A method and system for fault diagnosis of a transmission system

Abstract

This invention discloses a fault diagnosis method and system for a transmission system. The method includes: installing a triboelectric sensor on the coupling of the transmission system; collecting the output signals of the triboelectric sensor under different operating conditions of the transmission system; extracting and calibrating features from the output signals to obtain a dataset; training and testing a support vector machine using the dataset to obtain a fault diagnosis model; and performing intelligent fault diagnosis of the transmission system based on the fault diagnosis model. This invention employs a specific installation method to install the triboelectric sensor on the elastic body of the coupling, effectively increasing the amplitude of the triboelectric sensor's output signal. This results in significant differences in the characteristics of the output signal under different operating conditions, effectively improving the accuracy of subsequent model identification of fault types. Furthermore, the method is compact, easy to install, and has a wide range of applications. Combined with machine learning algorithms, it achieves automatic identification of transmission system fault types, which is beneficial for improving the intelligence level of industrial automation.

Description

Technical Field

[0001] This invention relates to the field of transmission system fault diagnosis technology, and in particular to a method and system for diagnosing transmission system faults. Background Technology

[0002] With the development of intelligent manufacturing and artificial intelligence in industry, transmission system sensors play a crucial role in modern industry. They are widely used to monitor and control the operating status of transmission systems, ensuring the safety and reliability of equipment. In recent years, their development has trended towards higher precision, higher sensitivity, multi-sensor data fusion, and automated processing.

[0003] Currently, there are many solutions for fault diagnosis of transmission systems, most of which use vibration probes as sensors. Although they are widely applicable and can respond to various faults, their vibration signals are mixed, the causes of vibration are complex and difficult to identify, and the difference between normal signals and fault signals is very small. Therefore, it is difficult to analyze the fault type through its vibration signals.

[0004] Therefore, existing technologies still need to be improved and developed. Summary of the Invention

[0005] The main objective of this invention is to provide a fault diagnosis method and system for a transmission system, aiming to solve the problem that in the prior art, when using sensors to diagnose faults in a transmission system, the difference between health signals and fault signals is very small, which makes it difficult to identify the fault type.

[0006] To achieve the above objectives, this application provides a fault diagnosis method for a transmission system, the fault diagnosis method for the transmission system comprising:

[0007] Install a triboelectric sensor on the coupling of the transmission system;

[0008] The output signal of the triboelectric sensor is collected under different operating conditions of the transmission system;

[0009] The output signal is subjected to feature extraction and calibration to obtain a dataset;

[0010] The support vector machine was trained and tested using the dataset to obtain a fault diagnosis model.

[0011] The target output signal of the triboelectric sensor in the transmission system to be diagnosed is acquired in real time, and the target output signal is input into the fault diagnosis model to perform transmission system fault diagnosis, and the fault type of the transmission system to be diagnosed is output.

[0012] Optionally, the coupling includes two steel claws and a quincunx-shaped elastic body serving as an intermediate body connecting the two steel claws;

[0013] The installation of a triboelectric sensor on the coupling of the transmission system specifically involves:

[0014] The triboelectric sensor is embedded in the gap between the plum blossom-shaped elastomer and the steel claw, and the triboelectric sensor is installed on any of the several extrusion surfaces of the plum blossom-shaped elastomer in a specific installation manner; wherein, the specific installation manner refers to the installation manner in which the amplitude of the output signal of the triboelectric sensor is greater than a preset value and the difference in the characteristics of the output signal of the triboelectric sensor under different working conditions of the transmission system is greater than a threshold.

[0015] Optionally, the triboelectric sensor consists of a three-layer structure of a triboelectric plate, a conductive copper foil, and a PI insulating tape. The triboelectric sensor is used to collect vibration signals between the plum blossom-shaped elastomer and the steel claw, and convert the vibration signals into voltage signals.

[0016] Optionally, the output signal of the triboelectric sensor under different operating conditions of the acquisition transmission system specifically includes:

[0017] The output signal of the triboelectric sensor is collected under the healthy condition of the transmission system and used as a health signal;

[0018] The output signal of the triboelectric sensor is collected under simulated fault conditions of the transmission system and used as a fault signal;

[0019] The simulated fault conditions include belt drive slippage, drive shaft misalignment, and gear wear.

[0020] Optionally, the step of extracting and calibrating features from the output signal to obtain a dataset specifically includes:

[0021] Perform a Fast Fourier Transform on the output signal to obtain the frequency domain signal corresponding to the output signal;

[0022] The time-domain feature value of the output signal and the frequency-domain feature value of the corresponding frequency-domain signal are extracted respectively. The time-domain feature value and the frequency-domain feature value both include standard deviation, root mean square, waveform factor and peak-to-peak value.

[0023] The health signal and fault signal in the output signal are calibrated to obtain the tag of the output signal;

[0024] A dataset is constructed based on the feature values ​​and labels of the output signal.

[0025] Optionally, the step of training and testing the support vector machine using the dataset to obtain the fault diagnosis model specifically includes:

[0026] The dataset is divided into a training set and a test set;

[0027] Based on the improved particle swarm optimization algorithm, the optimal penalty factor and optimal kernel function parameters of the support vector machine are determined;

[0028] Based on the optimal penalty factor and the optimal kernel function parameters, the support vector machine is trained using the training set to obtain a classification model of the support vector machine;

[0029] The support vector machine classification model is tested using the test set to verify its performance. The verified support vector machine classification model is then used as a fault diagnosis model for transmission system fault diagnosis.

[0030] In addition, to achieve the above objectives, the present invention also provides a fault diagnosis system for a transmission system, used to implement the steps in any of the above-described fault diagnosis methods for a transmission system, wherein the fault diagnosis system for the transmission system includes a transmission system test bench and a triboelectric sensor.

