A bolt detection model construction and bolt loosening detection method

Abstract

The application relates to the technical field of nondestructive testing and structure health monitoring, and discloses a bolt detection model construction and bolt loosening detection method, which comprises the following steps: applying a periodic modulated thermal excitation to a target bolt; synchronously adopting a structured light three-dimensional measurement technology to collect a bolt surface dynamic micro-deformation image sequence caused by the thermal excitation; solving the image sequence into a phase time sequence signal; taking the modulation frequency of the thermal excitation as a reference to perform a digital phase-locked demodulation operation on the phase time sequence signal, extract a response amplitude graph and a response phase delay graph, and combine the two graphs into a thermal force response fingerprint representing the bolt state; and finally, comparing the real-time acquired fingerprint with a pre-stored reference fingerprint under a fastening state, and quantitatively diagnosing the loosening state of the bolt according to the difference. The application can realize accurate quantitative detection of early micro loosening of the bolt in a non-contact and high-sensitivity manner, and has strong environmental interference resistance and engineering application value.

Description

Technical Field

[0001] This invention relates to the field of nondestructive testing and structural health monitoring technology, specifically to a bolt detection model construction and bolt loosening detection method. Background Technology

[0002] Bolted connections, as the most basic and widely used mechanical connection method, play a crucial role in key infrastructure and high-end equipment such as aerospace, rail transportation, energy equipment, and large bridges. The reliability of these connection points directly affects the safety and stable operation of the entire structure. However, under complex service environments such as vibration, impact, alternating loads, and temperature changes, bolted connections inevitably experience preload decay and even loosening, posing serious safety hazards and potentially leading to catastrophic structural failures. Therefore, regular and reliable inspection of bolted connections is a critical step in ensuring equipment safety and preventing accidents.

[0003] Currently, methods for detecting loose bolts have many limitations. Traditional methods, such as torque wrench re-inspection or tapping and listening, heavily rely on the operator's experience and sense of responsibility. These methods are not only labor-intensive and inefficient, but also often require equipment downtime, leading to high maintenance costs. More importantly, these contact-based methods are highly subjective, making it difficult to achieve quantitative and standardized assessments, and their reliability is questionable.

[0004] To overcome these shortcomings, the industry has developed a series of sensor-based monitoring technologies. For example, ultrasonic or acoustic emission-based detection methods determine preload by measuring changes in the propagation characteristics of stress waves in the bolt. While these methods offer high accuracy, they typically require sensors mounted on the bolts, making system deployment complex and costly, and difficult to apply to large groups of bolts. Furthermore, the sensors themselves can become new sources of failure. Another type of method is machine vision-based, which determines whether rotation has occurred by comparing and analyzing images of the bolt or markings on it. However, these methods generally have low sensitivity, only detecting macroscopic, visually perceptible rotation of the bolt. They are ineffective in detecting early, minor loosening stages where preload has decreased but no obvious rotation has yet occurred, failing to meet the need for early warning. They are also susceptible to interference from environmental factors such as changes in ambient light and surface oil contamination. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a bolt detection model construction and bolt loosening detection method, which solves the problem that existing detection technologies are unable to perform high-sensitivity and high-reliability quantitative detection of early-stage minor loosening of bolts under non-contact conditions.

[0006] To achieve the above objectives, the first aspect of the present invention provides a bolt loosening detection method, which includes the following steps:

[0007] Step S1: Apply a periodically varying modulated thermal excitation to the target bolt. Specifically, use a programmable power heat source to adjust its instantaneous power output. The target bolt is heated according to a preset periodic function. In one embodiment, the instantaneous power... The following relationship must be satisfied:

[0008] ;

[0009] in, For time, Average power, To modulate the amplitude, Let be the angular frequency.

[0010] Step S2: While applying the modulated thermal excitation, a structured light image sequence of the target bolt and its base area is acquired simultaneously. Specifically, a structured light measurement system is used, which includes a digital grating projector that projects an coded stripe pattern onto the target area. An industrial camera continuously acquires distorted stripe images reflected from the target surface at a frame rate higher than the modulated thermal excitation frequency to form a structured light image sequence.

