Intelligent slab rubber bearing capable of realizing in-situ detection of damage and identification method thereof

Abstract

This invention discloses an intelligent plate-type rubber bearing capable of in-situ damage detection, comprising, from top to bottom, a top plate, a sliding plate, a stainless steel cold-rolled plate, a middle steel plate, a rubber plate, and a bottom plate. The top and bottom plates are grooved to accommodate sensors, a vibrator, and sensing wires. The vibrator is connected to an external signal excitation module via sensing wires, and the sensors are connected to an external signal acquisition module via sensing wires. The signal excitation and acquisition modules are connected to a remote data transmission module, which processes the acquired detection signals into a cloud-based signal processing module to identify the bearing's condition. This invention also discloses a method for identifying the intelligent plate-type rubber bearing capable of in-situ damage detection. This invention can accurately identify the axial compression, deformation, and damage state of the bearing.

Description

Technical Field

[0001] This invention relates to the field of structural damage identification and health monitoring technology, and in particular to an intelligent plate rubber bearing capable of in-situ damage detection and its identification method. Background Technology

[0002] Bridge bearings are crucial components of bridges, playing a vital role in ensuring bridge structure and normal traffic flow. Their functions are primarily threefold: 1) connecting the superstructure and superstructure, allowing for free rotation and deformation; 2) coordinating the deformation of the superstructure to accommodate expansion and contraction due to temperature and humidity changes; and 3) restricting displacement. In continuous structures, irregular structures, skewed, curved, and sloping bridges, horizontal supports are needed at appropriate locations to prevent overall structural displacement caused by temperature changes, seismic forces, and other factors. Fixed bearings, unidirectional sliding bearings, and anti-falling beam devices play important roles in these aspects. Among bridge bearings, rubber bearings are widely used in bridge construction due to their space-saving design, ease of processing and installation, and low cost.

[0003] As a crucial component of bridges, damage to bridge bearings can alter the stress state of the bridge superstructure and even affect normal traffic safety. Damage to rubber bearings includes: partial or complete detachment, rubber aging and cracking, and shear deformation exceeding allowable values. In recent years, an increasing number of bridge rubber bearings have developed damage, with some bearings even showing signs of damage before the bridge is opened to traffic. With the increasing number of bridges and the widespread use of rubber bearings in my country, accurately assessing the damage to rubber bearings is of paramount importance to ensuring the safety of bridge structures.

[0004] Chinese invention patent application number 201911244894.8 discloses a seismic isolation bearing with self-detection function and its self-detection method. The method employs piezoelectric wave detection to detect internal damage to the bearing. An exciter is installed at the top, and a sensor is installed at the bottom. The exciter emits vibration signals, and the sensor receives these signals. By processing the vibration signals, features are extracted to reflect the damage state of the bearing. However, this method has the following drawbacks:

[0005] 1. Limited to localized detection. Existing solutions only use one sensor and vibrator. Due to the strong directional nature of the detection signal, this setup can only detect damage in a specific area and cannot achieve comprehensive detection of the support.

[0006] 2. Low accuracy in damage identification. Changes in support deformation and axial pressure also affect the vibration signal, thus impacting the effectiveness of support damage detection. Specifically, when the support only experiences axial pressure changes without actual damage, the characteristics of the vibration signal will change, potentially leading to a misinterpretation that the support has been damaged.

[0007] 3. Difficulty in identifying the specific type and extent of damage. While forward analysis and feature extraction of vibration signals can identify changes in the detected signal and indicate a change in the support's condition, it is difficult to determine the specific type and extent of damage to the support. Summary of the Invention

[0008] To address the problems existing in the prior art, the purpose of this invention is to provide an intelligent plate rubber bearing and its identification method that can realize in-situ damage detection. This invention can accurately identify the axial compression, deformation and damage state of the bearing.

[0009] To achieve the above objectives, the technical solution adopted by the present invention is: an intelligent plate-type rubber bearing capable of in-situ damage detection, comprising, from top to bottom, a top plate, a sliding plate, a stainless steel cold-rolled plate, a middle steel plate, a rubber plate, and a bottom plate. The top and bottom plates are grooved to accommodate sensors, a vibrator, and sensing wires. The vibrator is connected to an external signal excitation module via sensing wires, and the sensors are connected to an external signal acquisition module via sensing wires. The signal excitation module and the signal acquisition module are connected to a remote data transmission module, and the acquired detection signals are input to a signal processing module in the cloud for processing, thereby identifying the state of the bearing.

