Device and method for realizing contact sliding quick detection by using PVDF piezoelectric sensor

Abstract

The application provides a device and method for realizing quick detection of contact sliding by using a PVDF piezoelectric sensor, which comprises a PVDF piezoelectric sensor, a signal acquisition module and a controller, wherein the controller stores a sliding detection and recognition algorithm; the PVDF piezoelectric sensor is connected with the controller through the signal acquisition module; the signal acquisition module converts the electric signal of the PVDF piezoelectric sensor into a voltage signal and converts the voltage signal into a digital signal processed by the controller; and the controller executes quick detection of contact sliding by using the sliding detection and recognition algorithm; the application can realize quick detection of contact sliding by using a neural network algorithm based on the PVDF piezoelectric sensor, and solve the problems of slow detection speed and low accuracy rate of the detection operation in the traditional technology.

Description

Technical Field

[0001] The present invention relates to the field of robot tactile perception, and in particular to a device and method for realizing rapid detection of contact sliding using a PVDF piezoelectric sensor. Background Art

[0002] Tactile sensors provide feedback on tactile information, such as force, temperature, and slip, to aid in decision-making. Slip detection helps the robot identify the contact state between the gripper and the workpiece, dynamically adjusting the gripping force to successfully complete the grasping task. However, robots lack inherent tactile perception and require the aid of tactile sensors to detect slippage. Generally, the response time for human gripping force adjustment is approximately 75ms, so slip detection requires rapid speed.

[0003] Currently, there are methods for detecting slippage, such as patent number CN117629386A, which describes an initial slippage detection method for a manipulator that does not require a contact sensor. This invention discloses a slippage detection method, specifically as follows: a vibration measurement unit is installed on the back or side of the manipulator's fingers where they are not in contact with the grasped object to detect the vibration during initial slippage. Furthermore, by establishing a vibration model for the manipulator's grip, slippage detection based on contact force changes is converted to slippage detection based on two-dimensional vibration. Slippage detection is then achieved by further determining a slippage threshold. However, this method has certain limitations. The measurement of vibration signals requires the use of a displacement sensor, velocity sensor, acceleration sensor, gyroscope, and inertial measurement unit. The signal acquisition and processing of multiple sensing elements takes a considerable amount of time, resulting in a slow slippage detection speed. The slippage threshold is determined based on a correlation coefficient when there is no slippage, and there are multiple methods for determining it. The setting of the slippage threshold affects the accuracy of slippage detection. The present invention is based on a PVDF piezoelectric sensor and utilizes a neural network algorithm to achieve rapid detection of contact slippage. Summary of the Invention

[0004] The present invention proposes a device and method for realizing rapid detection of contact sliding using PVDF piezoelectric sensors. Based on the PVDF piezoelectric sensors, the invention can realize rapid detection of contact sliding using a neural network algorithm, thereby solving the problems of slow detection speed and low accuracy of such detection operations under traditional technologies.

[0005] The present invention adopts the following technical solutions.

[0006] A device for realizing rapid contact slip detection using a PVDF piezoelectric sensor includes a PVDF piezoelectric sensor, a signal acquisition module, and a controller, wherein the controller stores a slip detection and recognition algorithm; the PVDF piezoelectric sensor is connected to the controller via the signal acquisition module; the signal acquisition module converts the electrical signal of the PVDF piezoelectric sensor into a voltage signal, and then converts the voltage signal into a digital signal processed by the controller; the controller performs rapid contact slip detection using the slip detection and recognition algorithm.

[0007] The signal acquisition module includes a charge-voltage conversion module and an analog-to-digital conversion module connected thereto.

[0008] When the charge-to-voltage conversion module converts the electrical signal of the PVDF piezoelectric sensor into a voltage signal, the sensor is regarded as an equivalent circuit of a charge source connected in parallel with a capacitor and a resistor, and the charge-to-voltage conversion is achieved in the form of an integration circuit;

[0009] The analog-to-digital conversion module includes a development board, which collects the voltage signal output by the charge-to-voltage conversion module through the analog input pin, converts the voltage signal into a digital signal, and sends the digital signal to the controller through the communication interface.

[0010] The slip detection and recognition algorithm includes a neural network model for recognizing slip signals.