[0011] Step S3: Based on the structured light image sequence, calculate the phase timing signal reflecting surface deformation information. Specifically, for each frame or group of images in the structured light image sequence, use a phase shift algorithm to calculate the wrapping phase map at that moment, and then use a spatial or temporal phase unwrapping algorithm to restore the wrapping phase to a continuous absolute phase, thereby obtaining a sequence of absolute phase maps that change continuously in time. This sequence is the phase timing signal.

[0012] Step S4: Demodulate the phase timing signal to extract a real-time thermodynamic response fingerprint with the same frequency as the modulated thermal excitation. Specifically, this step is based on the digital lock-in amplification principle to demodulate the phase timing signal. Processing is performed to extract the angular frequency. The relevant response components, this step may include:

[0013] Calculate in-phase components and orthogonal components .

[0014] In one embodiment, the response amplitude diagram and response phase delay diagram Calculated in the following way:

[0015] ;

[0016] ;

[0017] in, These are the pixel coordinates in the image. For the in-phase component, The orthogonal component is referred to here.

[0018] Step S5: Compare the real-time thermal response fingerprint with the pre-stored reference thermal response fingerprint obtained under the bolt tightening state, and determine the loosening state of the target bolt based on the difference. Specifically, the reference thermal response fingerprint is obtained by performing steps S1 to S4 under the confirmed tightening state of the target bolt. The comparison process may include:

[0019] The difference spectrum between the real-time thermal response fingerprint and the reference thermal response fingerprint is calculated, and a scalarized loosening index is constructed based on the difference spectrum. Finally, the loosening index is compared with one or more preset thresholds to classify and diagnose the loosening state, for example, to determine it as a normal, warning or alarm state.

[0020] A second aspect of the present invention provides a bolt detection model construction and bolt loosening detection system, comprising:

[0021] The modulated thermal excitation module is used to apply periodically varying modulated thermal excitation to the target bolt.

[0022] The image acquisition module is used to simultaneously acquire a sequence of structured light images of the target bolt and its base area while the modulated thermal excitation is applied.

[0023] The phase calculation module, connected to the image acquisition module, is used to calculate the phase timing signal reflecting surface deformation information based on the structured light image sequence.

[0024] The response demodulation module, connected to the phase calculation module, is used to demodulate the phase timing signal to extract a real-time thermodynamic response fingerprint with the same frequency as the modulated thermal excitation.

[0025] The comparison and diagnosis module, connected to the response demodulation module, is used to compare the real-time thermal response fingerprint with a pre-stored reference thermal response fingerprint obtained under the bolt tightening state, and determine the loosening state of the target bolt based on the comparison result.

[0026] This invention provides a bolt detection model construction and a bolt loosening detection method. It has the following beneficial effects:

[0027] 1. This invention applies active, periodically modulated thermal excitation to the target bolt and extracts the response signal using the phase-locked demodulation principle. This makes the detection results insensitive to static deposits on the target bolt surface and gradual changes in ambient light. Specifically, since this invention only extracts the response signal with the same frequency as the modulated thermal excitation, static surface deposits such as oil stains and dust, as well as interference factors such as ambient light fluctuations whose frequency is unrelated to the modulation frequency, are effectively filtered out during the phase-locked demodulation process, thereby improving the reliability of the detection results in complex industrial environments.

[0028] 2. This invention, by measuring the thermoelastic deformation response induced by thermal excitation, can reflect the change in contact thermal resistance between the bolt and the base, thereby detecting changes in physical properties caused by early, minor loosening before significant changes in geometric morphology are apparent. Specifically, early loosening of the bolt first leads to a change in the contact thermal resistance at the contact surface with the base, which in turn affects the heat conduction efficiency at the contact interface, ultimately reflected in the response amplitude and phase delay diagrams measured by this invention. Therefore, this invention does not rely on the measurement of the bolt's macroscopic geometric displacement, but rather achieves early loosening detection through the inversion of physical property parameters.