[0010] As a further improvement of the present invention, a thread is provided at the bottom of the slot for placing the sensor and the exciter, and the sensor and the exciter are threadedly connected to the top plate and the bottom plate.

[0011] As a further improvement of the present invention, epoxy resin is used to reinforce the transducer of the sensor at the threaded connection, and the sensor, exciter and sensing wire are wrapped with epoxy resin.

[0012] As a further improvement of the present invention, five sensors and exciters are respectively provided on the top plate and the bottom plate.

[0013] As a further improvement of the present invention, a sliding plate sealing ring is provided on the outer periphery of the sliding plate, a rubber sealing ring is provided on the outer periphery of the middle steel plate, and a brass sealing ring is provided on the top outer periphery of the rubber plate.

[0014] The present invention also provides a method for identifying intelligent plate rubber bearings capable of in-situ damage detection as described above, comprising the following steps:

[0015] Step 1: Perform a calibration forward test on the support to obtain the detection signal signal(k,l,m,n) under different states, where (k,l,m,n) represents the state of axial compression, deformation, peeling damage and aging damage of the support; the dimension of the detection signal signal(k,l,m,n) is [i,j,length], where i represents the number of exciters, j represents the number of sensors, and length represents the length of the detection signal;

[0016] Step 2: Perform Fourier transform and wavelet transform on the detected signals under different states to obtain the amplitude spectrum matrix FFT(k,l,m,n) and the wavelet time-frequency plot matrix wavelet(k,l,m,n) of the detected signals. The dimension of the amplitude spectrum matrix is ​​[i,j,F_length], where F_length is the number of points in the Fourier transform, and the dimension of the wavelet time-frequency plot matrix is ​​[i,j,num_wave,length], where num_wave represents the number and frequency range of wavelet envelopes.

[0017] Step 3: Establish a two-stage deep learning network prediction model DM with multiple inputs and multiple outputs. Based on the state of the detected signal, the model identifies the damage and normal use state of the structure. The inputs of the network prediction model DM are the detected signal signal(k,l,m,n), the amplitude spectrum matrix FFT(k,l,m,n), and the wavelet time-frequency plot matrix wavelet(k,l,m,n).

[0018] Step 4: Take multi-stage measures to identify the state of the support, and use the identification results as feature inputs to supplement the identification of subsequent states.

[0019] As a further improvement of the present invention, the calibration positive test experiment of the support in step 1 is as follows:

[0020] Axial compression tests were conducted on the support, with staged loading ranging from 0 MPa to the support's ultimate axial compression. Each stage of loading involved an amplitude of 1 MPa. At the end of each stage, the axial compression was kept constant, and multiple exciters were sequentially excited to obtain the sensor signals. The resulting signal matrix is ​​as follows:

[0021]

[0022] Among them AD ij The detection signal is represented by exciter i and sensor j, obtained from the axial pressure test. When the axial pressure is kMPa, the signal matrix of the detection signal is AD(k).

[0023] Deformation tests were conducted on the support under constant design axial compression. The tests involved staged loading, with the deformation range from 0% to the ultimate shear strain, and each loading increment representing 10% of the shear strain. At the end of each loading stage, the deformation was kept constant, and multiple vibrators were sequentially excited to obtain the sensor signals. The resulting signal matrix is ​​as follows:

[0024]

[0025] Among them SD ij The detection signal obtained from the deformation detection test when the exciter is i and the sensor is j; when the deformation is l%, the signal matrix of the detection signal is SD(l);

[0026] A peel damage test was conducted on the bearing, and tests were performed on bearings with different degrees of damage. Multiple exciters were excited sequentially, and the sensor acquisition signals were obtained. The resulting signal matrix is ​​as follows:

[0027]

[0028] RDD ij The detection signal obtained from the peel damage detection test when the exciter is i and the sensor is j; when the peel damage area is m%, the signal matrix of the detection signal is RDD(m);

[0029] Aging damage tests were conducted on the supports, and tests were performed on supports with different degrees of damage. Multiple exciters were sequentially excited, and the sensor acquisition signals were obtained. The resulting signal matrix is ​​as follows:

[0030]

[0031] ADD ij The detection signal obtained from the peel damage detection test when the exciter is i and the sensor is j is represented by ADD(n); when the degree of aging damage is n%, the signal matrix of the detection signal is ADD(n).