[0011] A method for realizing rapid detection of contact slip using a PVDF piezoelectric sensor, using the above-mentioned device for realizing rapid detection of contact slip using a PVDF piezoelectric sensor, wherein the method is based on the PVDF piezoelectric sensor and realizes rapid detection of contact slip using a neural network algorithm including a neural network model;

[0012] The controller runs the sliding detection and recognition algorithm to implement the following steps for fast contact sliding detection:

[0013] Step S1: The controller loads the trained neural network model when starting the program;

[0014] Step S2: the controller establishes communication with the signal acquisition module;

[0015] Step S3: the electrical signal of the PVDF piezoelectric sensor is converted by the signal acquisition module and acquired by the controller to obtain a digital signal;

[0016] Step S4: After the controller collects the sensor signal for one cycle, it immediately calculates the time-frequency domain eigenvalues ​​of the signal within the cycle, uses the eigenvalues ​​as input to the neural network model, obtains the sliding detection and recognition results, and realizes rapid detection of contact sliding.

[0017] The neural network model establishment process includes the following steps:

[0018] Step A1: collecting a large number of sensor signals in two contact states, namely, contact sliding and contact non-sliding, to form a sensor signal data set;

[0019] Step A2: performing noise reduction processing on the collected sensor signals one by one; the sensor signals are time series, and the signal noise reduction is achieved using the VMD decomposition and reconstruction method;

[0020] Step A3: extracting signal features and constructing a signal feature set, extracting multiple time domain and multiple frequency domain signal features for characterizing the signal features of the PVDF piezoelectric sensor from the noise-reduced signal obtained in step A2;

[0021] Step A4: Optimize the parameters of the neural network model. Use the Dung Beetle Optimization (DBO) algorithm to optimize the parameters of the Long Short-Term Memory (LSTM) network to improve the convergence speed and recognition accuracy of the neural network. Use the signal feature set obtained in step A3 as the training set and test set of the neural network. Use the error rate of neural network recognition as the fitness function. Obtain the optimal parameter value through update iteration.

[0022] Step A5: Construct a neural network model using the signal feature set obtained in step A3 and the LSTM network parameters optimized in step A4.

[0023] The specific method of signal denoising in step A2 is as follows: first, the Archimedes Optimization Algorithm (AOA) is used to optimize the variational mode decomposition (VMD) parameters [K, α]. The minimum envelope entropy is used as the fitness function during the optimization process. When the Archimedes optimization algorithm is initialized, each population object contains [K, α]. The [K, α] corresponding to each object is used to perform VMD decomposition on the sensor signal to obtain the minimum envelope entropy. The minimum envelope entropy is the fitness value, and the [K, α] corresponding to the object with the smallest fitness value is selected as the optimal object. During each iteration, the [K, α] corresponding to each population object is updated, and then the fitness value of each population object is recalculated. If the fitness value of the new object is smaller than the fitness value of the current best object, the best object is updated to the [K, α] corresponding to the new object. At the end of the iteration, the [K, α] corresponding to the best object is the optimal parameter obtained by optimization. The optimal parameters are obtained through iterative updating, and then the sensor signal is subjected to VMD decomposition using the optimized optimal parameters. According to the correlation coefficient values ​​between each modal component obtained by VMD decomposition and the original signal, the modal components with correlation coefficient values ​​greater than 0.2 are selected to reconstruct the sensor signal to obtain the denoised signal.

[0024] In step A3, 12 time domain and 4 frequency domain signal features are extracted, namely peak-to-peak value F1, average value F2, absolute average value F3, variance F4, standard deviation F5, root mean square F6, kurtosis F7, skewness F8, shape factor F9, peak factor F 10 , pulse factor F 11 , margin factor F 12 , average frequency F 13 , center frequency F 14 , frequency root mean square F 15 , frequency standard deviation F 16 The 16 eigenvalue calculation formulas are shown in formulas (1) to (16) in the following table:

[0025]

[0026] Where: x(n) represents the time series of the signal, n = 1, 2, ..., N, N represents the number of sample points; through discrete Fourier transform, the time series is converted into a spectrum, X k The spectrum of the signal is shown, k=0,1,…,N-1, f k =k / N×fs, fs represents the sampling frequency of the signal.

[0027] The time domain and frequency domain signal features extracted in step A3 are conventional expression indicators of mechanical vibration signals. When performing contact sliding rapid detection, the detection surface of the PVDF piezoelectric sensor is in direct contact with the object to be detected.