[0029] 3. This invention combines high-precision structured light phase measurement technology with phase-locked demodulation algorithm to achieve high signal-to-noise ratio detection of weak deformation signals. Specifically, the structured light measurement unit can acquire the phase timing signal of the entire field at high resolution, providing an accurate raw data foundation for deformation measurement; the phase-locked demodulation algorithm used in the response demodulation module can accurately extract the periodic deformation component related to the bolt state caused by modulation thermal excitation from the noisy phase timing signal, thereby improving the signal-to-noise ratio and overall measurement accuracy. Attached Figure Description

[0030] Figure 1 This is a flowchart of the method of the present invention;

[0031] Figure 2 This is a system architecture diagram of the present invention;

[0032] Figure 3 This is a schematic diagram of the modulation thermal excitation application of the present invention;

[0033] Figure 4 This is a flowchart of the phase timing signal calculation of the present invention;

[0034] Figure 5 This is a flowchart of the thermal response fingerprint extraction process of the present invention;

[0035] Figure 6 This is a flowchart of the state comparison and diagnosis process of the present invention.

[0036] Figure 7This is a schematic diagram of the configuration of the synchronous image acquisition system of the present invention.

[0037] Among them, 10 is the modulation thermal excitation module; 20 is the image acquisition module; 30 is the phase calculation module; 40 is the response demodulation module; and 50 is the comparison and diagnosis module. Detailed Implementation

[0038] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0039] Example:

[0040] Please see the appendix Figure 1 -Appendix Figure 7 This invention provides a method for constructing a bolt detection model and detecting bolt loosening, including:

[0041] S100, Apply periodically varying modulated thermal excitation to the target bolt;

[0042] In step S100, the target bolt is subjected to periodically varying modulated thermal excitation by the modulated thermal excitation module 10. The modulated thermal excitation module 10 includes a heat source and a beam shaping unit. The heat source can be a power-programmable semiconductor laser or an infrared heating lamp with a focusing lens. The beam shaping unit is used to shape the energy beam emitted by the heat source into a light spot with a specific size and uniform energy density, and project it onto the region of interest covering the head of the target bolt and part of its base.

[0043] In this embodiment, the controller of the modulated thermal excitation module 10 is configured to cause its output instantaneous power to... Over time According to the preset periodic function variation, a specific functional form is sinusoidal modulation, and its expression is:

[0044] ;

[0045] in, For heat source at all times The instantaneous power output; The average power during the modulation period; The modulation amplitude of the power; The angular frequency of the modulation.

[0046] The setting of parameters has its technical considerations, such as average power. and modulation amplitude The value was set to ensure that the total excitation power was sufficient to induce thermoelastic deformation on the target surface that could be detected by subsequent steps, while remaining well below the threshold that would cause plastic deformation or surface damage to the target bolt material. This ensured the non-destructive nature of the detection.

[0047] Modulation angular frequency The selection of the frequency is intended to place the thermal excitation signal within a low-ambient-noise frequency band, for example, avoiding 50Hz / 60Hz power frequency interference commonly found in industrial sites caused by power supply systems, and mechanical vibrations of specific frequencies generated by the operation of mechanical equipment. In the implementation, the modulation frequency... .

[0048] (in The frequency range can be set to a low range, such as 0.1 Hz to 10 Hz. This frequency range ensures that the heat has enough time to be conducted inside the target in each modulation cycle, while effectively distinguishing it from high-frequency mechanical vibration noise.

[0049] The modulated thermal excitation applied in this manner determines the angular frequency. This will serve as the reference frequency for phase-locked demodulation in the subsequent step S400. The direct physical result of this step is to induce a microscopic, periodic thermoelastic deformation field with the same frequency as the excitation source on the surface of the target bolt and its base region. The characteristics of this deformation field are directly related to the thermophysical parameters of the target region.

[0050] S200: While applying modulated thermal excitation, the structured light image sequence of the target bolt and its base area is acquired simultaneously;

[0051] In step S200, while the modulation thermal excitation module 10 applies modulation thermal excitation to the target bolt, the image acquisition module 20 synchronously acquires a sequence of structured light images of the target bolt and its base area. The image acquisition module 20 includes a digital grating projector and an industrial camera, which together constitute a structured light measurement unit.

[0052] The system in this embodiment further includes a synchronization control unit, which is connected to the modulation thermal excitation module 10 and the image acquisition module 20. During operation, the synchronization control unit outputs a control signal, which drives the modulation thermal excitation module 10 to output power according to a preset modulation function. On the other hand, it strictly triggers the image acquisition operation of the image acquisition module 20 according to the phase of the modulation function. This synchronization mechanism ensures that during subsequent data processing, the timestamp of each frame of the acquired image has a definite correspondence with the instantaneous phase of the thermal excitation.