[0032] As a further improvement of the present invention, in step 1, the detection signal signal(k,l,m,n) is specifically as follows:

[0033]

[0034] As a further improvement to the present invention, step 3 is specifically as follows:

[0035] The detection signal signal(k,l,m,n) and the amplitude spectrum matrix FFT(k,l,m,n) are merged to form a new feature vector vector1 with dimensions [i,j,length+F_length]. The feature vector vector1 and the wavelet time-frequency plot matrix are processed by a one-dimensional convolutional neural network and a two-dimensional convolutional neural network, respectively. Convolution, pooling and fully connected operations are used to recombine the signals to form a new fused feature vector vector2.

[0036] As a further improvement of the present invention, step 4 is specifically as follows:

[0037] The network prediction model DM includes sub-models SDM1, SDM2, SDM3, SDM4, and SDM5. Sub-model SDM1 is a multi-dimensional signal input processing module that processes three types of signal inputs in a unified manner and outputs a fused feature vector. Its inputs are the detection signal (k,l,m,n), the amplitude spectrum matrix FFT (k,l,m,n), and the wavelet time-frequency plot matrix wavelet (k,l,m,n), and its output is the feature vector vector2. Sub-model SDM2 is a support peeling damage identification model, and its input is the feature vector vector2. The output is the peel damage identification result of the support; sub-model SDM3 is the aging damage identification model of the support, its input is feature vector vector2 and the peel damage identification result, and the output is the aging damage identification result of the support; sub-model SDM4 is the deformation identification model of the support, its input is feature vector vector2, peel damage identification result and aging damage identification result, and the output is the shear deformation of the support; sub-model SDM5 is the axial compression identification model of the support, its input is feature vector vector2, peel damage identification result, aging damage identification result and deformation identification result, and the output is the axial compression identification result of the support.

[0038] The beneficial effects of this invention are:

[0039] 1. The present invention has multiple sensors and vibrators set in the center and around the support, which can realize comprehensive detection of internal damage of the support.

[0040] 2. This invention obtains the characteristic change law of the detection signal of the support under axial compression, deformation and damage by conducting calibration tests on the support in advance. Based on the measured signal, a deep learning prediction model is established, which can accurately identify the axial compression, deformation and damage state of the support, and can realize comprehensive detection of the axial compression, deformation and damage type and degree of the support. Attached Figure Description

[0041] Figure 1 This is a schematic diagram of the support structure in an embodiment of the present invention;

[0042] Figure 2 for Figure 1 Detail A structural diagram;

[0043] Figure 3 This is a flowchart illustrating the establishment of a two-stage deep learning network prediction model DM in an embodiment of the present invention.

[0044] Figure 4 This is a flowchart illustrating how the deep learning network prediction model DM identifies supports in an embodiment of the present invention.

[0045] Figure label:

[0046] 1. Top plate, 2. Slide plate sealing ring, 3. Stainless steel cold-rolled steel plate, 4. Slide plate, 5. Middle steel plate, 6. Rubber sealing ring, 7. Brass sealing ring, 8. Bottom plate, 9. Rubber plate. Detailed Implementation

[0047] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0048] Example

[0049] like Figure 1 As shown, an intelligent plate-type rubber bearing capable of in-situ damage detection includes, from top to bottom, a top plate 1, a sliding plate 4, a stainless steel cold-rolled plate 3, a middle steel plate 5, a rubber plate 9, and a bottom plate 8. Grooves are cut into the top plate 1 and the bottom plate 8 to house sensors, vibrators, and sensing wires. The vibrator is connected to an external signal excitation module via sensing wires, and the sensors are connected to an external signal acquisition module via sensing wires. The signal excitation module and the signal acquisition module are connected to a remote data transmission module, enabling online signal excitation and acquisition. The acquired detection signals are input to a cloud-based signal processing module for processing, thereby identifying the bearing's state to determine its axial compression, deformation, and damage status.

[0050] like Figure 2 As shown, in this embodiment, a thread is provided at the bottom of the slot for placing the sensor and the exciter, and the sensor and the exciter are threadedly connected to the top plate 1 and the bottom plate 8.

[0051] In this embodiment, epoxy resin is used to reinforce the transducer of the sensor at the threaded connection to prevent it from loosening during vibration. The sensor, exciter, and sensor wire are also wrapped with epoxy resin to enhance their waterproof and moisture-proof capabilities. Five sensors and exciters are respectively installed on the upper and lower parts of the support.