[0028] Compared with the existing technology, the present invention is based on a PVDF piezoelectric sensor and uses a neural network algorithm to achieve rapid detection of contact sliding, which has the advantages of fast detection speed and high accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0030] Attachment Figure 1 This is a schematic diagram of the direct contact between the sensor surface and the object provided by the present invention;

[0031] Attachment Figure 2 Schematic diagram of the contact sliding detection device provided by the present invention;

[0032] Attachment Figure 3 This is a flow chart of the steps for implementing rapid contact sliding detection provided by the present invention;

[0033] Attachment Figure 4 It is a flowchart of establishing a neural network for identifying sliding signals provided by the present invention;

[0034] In the picture:

[0035] 1. PVDF piezoelectric sensor; 2. Contact object; 3. Signal acquisition module; 4. Controller. DETAILED DESCRIPTION

[0036] To make the features and advantages of this patent more clearly understood, the following embodiments are specifically described in detail as follows:

[0037] It should be noted that the following detailed description is illustrative and is intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used in this specification have the same meaning as commonly understood by those skilled in the art to which this application belongs.

[0038] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0039] The following is a further detailed description of this embodiment with reference to the accompanying drawings:

[0040] As shown in the figure, the device for realizing rapid detection of contact sliding using a PVDF piezoelectric sensor includes a PVDF piezoelectric sensor, a signal acquisition module, and a controller, wherein the controller stores a sliding detection and recognition algorithm; the PVDF piezoelectric sensor is connected to the controller via the signal acquisition module; the signal acquisition module converts the electrical signal of the PVDF piezoelectric sensor into a voltage signal, and converts the voltage signal into a digital signal processed by the controller; the controller performs rapid detection of contact sliding using the sliding detection and recognition algorithm.

[0041] The signal acquisition module includes a charge-voltage conversion module and an analog-to-digital conversion module connected thereto.

[0042] When the charge-to-voltage conversion module converts the electrical signal of the PVDF piezoelectric sensor into a voltage signal, the sensor is regarded as an equivalent circuit of a charge source connected in parallel with a capacitor and a resistor, and the charge-to-voltage conversion is achieved in the form of an integration circuit;

[0043] The analog-to-digital conversion module includes a development board, which collects the voltage signal output by the charge-to-voltage conversion module through the analog input pin, converts the voltage signal into a digital signal, and sends the digital signal to the controller through the communication interface.

[0044] The slip detection and recognition algorithm includes a neural network model for recognizing slip signals.

[0045] A method for realizing rapid detection of contact slip using a PVDF piezoelectric sensor, using the above-mentioned device for realizing rapid detection of contact slip using a PVDF piezoelectric sensor, wherein the method is based on the PVDF piezoelectric sensor and realizes rapid detection of contact slip using a neural network algorithm including a neural network model;

[0046] The controller runs the sliding detection and recognition algorithm to implement the following steps for fast contact sliding detection:

[0047] Step S1: The controller loads the trained neural network model when starting the program;

[0048] Step S2: the controller establishes communication with the signal acquisition module;

[0049] Step S3: the electrical signal of the PVDF piezoelectric sensor is converted by the signal acquisition module and acquired by the controller to obtain a digital signal;

[0050] Step S4: After the controller collects the sensor signal for one cycle, it immediately calculates the time-frequency domain eigenvalues ​​of the signal within the cycle, uses the eigenvalues ​​as input to the neural network model, obtains the sliding detection and recognition results, and realizes rapid detection of contact sliding.

[0051] The neural network model establishment process includes the following steps:

[0052] Step A1: collecting a large number of sensor signals in two contact states, namely, contact sliding and contact non-sliding, to form a sensor signal data set;

[0053] Step A2: performing noise reduction processing on the collected sensor signals one by one; the sensor signals are time series, and the signal noise reduction is achieved using the VMD decomposition and reconstruction method;

[0054] Step A3: extracting signal features and constructing a signal feature set, extracting multiple time domain and multiple frequency domain signal features for characterizing the signal features of the PVDF piezoelectric sensor from the noise-reduced signal obtained in step A2;

[0055] Step A4: Optimize the parameters of the neural network model. Use the Dung Beetle Optimization (DBO) algorithm to optimize the parameters of the Long Short-Term Memory (LSTM) network to improve the convergence speed and recognition accuracy of the neural network. Use the signal feature set obtained in step A3 as the training set and test set of the neural network. Use the error rate of neural network recognition as the fitness function. Obtain the optimal parameter value through update iteration.