[0053] In the specific acquisition process, the digital grating projector projects a set of pre-coded sinusoidal fringe patterns onto the target area being measured, while the industrial camera continuously acquires distorted fringe images that are highly modulated and deformed by the target surface from another angle. This acquisition process is not a single snapshot, but continuously covers multiple complete modulation thermal excitation cycles, thereby forming a structured light image sequence containing information on the periodic changes in surface morphology over time. This sequence is the direct data input for the subsequent step S300 to perform phase calculation.

[0054] Setting the acquisition parameters is crucial for obtaining effective deformation information. The image acquisition frame rate of an industrial camera... It was set to a modulation frequency much higher than that of the modulation thermal excitation. Specifically, the frame rate of data acquisition .

[0055] The sampling theorem must be satisfied to ensure that the frequency of thermally excited events can be recorded without distortion. The periodic deformation, in the implementation method, Set as modulation frequency 10 times or more, in addition, the total collection time Set as modulation period (in The data should be an integer multiple of the total number of modulation cycles, such as 5 or 10 complete modulation cycles, to ensure that the subsequent phase-locked demodulation algorithm has sufficient data covering the complete cycle when performing integration operations.

[0056] S300. Based on the structured light image sequence, calculate the phase timing signal that reflects the surface deformation information;

[0057] In step S300, the structured light image sequence acquired in step S200 is processed by the phase calculation module 30 to calculate the phase time sequence signal reflecting the change of the three-dimensional morphology of the target surface over time.

[0058] In this implementation, the N-step phase shift method is used to calculate the phase. For each discrete time point in the acquired structured light image sequence, there is a set (N images) of fringe images projected by a digital grating projector with a fixed phase step. The phase calculation module 30 first processes this set of images to calculate the wrap-around phase map at that moment. .

[0059] The calculation formula is as follows:

[0060] ;

[0061] in, These are the pixel coordinates in the image; For example, the total number of phase shift steps. ; This is the index of the current phase shift step. ; In the first During phase shift, pixel points The light intensity value collected at the location.

[0062] Due to the computational characteristics of the arctangent function, the wrapped phase calculated in the above steps... The range is truncated at Within the interval, there exists The phase transition cannot directly reflect the true continuous surface morphology of an object. Therefore, it is necessary to use a phase unwrapping algorithm to restore it to a spatially continuous absolute phase. The phase calculation module 30 can execute a spatial phase unwrapping algorithm based on path integral or a temporal phase unwrapping algorithm using time series information to eliminate phase jumps and obtain the absolute phase map at that moment.

[0063] For each time-series image subset acquired in step S200, the phase calculation and unpacking process described above are repeated to obtain a series of temporally continuous absolute phase images. This image sequence constitutes the phase-time sequence signal. This signal accurately records every pixel within the target area. The phase value over time The minute fluctuations are the direct input data for subsequent extraction of thermal response fingerprints.

[0064] S400: Demodulate the phase timing signal to extract the real-time thermal response fingerprint at the same frequency as the modulated thermal excitation;

[0065] In step S400, the phase timing signal generated in step S300 is processed by the response demodulation module 40. Demodulation processing is performed to extract a real-time thermodynamic response fingerprint at the same frequency as the modulated thermal excitation. This processing adopts the principle of digital phase-locked amplification.

[0066] The response demodulation module 40 demodulates each pixel in the image. The corresponding phase timing signal Cross-correlation operations are performed with the in-phase reference signal and the quadrature reference signal, respectively. The frequencies of the in-phase reference signal and the quadrature reference signal are compared with the angular frequency of the modulated thermal excitation in step S100. Same. (Through the entire collection duration) Integrating the components, we can calculate the in-phase components. and orthogonal components .

[0067] The specific calculation formula is as follows:

[0068] ;

[0069] ;

[0070] in, For pixels Phase timing signal at the location; For time; This is the total signal acquisition time; Given the modulation angular frequency of the thermal excitation, the physical effect of this integral operation is to filter out all frequencies related to... Unrelated frequency components are retained, only the response portion of the signal induced by modulation thermal excitation is preserved.

[0071] Calculate the in-phase components and orthogonal components Subsequently, a real-time thermal response fingerprint can be generated, which includes a response amplitude map. and response phase delay diagram .

