency domain and utilizes a frequency domain filter. This process removes unnecessary noise components, resulting in a denoised signal. The Fourier transform denoising formula is as follows: ,in, It is the Fourier transform of the signal.

[0037] The present application utilizes the working mechanism of TMR sensors based on the principle of magnetic induction to achieve efficient non-contact detection of bolt loosening state, significantly reducing the structural damage and maintenance complexity caused by the physical modification (such as punching, welding, etc.) of the bolt in traditional detection methods. This non-destructive testing method not only maintains the integrity of the bolt, avoiding the necessity of excessive modification, but also ensures the efficiency and accuracy of the detection process. In addition, the device adopts a wireless passive working mode, avoiding the service life limitation of traditional battery-powered mode and the maintenance problems caused by battery replacement. The absence of external power supply enables the device to work stably for a long time without frequent battery replacement or power supply maintenance, greatly improving the reliability and durability of the system, especially suitable for applications where external power supply is not easily accessible. Compared with traditional active sensor systems, the device has a longer working cycle and lower maintenance cost, providing a solid technical guarantee for continuous monitoring in large-scale industrial environments.

[0038] The sensor design of the device is simple and compact, with flexible and diverse installation methods, without the need for complex cable wiring. Users can quickly and securely install the sensor on the bolt surface through structural glue, buckle-type fixation, or magnetic attraction, which greatly simplifies the manual deployment process, shortens the installation time, and greatly reduces the installation difficulty. With this convenient installation method, the device can be quickly applied in various industrial environments, especially in complex mechanical equipment and high-altitude work fields, significantly improving the deployment efficiency. To further enhance the functionality of the system, the sensor module can be used with an integrated antenna or low-power communication module to form a complete remote wireless monitoring system. This system not only enables timed collection and real-time monitoring of bolt status, but also has remote early warning upload function. When the bolt is loose or abnormal, the system will automatically send an early warning signal and transmit data to the monitoring center through wireless means, ensuring that equipment managers can monitor the equipment status in real time and respond quickly. This function makes the device have high application value in the fields of structural health monitoring, fault prediction, and equipment maintenance, providing accurate data support for equipment monitoring in industrial production, thereby realizing intelligent and remote maintenance management mode, further improving the safety and reliability of equipment operation.

[0039] Example 2 The embodiment provides a high-precision bolt loosening detection method based on a TMR sensor spatial array, adopts the high-precision bolt loosening detection device based on the TMR sensor spatial array in the embodiment 1, and the method comprises the following steps: placing a first permanent magnet 10 on a nut 101 of a bolt 100 to be detected; determining the installation positions of a plurality of TMR sensors 20 according to the distance between the nut 101 and a screw cap 1021 of a screw 102; performing feature processing on the output voltages of the plurality of TMR sensors 20 to obtain physical features; and inputting the output voltages of the plurality of TMR sensors 20 and the physical features into a trained deep learning model to predict the rotation angle of the bolt 100 to be detected in the horizontal direction and the displacement of the bolt 100 to be detected in the vertical direction.

[0040] The physical features comprise: the resistance change amount of each TMR sensor, the bolt horizontal rotation angle estimation value independently calculated based on the output voltage of each TMR sensor in the outer circumferential direction of the nut, the bolt vertical displacement estimation value independently calculated based on the output voltage of each TMR sensor in the axial direction of the nut, the output voltage spatial variance of all TMR sensors in the outer circumferential direction of the nut, and the difference value of the output voltages of any two TMR sensors in the axial direction of the nut. The calculation methods of each physical feature are introduced as follows: In the bolt loosening detection, the relationship between the resistance change of a TMR (tunnel magnetoresistance) sensor and the rotation angle of the bolt in the horizontal direction is as follows:

[0041] The formula shows that the magnetic sensitive resistance change of the TMR sensor is in a linear relationship with the rotation angle of the bolt. Wherein, is the magnetic sensitive resistance of the TMR sensor at the current moment, is the initial resistance of the TMR sensor, is the sensitivity coefficient of the TMR sensor.

[0042] The relationship between the output voltage of the TMR sensor and the magnetic sensitive resistance change of the TMR sensor is as follows:

[0043] Wherein, is the sensitivity constant of the TMR sensor, representing the proportional relationship between the sensor output voltage signal and the sensor magnetic sensitive resistance change.

[0044] ​​​​The present application has multiple TMR sensors, and the embodiment takes 8 TMR sensors as an example for introduction, 4 TMR sensors are uniformly arranged in the outer circumferential direction (R direction) of the bolt, 2 TMR sensors are arranged in the axial direction of the bolt, the resistance change of the first TMR sensor is i :

[0045] Among them, represents the initial resistance of the TMR sensor, and the initial resistance of all TMR sensors in the present application is the same. represents the sensitivity coefficient of the TMR sensor, and in the present application, the sensitivity coefficients of all TMR sensors in the outer circumferential direction of the nut are the same. The sensitivity coefficients of all TMR sensors in the axial direction of the nut are the same. is the rotation angle of the nut in the horizontal direction.