[0056] Step A5: Construct a neural network model using the signal feature set obtained in step A3 and the LSTM network parameters optimized in step A4.

[0057] The specific method of signal denoising in step A2 is as follows: first, the Archimedes Optimization Algorithm (AOA) is used to optimize the variational mode decomposition (VMD) parameters [K, α]. The minimum envelope entropy is used as the fitness function during the optimization process. When the Archimedes optimization algorithm is initialized, each population object contains [K, α]. The [K, α] corresponding to each object is used to perform VMD decomposition on the sensor signal to obtain the minimum envelope entropy. The minimum envelope entropy is the fitness value, and the [K, α] corresponding to the object with the smallest fitness value is selected as the optimal object. During each iteration, the [K, α] corresponding to each population object is updated, and then the fitness value of each population object is recalculated. If the fitness value of the new object is smaller than the fitness value of the current best object, the best object is updated to the [K, α] corresponding to the new object. At the end of the iteration, the [K, α] corresponding to the best object is the optimal parameter obtained by optimization. The optimal parameters are obtained through iterative updating, and then the sensor signal is subjected to VMD decomposition using the optimized optimal parameters. According to the correlation coefficient values ​​between each modal component obtained by VMD decomposition and the original signal, the modal components with correlation coefficient values ​​greater than 0.2 are selected to reconstruct the sensor signal to obtain the denoised signal.

[0058] In step A3, 12 time domain and 4 frequency domain signal features are extracted, namely peak-to-peak value F1, average value F2, absolute average value F3, variance F4, standard deviation F5, root mean square F6, kurtosis F7, skewness F8, shape factor F9, peak factor F 10 , pulse factor F 11 , margin factor F 12 , average frequency F 13 , center frequency F 14 , frequency root mean square F 15 , frequency standard deviation F 16 The 16 eigenvalue calculation formulas are shown in formulas (1) to (16) in the following table:

[0059]

[0060]

[0061] Where: x(n) represents the time series of the signal, n = 1, 2, ..., N, N represents the number of sample points; through discrete Fourier transform, the time series is converted into a spectrum, X k The spectrum of the signal is shown, k=0,1,…,N-1, f k =k / N×fs, fs represents the sampling frequency of the signal.

[0062] The time domain and frequency domain signal features extracted in step A3 are conventional expression indicators of mechanical vibration signals. When performing contact sliding rapid detection, the detection surface of the PVDF piezoelectric sensor is in direct contact with the object to be detected.

[0063] Example:

[0064] The following is a further detailed description of this embodiment with reference to the accompanying drawings:

[0065] This embodiment provides a device and method for realizing rapid detection of contact slip using a PVDF piezoelectric sensor.

[0066] like Figure 1 As shown, in this embodiment, the surface of the PVDF piezoelectric sensor is in direct contact with the object, including the PVDF piezoelectric sensor 1 and the contact object 2. The contact object 2 applies pressure to the surface of the PVDF piezoelectric sensor 1. When the contact object 2 and the surface of the PVDF piezoelectric sensor 1 do not slide relative to each other, the sensor signal collected is a contact non-sliding signal. When the contact object 2 and the surface of the PVDF piezoelectric sensor 1 slide relative to each other, the sensor signal collected is a contact sliding signal.

[0067] like Figure 2 As shown, the present invention provides a device for rapid contact slip detection using a PVDF piezoelectric sensor. The device comprises a PVDF piezoelectric sensor 1, a signal acquisition module 3, and a controller 4, wherein the controller includes a slip detection and recognition algorithm. The PVDF piezoelectric sensor 1 provides tactile feedback in the form of an electrical signal. The signal acquisition module 3 converts the sensor's electrical signal into a voltage signal, which is then converted into a digital signal that can be processed by the controller. The controller 4 is a laptop computer, which runs the slip detection and recognition algorithm to achieve rapid contact slip detection.