[0072] Response amplitude diagram This represents the amplitude of the thermoelastic deformation at each pixel point under the modulation frequency, and its calculation formula is:

[0073] ;

[0074] Response phase delay diagram This represents the phase delay of the thermoelastic deformation at each pixel relative to the thermal excitation source at the modulation frequency. The calculation formula is:

[0075] ;

[0076] By performing the above calculations on each pixel within the field of view, two complete feature maps with the same resolution as the original image can be obtained:

[0077] The response amplitude diagram and the response phase delay diagram together constitute a real-time thermodynamic response fingerprint of the current bolt state, and serve as input for comparison and diagnosis in the subsequent step S500.

[0078] S500 compares the real-time thermal response fingerprint with the pre-stored reference thermal response fingerprint obtained under the bolt tightening state, and determines the loosening state of the target bolt based on the difference.

[0079] In step S500, the real-time thermal response fingerprint extracted in step S400 is compared with the pre-stored benchmark thermal response fingerprint by the comparison and diagnosis module 50, and the loosening state of the target bolt is determined based on the comparison result.

[0080] In an initial calibration phase, the target bolt needs to be in a confirmed, compliant tightening state. Steps S100 to S400 are then executed to obtain its corresponding response amplitude map and response phase delay map. These two maps are stored in the storage unit of the comparison and diagnostic module 50 as a reference thermodynamic response fingerprint for subsequent comparisons, and are denoted as the reference amplitude map. and reference phase delay diagram .

[0081] During real-time detection, the comparison and diagnostic module 50 first calculates the difference between the real-time thermal response fingerprint and the reference thermal response fingerprint. This calculation is performed pixel-by-pixel within a specified region of interest, which is typically defined as covering the edge of the bolt head and its contact ring with the base. An amplitude difference map is generated by subtracting the real-time map from the reference map pixel by pixel. Phase delay difference map .

[0082] Subsequently, in order to convert the two-dimensional difference map information into a single scalar value for easier judgment, the comparison and diagnosis module 50 constructs a loosening index. This index is obtained by statistically analyzing and fusing the differences within the region of interest. In the implementation method, the loosening index is... It is calculated by weighted summation of the mean values ​​of amplitude difference and phase delay difference:

[0083] ;

[0084] in, This represents the mean of the amplitude difference spectrum within the region of interest. This represents the mean value of the phase delay difference map within the region of interest. and The preset weighting coefficients, and The values ​​of these two weighting coefficients are determined based on experimental data to reflect the differences in sensitivity of the two features, amplitude and phase delay, to a specific type of bolt loosening state.

[0085] Finally, the comparison and diagnosis module 50 will calculate the loosening index. It compares with one or more preset thresholds to achieve graded diagnosis of loosening status, for example, setting a warning threshold. and alarm threshold .

[0086] like If so, the bolt condition is considered normal; if If the bolt condition is determined to be a warning; if If the condition is confirmed, the bolt status is determined to be alarm-indicating. This final diagnostic result serves as the output of this detection method and can be transmitted to the upper-level monitoring system or directly displayed to the operator.

[0087] See attached document Figure 2 A bolt detection model construction and bolt loosening detection system is disclosed. The system includes a modulation thermal excitation module 10, an image acquisition module 20, a phase calculation module 30, a response demodulation module 40, and a comparison and diagnosis module 50.

[0088] The modulation thermal excitation module 10 performs step S100, applying a modulation frequency to the target bolt. The periodic thermal excitation, the image acquisition module 20, its function is to execute step S200, under synchronous control, to acquire a sequence of structured light images reflecting the morphological changes of the target surface under thermal excitation, and output the sequence data to the phase calculation module 30.

[0089] The phase calculation module 30 performs step S300. This module receives a structured light image sequence from the image acquisition module 20 and calculates the absolute phase map at each moment by executing an N-step phase shifting method and a phase unwrapping algorithm, thereby constructing a phase-time sequence signal. The signal is then transmitted to the response demodulation module 40.

[0090] The response demodulation module 40 performs step S400, which involves receiving phase timing signals. and modulated angular frequency Using the reference frequency, digital phase-locked demodulation is performed. The output of this module is a real-time thermal response fingerprint, i.e., a response amplitude diagram. and response phase delay diagram .