[0046] The present application measures the rotation angle of the bolt in the horizontal direction through multiple TMR sensors uniformly arranged in the outer circumferential direction of the bolt, and measures the displacement of the bolt in the vertical direction through multiple TMR sensors uniformly arranged in the axial direction of the bolt. Therefore, the output voltage of the first TMR sensor is represented as: i

[0047] The rotation angle of the nut in the horizontal direction measured by the first TMR sensor is obtained by moving the term to get: i

[0048] represents the sensitivity constant of the first TMR sensor in the outer circumferential direction of the bolt, i represents the sensitivity coefficient of the first TMR sensor in the outer circumferential direction of the bolt, represents the rotation angle of the nut in the horizontal direction measured by the first TMR sensor, i represents the output voltage reference value of the first TMR sensor in the bolt fastening state. i i The output voltage of each TMR sensor in the axial direction of the nut is:

[0049]

[0050] The rotation angle of the nut in the horizontal direction measured by the first TMR sensor is obtained by moving the term to get: j ​​​​​​​​​​The displacement of the nut in the vertical direction measured by the i-th TMR sensor is:

[0051] wherein, represents the change of the magnetic induction intensity experienced by the i-th TMR sensor in the axial direction of the nut as the displacement of the nut in the vertical direction changes. j represents the sensitivity constant of the i-th TMR sensor in the axial direction of the bolt, represents the sensitivity constant of the i-th TMR sensor in the axial direction of the bolt, j represents the sensitivity constant of the i-th TMR sensor in the axial direction of the bolt, represents the sensitivity constant of the i-th TMR sensor in the axial direction of the bolt, j represents the sensitivity constant of the i-th TMR sensor in the axial direction of the bolt, represents the displacement of the nut in the vertical direction measured by the i-th TMR sensor, j represents the displacement of the nut in the vertical direction measured by the i-th TMR sensor, represents the reference value of the output voltage of the i-th TMR sensor in the bolt fastened state. The vertical direction in the present invention is parallel to the axial direction of the nut, the horizontal direction is parallel to the circumferential direction of the nut, and the vertical direction and the horizontal direction are perpendicular to each other. j

[0052] The spatial variance of the output voltage of all TMR sensors in the circumferential direction of the nut is:

[0053] wherein, is the mean value of the output voltage of all TMR sensors in the circumferential direction of the nut, , is the total number of TMR sensors in the circumferential direction of the nut.

[0054] In the bolt fastened state, the magnetic field distribution is symmetrical, the output values of each TMR sensor in the circumferential direction of the nut are close, and the variance is very small. When the bolt is horizontally rotated or tilted, the symmetry of the magnetic field is destroyed, the output voltage of each TMR sensor in the circumferential direction of the nut is different, and the variance significantly increases. Therefore, the spatial variance of the output voltage of all TMR sensors in the circumferential direction of the nut is a strong indicator for detecting the loosening of the nut.

[0055] The difference between the output voltages of any two TMR sensors in the axial direction of the nut is:

[0056] wherein, , ​respectively, represent the difference between any two TMR sensor output voltages in the nut axial direction, assuming that there are J TMR sensors evenly arranged in the nut axial direction, , , , .

[0057] When the nut moves in the vertical direction, the distance between the first permanent magnet 10 on the nut and all TMR sensors in the nut axial direction changes synchronously, and all difference values will drift uniformly. When the nut is tilted, the difference values of different azimuth angles will show different change patterns. Therefore, the difference between any two TMR sensor output voltages in the nut axial direction is a key indicator to distinguish vertical displacement and tilt.

[0058] The training process of the deep learning model includes: On a calibration platform, the output voltages of the plurality of TMR sensors 20 are read by moving the nut 101 of the bolt 100 to be detected, and the rotation angle of the bolt 100 to be detected in the horizontal direction and the displacement in the vertical direction are read. The output voltages are processed to obtain physical characteristics, and training samples are constructed, including output voltages, physical characteristics, rotation angle and axial displacement of the bolt 100 to be detected. The output voltages and physical characteristics are taken as inputs, and the rotation angle of the bolt 100 to be detected in the horizontal direction and the displacement in the vertical direction are taken as output labels, and the deep learning model is trained to obtain the trained deep learning model.

[0059] The above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A high-precision bolt loosening detection device based on a TMR sensor spatial array, characterized in that: The device includes a first permanent magnet (10) and a TMR sensor (20). Multiple TMR sensors (20) are evenly arranged along the axial direction and the outer circumference of the bolt. The TMR sensor (20) includes a second permanent magnet (21) and a TMR sensor chip (23). When the first permanent magnet (10) moves relative to the second permanent magnet (21), the magnetic field of the TMR sensor chip (23) changes.

2. The high-precision bolt loosening detection device based on a TMR sensor spatial array according to claim 1, characterized in that: The bolt's rotation angle in the horizontal direction is detected by multiple TMR sensors (20) evenly arranged in the outer circumference direction of the bolt, and the bolt's displacement in the vertical direction is detected by multiple TMR sensors (20) evenly arranged in the axial direction of the bolt.