[0068] like Figure 3 As shown in FIG, the present invention provides a flowchart of the steps for implementing rapid contact sliding detection, and the specific steps are as follows:

[0069] Step 1: The controller loads the trained neural network model when starting the program;

[0070] Step 2: The controller establishes communication with the signal acquisition module;

[0071] Step 3: The electrical signal of the PVDF piezoelectric sensor is converted by the signal acquisition module and acquired by the controller to obtain a digital signal;

[0072] Step 4: After the controller collects the sensor signal for one cycle, it immediately calculates the time-frequency domain eigenvalues ​​of the signal within the cycle, uses the eigenvalues ​​as the input of the neural network model, obtains the sliding detection and recognition results, and realizes the rapid detection of contact sliding.

[0073] The invention is characterized in that, based on a PVDF piezoelectric sensor, a neural network algorithm is used to realize rapid detection of contact sliding.

[0074] Furthermore, the signal acquisition module includes two submodules: a charge-to-voltage conversion module and an analog-to-digital conversion module. The charge-to-voltage conversion module converts the electrical signal of the PVDF piezoelectric sensor 1 into a voltage signal by treating the sensor as an equivalent circuit consisting of a charge source in parallel with a capacitor and a resistor, and implementing charge-to-voltage conversion in the form of an integration circuit. The analog-to-digital conversion module uses an Arduino UNO development board to acquire the voltage signal output by the charge-to-voltage conversion module through the analog input pins of the Arduino UNO development board, converts the voltage signal into a digital signal, and sends the digital signal to the controller via the communication interface.

[0075] Furthermore, if Figure 4 As shown in the flowchart of establishing a neural network for identifying sliding signals provided by the present invention, the steps of establishing a neural network model are as follows:

[0076] Step 1: Collect a large number of sensor signals in two contact states: contact sliding and contact non-sliding to form a sensor signal data set.

[0077] Step 2: De-noise the sensor signals one by one. The collected sensor digital signals are time series. The VMD decomposition and reconstruction method is used to achieve signal noise reduction. The specific steps are as follows: First, the Archimedes Optimization Algorithm (AOA) is used to optimize the variational mode decomposition (VMD). The parameters [K, α] of the VMD (Mode Decomposition) algorithm are determined. Minimum envelope entropy is used as the fitness function during the optimization process. Each population object in the Archimedean optimization algorithm is initialized with [K, α]. VMD decomposition is performed on the sensor signal using the [K, α] corresponding to each object to determine the minimum envelope entropy, which is the fitness value. The [K, α] corresponding to the object with the smallest fitness value is selected as the optimal object. During each iteration, the [K, α] corresponding to each population object is updated, and the fitness value of each population object is recalculated. If the fitness value of the new object is smaller than that of the current optimal object, the optimal object is updated to the [K, α] corresponding to the new object. At the end of the iteration, the [K, α] corresponding to the optimal object is the optimal parameter obtained through optimization. The optimal parameters are then iteratively updated to obtain the optimal parameters. VMD decomposition is then performed on the sensor signal using the optimized parameters. Based on the correlation coefficients between the modal components obtained by VMD decomposition and the original signal, modal components with correlation coefficients greater than 0.2 are selected to reconstruct the sensor signal, resulting in a denoised signal.

[0078] Step 3: Extract signal features and construct a signal feature set. Extract 12 time domain and 4 frequency domain signal features from the noise-reduced signal obtained in step 2. These 16 signal features are commonly used indicators in mechanical vibration signals and can characterize the sensor signal characteristics, namely: peak-to-peak value F1, average value F2, absolute average value F3, variance F4, standard deviation F5, root mean square F6, kurtosis F7, skewness F8, waveform factor F9, peak factor F 10 , pulse factor F 11 , margin factor F 12 , average frequency F 13 , center frequency F 14 , frequency root mean square F 15 , frequency standard deviation F 16 The 16 eigenvalue calculation formulas are shown in the following formulas (1) to (16):

[0079] F1=max[x(n)]-min[x(n)] (Formula 1)

[0080]

[0081]

[0082]

[0083]

[0084]

[0085]

[0086]

[0087] Where: x(n) represents the time series of the signal, n = 1, 2, ..., N, N represents the number of sample points; through discrete Fourier transform, the time series is converted into a spectrum, X k The spectrum of the signal is shown, k=0,1,…,N-1, f k =k / N×fs, fs represents the sampling frequency of the signal.