[0052] In this embodiment, a sliding plate sealing ring 2 is provided on the periphery of the sliding plate 4, a rubber sealing ring 6 is provided on the periphery of the middle steel plate 5, and a brass sealing ring 7 is provided on the top periphery of the rubber plate 9.

[0053] This embodiment also provides a method for identifying intelligent plate rubber bearings that can achieve in-situ damage detection, as described above. First, calibration testing is required to establish a bearing state identification model. The algorithm establishment process is as follows:

[0054] Axial compression tests were conducted on the aforementioned supports. The axial compression tests were performed with staged loading, ranging from 0 MPa to the support's ultimate axial compression, with each loading amplitude being 1 MPa. At the end of each loading stage, the axial compression was kept constant, and five exciters were sequentially excited to obtain the sensor's acquired signals. The resulting signal matrix is ​​as follows:

[0055]

[0056] Among them AD ij The detection signal obtained from the axial compression test when the exciter is i and the sensor is j.

[0057] When the axial pressure is kMPa, the signal matrix of the detected signal is AD(k).

[0058] Deformation tests were conducted on the aforementioned supports under constant design axial compression. The tests involved staged loading, with the deformation range from 0% to the ultimate shear strain, and each loading increment representing 10% of the shear strain. At the end of each loading stage, the deformation was kept constant, and five vibrators were sequentially excited to obtain the sensor signals. The resulting signal matrix is ​​as follows:

[0059]

[0060] Among them SD ij The detection signal obtained from the deformation detection test when the exciter is i and the sensor is j represents the signal obtained from the deformation detection test.

[0061] When the deformation is l%, the signal matrix of the detected signal is SD(l).

[0062] Peel damage tests were conducted on the aforementioned supports. The degree of peel damage was divided into 10 levels according to the damaged area. Tests were performed on supports with different damage levels, and five exciters were sequentially excited to obtain the sensor acquisition signals. The resulting signal matrix is ​​as follows:

[0063]

[0064] RDD ij The detection signal obtained from the peel damage detection test when the exciter is i and the sensor is j represents the exciter.

[0065] When the area of ​​peel damage is m%, the signal matrix of the detection signal is RDD(m).

[0066] Aging damage tests were conducted on the aforementioned supports, with the degree of aging damage divided into 10 levels. Tests were performed on supports with different damage levels, sequentially exciting five vibrators to obtain the sensor acquisition signals. The resulting signal matrix is ​​as follows:

[0067]

[0068] ADD ij The detection signal obtained from the peel damage detection test when the exciter is i and the sensor is j represents the exciter.

[0069] When the degree of aging damage is n%, the signal matrix of the detected signal is ADD(n).

[0070] Through the above forward testing experiments, the detection signal signal(k,l,m,n) under different states was obtained, where (k,l,m,n) represents the state of axial compression, deformation, peeling damage, and aging damage of the support. The dimension of the detection signal signal(k,l,m,n) is [5,5,length], where length represents the length of the detection signal.

[0071]

[0072] Fourier transform and wavelet transform are performed on the detected signals under different states to obtain the amplitude spectrum matrix FFT(k,l,m,n) and the wavelet time-frequency plot matrix wavelet(k,l,m,n). The amplitude spectrum matrix has dimensions [5,5,F_length], where F_length is the number of points in the FFT. The wavelet time-frequency plot matrix has dimensions [5,5,num_wave,length], where num_wave represents the number and frequency range of wavelet envelopes.

[0073] A two-stage deep learning network prediction model DM with multiple inputs and multiple outputs is established to identify structural damage and normal use status based on the state of the detected signals. Figure 3As shown, the model's input consists of the detection signal signal(k,l,m,n), the amplitude spectrum matrix FFT(k,l,m,n), and the wavelet time-frequency plot matrix wavelet(k,l,m,n). Since the model's input comes from three types of signals, processing is required for accurate signal extraction. The basic elements of both the detection signal and its amplitude spectrum matrix are one-dimensional vectors; therefore, these two vectors are merged to form a new feature vector vector1 with dimensions [5,5,length+F_length]. The basic elements of the wavelet time-frequency plot are two-dimensional images and cannot be directly merged with a one-dimensional vector. Therefore, a one-dimensional convolutional neural network and a two-dimensional convolutional neural network are used to process the feature vector vector1 and the wavelet time-frequency plot, respectively. Convolution, pooling, and fully connected operations are used to recombine the signals into a new fused feature vector vector2.