[0091] The comparison and diagnosis module 50 performs step S500. This module receives the real-time thermal response fingerprint from the response demodulation module 40 and retrieves the pre-stored reference thermal response fingerprint from its internal storage unit. By calculating the difference between the two, a loosening index is constructed based on the difference. Finally, the index is compared with a preset threshold, and a diagnostic result on the loosening state of the target bolt is output.

[0092] In summary, the technical solution of this invention actively applies modulated thermal excitation and uses structured light phase measurement technology to accurately capture the microscopic thermoelastic deformation caused by it. Then, it combines a phase-locked demodulation algorithm to extract the thermodynamic response fingerprint closely related to the bolt tightness. Finally, by comparing it with the reference state, it achieves quantitative and reliable non-contact detection of the bolt loosening state.

Claims

1. A bolt detection model construction and bolt loosening detection method, characterized in that, Includes the following steps: Apply periodically varying modulated thermal excitation to the target bolt; While applying the modulated thermal excitation, a sequence of structured light images of the target bolt and its base region is simultaneously acquired; Based on the structured light image sequence, the phase timing signal reflecting surface deformation information is calculated; The phase timing signal is demodulated to extract a real-time thermodynamic response fingerprint with the same frequency as the modulated thermal excitation; The real-time thermal response fingerprint is compared with a pre-stored reference thermal response fingerprint obtained under the bolt tightening state, and the loosening state of the target bolt is determined based on the difference.

2. The bolt detection model construction and bolt loosening detection method according to claim 1, characterized in that, Both the real-time thermal response fingerprint and the reference thermal response fingerprint include: Response amplitude diagram and response phase delay diagram.

3. The bolt detection model construction and bolt loosening detection method according to claim 1, characterized in that, The specific steps for demodulating the phase timing signal are as follows: For phase-time signals Angular frequency based on modulated thermal excitation Calculate their in-phase components respectively. and orthogonal components This constitutes the real-time thermal response fingerprint.

4. The bolt detection model construction and bolt loosening detection method according to claim 2, characterized in that, The response amplitude diagram and response phase delay diagram Calculated in the following way: ； ； in, These are the pixel coordinates in the image. For in-phase components, These are orthogonal components.

5. The bolt detection model construction and bolt loosening detection method according to claim 1, characterized in that, The instantaneous power of the modulated thermal excitation satisfy: ； in, For time, Average power, To modulate the amplitude, ω is the angular frequency.

6. The bolt detection model construction and bolt loosening detection method according to claim 1, characterized in that, The steps for calculating the phase timing signal include: The wrapped phase diagram at each moment is calculated using the N-step phase shift method, and then the continuous absolute phase is obtained through the phase unwrapping algorithm, thus forming the phase timing signal.

7. The bolt detection model construction and bolt loosening detection method according to claim 1, characterized in that, The step of comparing the real-time thermal response fingerprint with the reference thermal response fingerprint includes: Calculate the difference spectrum between the real-time thermal response fingerprint and the baseline thermal response fingerprint, and construct a scalarized loosening index based on the difference spectrum.

8. The bolt detection model construction and bolt loosening detection method according to claim 7, characterized in that, The construction of the loosening index includes: The differences between the response amplitude map and the response phase delay map are weighted and fused.

9. The bolt detection model construction and bolt loosening detection method according to claim 7, characterized in that, The step of determining the loosening state based on the difference includes: The loosening index is compared with a preset warning threshold or alarm threshold to classify and diagnose the loosening state.

10. A bolt detection model construction and bolt loosening detection system, comprising a bolt detection model construction and bolt loosening detection method according to any one of claims 1-9, characterized in that, include: The modulated thermal excitation module is used to apply periodically varying modulated thermal excitation to the target bolt. The image acquisition module is used to simultaneously acquire a sequence of structured light images of the target bolt and its base area while the modulated thermal excitation is applied; The phase calculation module is used to calculate the phase timing signal reflecting surface deformation information based on the structured light image sequence. The response demodulation module is used to demodulate the phase timing signal to extract a real-time thermodynamic response fingerprint with the same frequency as the modulated thermal excitation. The comparison and diagnosis module is used to compare the real-time thermal response fingerprint with a pre-stored reference thermal response fingerprint obtained under the bolt tightening state, and determine the loosening state of the target bolt based on the comparison result.

Images