[0088] Step 4: Optimize the parameters of the neural network model. Use the dung beetle optimization algorithm to optimize the long short-term memory network parameters to improve the convergence speed and recognition accuracy of the neural network. Use the signal feature set obtained in step 3 as the training set and test set of the neural network. Use the error rate of neural network recognition as the fitness function. Obtain the optimal parameter value through update iteration.

[0089] Step 5: Build a neural network model using the signal feature set obtained in step 3 and the LSTM network parameters optimized in step 4.

[0090] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any other manner. Any person skilled in the art may utilize the above-disclosed technical content to modify or modify the present invention into equivalent embodiments. However, any simple modifications, equivalent variations, and modifications to the above embodiments that do not depart from the technical content of the present invention and are based on the technical essence of the present invention remain within the scope of protection of the present invention.

Claims

1. A device for rapid contact and sliding detection using a PVDF piezoelectric sensor, characterized in that: The device comprises a PVDF piezoelectric sensor, a signal acquisition module, and a controller. The controller stores a sliding detection and recognition algorithm. The PVDF piezoelectric sensor is connected to the controller via the signal acquisition module. The signal acquisition module converts the electrical signal of the PVDF piezoelectric sensor into a voltage signal, and then converts the voltage signal into a digital signal processed by the controller. The controller performs contact sliding rapid detection using the sliding detection and recognition algorithm. The slip detection and recognition algorithm includes a neural network model for identifying slip signals; The specific steps to achieve fast contact sliding detection are as follows: Step 1: The controller loads the trained neural network model when starting the program; Step 2: The controller establishes communication with the signal acquisition module; Step 3: The electrical signal of the PVDF piezoelectric sensor is converted by the signal acquisition module and acquired by the controller to obtain a digital signal; Step 4: After the controller collects the sensor signal for one cycle, it immediately calculates the time-frequency domain eigenvalues ​​of the signal within the cycle, uses the eigenvalues ​​as input to the neural network model, obtains the slip detection and recognition results, and realizes rapid detection of contact slip; The neural network model establishment process includes the following steps: Step A1: collecting a large number of sensor signals in two contact states, namely, contact sliding and contact non-sliding, to form a sensor signal data set; Step A2: performing noise reduction processing on the collected sensor signals one by one; the sensor signals are time series, and the signal noise reduction is achieved using the VMD decomposition and reconstruction method; Step A3: extracting signal features and constructing a signal feature set, extracting multiple time domain and multiple frequency domain signal features for characterizing the signal features of the PVDF piezoelectric sensor from the noise-reduced signal obtained in step A2; Step A4: Optimize the parameters of the neural network model. Use the Dung Beetle Optimization (DBO) algorithm to optimize the parameters of the long short-term memory (LSTM) network to improve the convergence speed and recognition accuracy of the neural network. Use the signal feature set obtained in step A3 as the training set and test set of the neural network. Use the error rate of the neural network recognition as the fitness function. Optimal parameter values ​​are obtained through iterative updating. Step A5: Construct a neural network model using the signal feature set obtained in step A3 and the LSTM network parameters optimized in step A4. The time domain and frequency domain signal features extracted in step A3 are conventional expression indicators of mechanical vibration signals. When performing contact sliding rapid detection, the detection surface of the PVDF piezoelectric sensor is in direct contact with the object to be detected.

2. The device for realizing rapid contact and sliding detection using a PVDF piezoelectric sensor according to claim 1, characterized in that: The signal acquisition module includes a charge-voltage conversion module and an analog-to-digital conversion module connected thereto.

3. The device for realizing rapid contact and sliding detection using a PVDF piezoelectric sensor according to claim 2, characterized in that: When the charge-voltage conversion module converts the electrical signal of the PVDF piezoelectric sensor into a voltage signal, the sensor is regarded as an equivalent circuit in which a charge source is connected in parallel with a capacitor and a resistor, and the charge-voltage conversion is realized in the form of an integration circuit.

4. The device for realizing rapid contact and sliding detection using a PVDF piezoelectric sensor according to claim 2, characterized in that: The analog-to-digital conversion module includes a development board, which collects the voltage signal output by the charge-to-voltage conversion module through the analog input pin, converts the voltage signal into a digital signal, and sends the digital signal to the controller through the communication interface.