[0074] The deep learning network prediction model DM has four output dimensions: axial compression, deformation, peel damage, and aging damage. Since different states have varying degrees of influence on the bearing's detection signal, the order of influence is peel damage > aging damage > deformation > axial compression. Therefore, identifying all four states of the bearing in parallel is difficult to obtain accurate results. To address this issue, a multi-stage approach is needed to identify the bearing's states and use the identified results as feature inputs to supplement the identification of subsequent states.

[0075] In summary, such as Figure 4As shown, the neural network prediction model DM is divided into five sub-models SDM1-5. Sub-model SDM1 is a multi-dimensional signal input processing module that processes the three types of signal inputs in a unified manner and outputs a fused feature vector. Its inputs are the detection signal signal(k,l,m,n), the amplitude spectrum matrix FFT(k,l,m,n), and the wavelet time-frequency plot matrix wavelet(k,l,m,n), and its output is the feature vector vector2. Sub-model SDM2 is a support peeling damage identification model. Its input is the feature vector vector2, and its output is the support peeling damage identification. The results are as follows: Sub-model SDM3 is the aging damage identification model for the support, with inputs being the feature vector vector2 and the identification results of peeling damage, and outputting the aging damage identification result of the support; Sub-model SDM4 is the deformation identification model for the support, with inputs being the feature vector vector2, the peeling damage identification result, and the aging damage identification result, and outputting the shear deformation of the support; Sub-model SDM5 is the axial compression identification model for the support, with inputs being the feature vector vector2, the peeling damage identification result, the aging damage identification result, and the deformation identification result, and outputting the axial compression identification result of the support. Through the multi-stage identification results, the coupling influence of each state of the support on the detection signal is effectively distinguished, realizing multi-state identification of the support under limited data.

[0076] The DM model can predict various states of the support based on the actual detection signals. Because the dataset contains multiple data types, the trained model is robust and can predict various states of the support, such as axial compression, deformation, peeling, and aging. Furthermore, data can be supplemented in real time during support operation, further enhancing model training and increasing prediction accuracy.

[0077] The embodiments described above are merely illustrative of specific implementations of the present invention, and while the descriptions are detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.

Claims

1. A method for identifying the bearing status of an intelligent plate rubber bearing capable of in-situ damage detection, characterized in that, The rubber support comprises, from top to bottom, a top plate, a sliding plate, a stainless steel cold-rolled plate, a middle steel plate, a rubber plate, and a bottom plate. The top and bottom plates are grooved to accommodate sensors, vibrators, and sensing wires. The vibrator is connected to an external signal excitation module via sensing wires, and the sensors are connected to an external signal acquisition module via sensing wires. The signal excitation module and the signal acquisition module are connected to a remote data transmission module, which inputs the acquired detection signals to a cloud-based signal processing module for processing, thereby identifying the support's status. The identification method includes the following steps: Step 1: Conduct a calibration test on the support to obtain the detection signals under different conditions. Where k, l, m, and n represent the axial compression, deformation, peeling damage, and aging damage states of the support, respectively; detection signals The dimension is [i,j,length], where i represents the number of exciters, j represents the number of sensors, and length represents the length of the detected signal; Step 2: Perform Fourier transform and wavelet transform on the detection signals under different states to obtain the amplitude spectrum matrix of the detection signals. Wavelet time-frequency plot matrix The magnitude spectrum matrix has dimensions [i,j,F_length], where F_length is the number of points in the FFT during Fourier transform, and the wavelet time-frequency plot matrix has dimensions [i,j,num_wave,W_length], where num_wave represents the number of wavelet envelopes and the frequency range. Step 3: Establish a two-stage deep learning network prediction model DM with multiple inputs and multiple outputs. This model identifies structural damage and normal operating conditions based on the state of the detection signal. The input to the network prediction model DM is the detection signal. Amplitude spectral matrix Wavelet time-frequency plot matrix ; Step 3 is described in detail below: Detect signal and amplitude spectrum matrix The two vectors are merged to form a new feature vector vector1 with dimensions [i,j,length+F_length]. The feature vector vector1 and the wavelet time-frequency plot matrix are processed by a one-dimensional convolutional neural network and a two-dimensional convolutional neural network, respectively. Convolution, pooling and fully connected operations are used to recombine the signals into a new fused feature vector vector2. Step 4: Take multi-stage measures to identify the state of the support, and use the identification results as feature inputs to supplement the identification of subsequent states; Step 4 is described in detail below: The network prediction model DM includes sub-models SDM1, SDM2, SDM3, SDM4, and SDM5. Sub-model SDM1 is a multi-dimensional signal input processing module that processes the three types of signal inputs in a unified manner and outputs a fused feature vector. Its input is the detection signal. Amplitude spectral matrix Wavelet time-frequency plot matrix The sub-model SDM2 is the support peel damage identification model, whose input is the feature vector vector2 and the output is the support peel damage identification result; the sub-model SDM3 is the support aging damage identification model, whose input is the feature vector vector2 and the peel damage identification result, and the output is the support aging damage identification result; the sub-model SDM4 is the support deformation identification model, whose input is the feature vector vector2, the peel damage identification result and the aging damage identification result, and the output is the support shear deformation; the sub-model SDM5 is the support axial compression identification model, whose input is the feature vector vector2, the peel damage identification result, the aging damage identification result and the deformation identification result, and the output is the support axial compression identification result.