3. The high-precision bolt loosening detection device based on a TMR sensor spatial array according to claim 1, characterized in that: The second permanent magnet (21) is fixed to the shell (24) by a cantilever beam (22), which is a single cantilever beam, a double cantilever beam, or a ring-shaped cantilever beam.

4. The high-precision bolt loosening detection device based on a TMR sensor spatial array according to claim 1, characterized in that: The device also includes a first metal plug (31) and a pad (32). The TMR sensor chip (23) is fixed to one end of the first metal plug (31) by the pad (32), and the other end of the first metal plug (31) passes through the housing (24) and is fixed to the housing (24) by the pad (32).

5. The high-precision bolt loosening detection device based on a TMR sensor spatial array according to claim 1, characterized in that: The device also includes a protective shell (40) and a second metal plug (33). One end of the second metal plug (33) is fixed to a pad (32) on the outside of the shell (24), and the other end of the second metal plug (33) passes through the protective shell (40) and is fixed to the protective shell (40) by the pad (32).

6. The high-precision bolt loosening detection device based on a TMR sensor spatial array according to claim 1, characterized in that: The device also includes a detection circuit (50) and a signal processing circuit (60). The input terminal of the detection circuit (50) is connected to the TMR sensor chip (23) to detect the output voltage of the TMR sensor chip (23). The input terminal of the signal processing circuit (60) is connected to the output terminal of the detection circuit (50) to filter, amplify and compress the output voltage to obtain the processed signal.

7. A high-precision bolt loosening detection method based on a TMR sensor spatial array, using the detection device described in any one of claims 1-6, characterized in that: The methods include: The first permanent magnet (10) is placed on the nut (101) of the bolt (100) to be tested. Based on the distance between the nut (101) and the nut (1021) of the screw (102), the installation position of multiple TMR sensors (20) is determined. The output voltage of multiple TMR sensors (20) is processed to obtain physical features. The output voltage and physical features of multiple TMR sensors (20) are input into the trained deep learning model to predict the rotation angle of the bolt (100) to be tested in the horizontal direction and the displacement in the vertical direction.

8. The high-precision bolt loosening detection method based on a TMR sensor spatial array according to claim 7, characterized in that: The physical characteristics include: the resistance change of each TMR sensor, the estimated horizontal rotation angle of the bolt calculated independently based on the output voltage of each TMR sensor in the outer circumferential direction of the nut, the estimated vertical displacement of the bolt calculated independently based on the output voltage of each TMR sensor in the axial direction of the nut, the spatial variance of the output voltage of all TMR sensors in the outer circumferential direction of the nut, and the difference between the output voltages of any two TMR sensors in the axial direction of the nut.

9. The high-precision bolt loosening detection method based on a TMR sensor spatial array according to claim 8, characterized in that: No. i The rotation angle of the nut in the horizontal direction measured by a TMR sensor. for: Indicates the first circumference of the bolt in the outer circumferential direction i The sensitivity constant of a TMR sensor, Indicates the first circumference of the bolt in the outer circumferential direction i The sensitivity coefficient of each TMR sensor, Indicates the first i The rotation angle of the nut in the horizontal direction measured by a TMR sensor. Indicates the first i The output voltage reference value of a TMR sensor under bolt-tightened condition; No. j The vertical displacement of the nut measured by a TMR sensor. for: in, The first in the axial direction of the nut j The output voltage of each TMR sensor Indicates the first bolt in the axial direction j The sensitivity constant of a TMR sensor, Indicates the first bolt in the axial direction j The sensitivity coefficient of each TMR sensor, Indicates the first j The vertical displacement of the nut measured by a TMR sensor. Indicates the first j The output voltage reference value of a TMR sensor under bolt-tightened condition; Spatial variance of output voltage of all TMR sensors in the outer circumferential direction of the nut for: in, This is the average output voltage of all TMR sensors along the outer circumference of the nut. , This represents the total number of TMR sensors along the outer circumference of the nut. The difference in output voltage between any two TMR sensors along the axial direction of the nut for: in, , These represent the output voltages of any two TMR sensors along the axial direction of the nut, respectively. The nut is assumed to have evenly distributed... J One TMR sensor, , , , .

10. The high-precision bolt loosening detection method based on a TMR sensor spatial array according to claim 8, characterized in that: The training process of a deep learning model includes: On a calibration platform, by moving the nut (101) of the bolt (100) to be tested, the rotation angle of the bolt (100) in the horizontal direction and the displacement in the vertical direction are read, and the output voltage of multiple TMR sensors (20) is recorded. The output voltage is processed to obtain physical features, and training samples are constructed. The training samples include the output voltage, physical features, the rotation angle of the bolt (100) under test in the horizontal direction and the displacement in the vertical direction. Using the output voltage and physical characteristics as inputs, and the rotation angle of the bolt (100) to be detected in the horizontal direction and the displacement in the vertical direction as output labels, a deep learning model is trained to obtain a well-trained deep learning model.

Images