5. A method for realizing rapid detection of contact sliding using a PVDF piezoelectric sensor, comprising: The method is based on a PVDF piezoelectric sensor and utilizes a neural network algorithm including a neural network model to achieve rapid detection of contact slip; The controller runs the sliding detection and recognition algorithm to implement the following steps for fast contact sliding detection: Step S1: The controller loads the trained neural network model when starting the program; Step S2: the controller establishes communication with the signal acquisition module; Step S3: the electrical signal of the PVDF piezoelectric sensor is converted by the signal acquisition module and acquired by the controller to obtain a digital signal; Step S4: After the controller collects the sensor signal for one cycle, it immediately calculates the time-frequency domain eigenvalues ​​of the signal within the cycle, uses the eigenvalues ​​as input to the neural network model, obtains the slip detection and recognition results, and realizes rapid detection of contact slip; The neural network model establishment process includes the following steps: Step A1: collecting a large number of sensor signals in two contact states, namely, contact sliding and contact non-sliding, to form a sensor signal data set; Step A2: performing noise reduction processing on the collected sensor signals one by one; the sensor signals are time series, and the signal noise reduction is achieved using the VMD decomposition and reconstruction method; Step A3: extracting signal features and constructing a signal feature set, extracting multiple time domain and multiple frequency domain signal features for characterizing the signal features of the PVDF piezoelectric sensor from the noise-reduced signal obtained in step A2; Step A4: Optimize the parameters of the neural network model. Use the Dung Beetle Optimization (DBO) algorithm to optimize the parameters of the long short-term memory (LSTM) network to improve the convergence speed and recognition accuracy of the neural network. Use the signal feature set obtained in step A3 as the training set and test set of the neural network. Use the error rate of the neural network recognition as the fitness function. Optimal parameter values ​​are obtained through iterative updating. Step A5: Construct a neural network model using the signal feature set obtained in step A3 and the LSTM network parameters optimized in step A4. The time domain and frequency domain signal features extracted in step A3 are conventional expression indicators of mechanical vibration signals. When performing contact sliding rapid detection, the detection surface of the PVDF piezoelectric sensor is in direct contact with the object to be detected.

6. The method for realizing rapid contact slip detection using a PVDF piezoelectric sensor according to claim 5, characterized in that: The specific method of signal noise reduction in step A2 is as follows: first, the Archimedean optimization algorithm AOA is used to optimize the parameters [K, α] of the variational mode decomposition VMD. The minimum envelope entropy is used as the fitness function during the optimization process. Each population object when the Archimedean optimization algorithm is initialized contains [K, α]. The [K, α] corresponding to each object is used to perform VMD decomposition on the sensor signal to obtain the minimum envelope entropy. The minimum envelope entropy is the fitness value. The [K, α] corresponding to the object with the smallest fitness value is taken as the optimal object. In each iteration, the [K, α] corresponding to each population object is updated, and then The fitness value of each population object is then recalculated. If the fitness value of the new object is smaller than the fitness value of the current best object, the best object is updated to the [K, α] corresponding to the new object. At the end of the iteration, the [K, α] corresponding to the best object is the best parameter obtained by optimization. The best parameters are obtained by iterative update, and then the sensor signal is subjected to VMD decomposition using the optimized best parameters. According to the correlation coefficient values ​​between each modal component obtained by VMD decomposition and the original signal, the modal components with correlation coefficient values ​​greater than 0.2 are selected to reconstruct the sensor signal to obtain the denoised signal.

7. The method for realizing rapid contact sliding detection using a PVDF piezoelectric sensor according to claim 5, characterized in that: In step A3, 12 time domain and 4 frequency domain signal features are extracted, namely peak-to-peak value F1, average value F2, absolute average value F3, variance F4, standard deviation F5, root mean square F6, kurtosis F7, skewness F8, shape factor F9, peak factor F 10 , pulse factor F 11 , margin factor F 12 , average frequency F 13 , center frequency F 14 , frequency root mean square F 15 , frequency standard deviation F 16 The 16 eigenvalue calculation formulas are shown in formulas (1) to (16) in the following table: Where: x(n) represents the time series of the signal, n = 1, 2, ..., N, N represents the number of sample points; through discrete Fourier transform, the time series is converted into a spectrum, X k The spectrum of the signal is shown, k=0,1,…,N-1, f k =k / N×fs, fs represents the sampling frequency of the signal.

Images