2. The method for identifying the bearing status of an intelligent plate rubber bearing capable of in-situ damage detection according to claim 1, characterized in that, In step 1, the calibration positive test experiment for the support is carried out as follows: Axial compression tests were conducted on the support, with staged loading ranging from 0 MPa to the support's ultimate axial compression. Each stage of loading involved an amplitude of 1 MPa. At the end of each stage, the axial compression was kept constant, and multiple exciters were sequentially excited to obtain the sensor signals. The resulting signal matrix is ​​as follows: ； in The detection signal is obtained from the axial pressure test when the exciter is i and the sensor is j; the signal matrix of the detection signal is as follows when the axial pressure is kMPa. ; Deformation tests were conducted on the support under constant design axial compression. The tests involved staged loading, with the deformation range from 0% to the ultimate shear strain, and each loading increment representing 10% of the shear strain. At the end of each loading stage, the deformation was kept constant, and multiple vibrators were sequentially excited to obtain the sensor signals. The resulting signal matrix is ​​as follows: ； in The detection signal obtained from the deformation detection test when the exciter is i and the sensor is j; the signal matrix of the detection signal when the deformation is l% is... ; A peel damage test was conducted on the bearing, and tests were performed on bearings with different degrees of damage. Multiple exciters were excited sequentially, and the sensor acquisition signals were obtained. The resulting signal matrix is ​​as follows: ； in The detection signal obtained from the peel damage detection test when the exciter is i and the sensor is j; the signal matrix of the detection signal when the peel damage area is m% is... ; Aging damage tests were conducted on the supports, and tests were performed on supports with different degrees of damage. Multiple exciters were sequentially excited, and the sensor acquisition signals were obtained. The resulting signal matrix is ​​as follows: ； in The detection signal obtained from the peel damage detection test when the exciter is i and the sensor is j; the signal matrix of the detection signal when the degree of aging damage is n% is... .

3. The method for identifying the bearing status of an intelligent plate rubber bearing capable of in-situ damage detection according to claim 2, characterized in that, In step 1, the detection signal Specifically as follows: 。 4. The method for identifying the bearing status of an intelligent plate rubber bearing capable of in-situ damage detection according to claim 1, characterized in that, A thread is provided at the bottom of the slot used to place the sensor and the vibrator, and the sensor and the vibrator are threadedly connected to the top plate and the bottom plate.

5. The method for identifying the bearing status of an intelligent plate rubber bearing capable of in-situ damage detection according to claim 4, characterized in that, The transducer of the sensor is reinforced with epoxy resin at the threaded connection, and the sensor, exciter, and sensing wire are wrapped with epoxy resin.

6. The method for identifying the bearing status of an intelligent plate rubber bearing capable of in-situ damage detection according to claim 5, characterized in that, Five sensors and vibrators are installed on the top plate and the bottom plate respectively.

7. The method for identifying the bearing status of an intelligent plate rubber bearing capable of in-situ damage detection according to claim 1, characterized in that, A skateboard sealing ring is installed on the outer perimeter of the skateboard, a rubber sealing ring is installed on the outer perimeter of the middle steel plate, and a brass sealing ring is installed on the top perimeter of the rubber plate.

Images