[0031] The transmission system test bench includes a base and a variable frequency motor, a speed sensor, a gearbox, a torque sensor, a conveyor pulley, a conveyor belt, a magnetic powder brake, and a control console mounted on the base. The motor shaft of the variable frequency motor is connected to the input shaft of the gearbox via a coupling. The coupling includes two steel claws and a plum blossom-shaped elastomer serving as an intermediate body connecting the two steel claws. The triboelectric sensor is embedded in the gap between the plum blossom-shaped elastomer and the steel claws. The triboelectric sensor is mounted on any surface of several extrusion surfaces of the plum blossom-shaped elastomer in a specific mounting manner.

[0032] Optionally, the triboelectric sensor consists of a three-layer structure of a triboelectric plate, a conductive copper foil, and a PI insulating tape. The triboelectric sensor is used to collect vibration signals between the plum blossom-shaped elastomer and the steel claw, and convert the vibration signals into voltage signals.

[0033] Optionally, the fault diagnosis system of the transmission system further includes a conductive slip ring, an oscilloscope, and a host computer. The conductive slip ring is installed on the motor shaft where the coupling is located. The triboelectric sensor is connected to one end of the conductive slip ring, the other end of the conductive slip ring is connected to one end of the oscilloscope, and the other end of the oscilloscope is connected to the host computer.

[0034] Optionally, the conductive slip ring is used to derive the output signal of the triboelectric sensor, the oscilloscope is used to acquire the output signal, the host computer uses MATLAB software to extract features and calibrate the output signal to obtain a dataset, and the host computer also uses a support vector machine algorithm to classify and identify the dataset and output the fault type of the transmission system.

[0035] Beneficial Effects: Compared with existing technologies, this invention provides a fault diagnosis method and system for transmission systems. The method includes: installing a triboelectric sensor on the coupling of the transmission system; collecting the output signals of the triboelectric sensor under different operating conditions of the transmission system; extracting and calibrating features from the output signals to obtain a dataset; training and testing a support vector machine using the dataset to obtain a fault diagnosis model; and performing intelligent fault diagnosis of the transmission system based on the fault diagnosis model. This invention employs a specific installation method to install the triboelectric sensor on the elastic body of the coupling, effectively increasing the amplitude of the triboelectric sensor's output signal, resulting in significant differences in output signal characteristics under different operating conditions. This effectively improves the accuracy of subsequent model identification of fault types. Furthermore, the invention is compact, easy to install, and widely applicable. Combined with machine learning algorithms, it achieves automatic identification of transmission system fault types, which is beneficial for improving the intelligence level of industrial automation. Attached Figure Description

[0036] Figure 1 This is a flowchart of a preferred embodiment of the conventional system fault diagnosis method of the present invention;

[0037] Figure 2 This is a structural diagram of the coupling and triboelectric sensor of the present invention;

[0038] Figure 3 This is a structural diagram of the elastomer of the present invention;

[0039] Figure 4 This is a schematic diagram of the installation of the double-sided electrodes on the elastomer of the present invention;

[0040] Figure 5 This is a schematic diagram of the installation of a single-sided electrode on the elastomer of the present invention;

[0041] Figure 6 This is a comparison chart of the average voltage of fault signals and healthy signals under the first to sixth installation methods of the single-sided electrode 5-piece triboelectric sensor of the present invention;

[0042] Figure 7 This is a comparison chart of the average voltage of fault signals and healthy signals under the 7th to 12th installation methods of the single-sided electrode 5-piece triboelectric sensor of the present invention;

[0043] Figure 8 This is a schematic diagram of the slippage fault structure of the belt drive system of the present invention;

[0044] Figure 9 This is a voltage diagram of health signal and fault signal under two installation methods of the present invention: "single-sided electrode 1-2-3-4-5-RLLRL" and "single-sided electrode 1-2-3-4-5-RLRLR".

[0045] Figure 10This is a data table diagram of the feature values ​​and output values ​​in the support vector machine algorithm of this invention;

[0046] Figure 11 This is a flowchart of the support vector machine algorithm of this invention;

[0047] Figure 12 This is a comparison chart of the prediction results of the training set and the test set in the support vector machine algorithm of this invention;

[0048] Figure 13 This is a graph showing the signal changes over 6000 consecutive cycles of operation according to the present invention;

[0049] Figure 14 This is a structural diagram of a preferred embodiment of the conventional system fault diagnosis system of the present invention;

[0050] Figure 15 This is a schematic diagram of the circuit connection of the triboelectric sensor of the present invention. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of this invention clearer and more explicit, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0052] With the rapid advancement of information technology and industrial manufacturing technology, modern industrial equipment is becoming increasingly large-scale, complex, high-speed, integrated, and intelligent, placing ever higher demands on its stability and safety. To prevent downtime caused by aging, mechanical failures, and other malfunctions in mechanical equipment and transmission systems, which could disrupt normal production and daily life, it is necessary to research suitable fault diagnosis methods to quickly identify fault types and perform timely repairs, thus avoiding significant economic losses. Fault diagnosis, as a key technology for maintaining the safe and reliable operation of equipment, aims to quickly and accurately determine the type of fault, thereby reducing equipment operation and maintenance costs.

[0053] The transmission system is a critical component in the operation of mechanical equipment, playing a vital role in transmitting kinetic energy. A high failure rate in the transmission system is a significant cause of equipment downtime; therefore, real-time monitoring and fault diagnosis of the transmission system are crucial for ensuring equipment operation. With the development of intelligent manufacturing and artificial intelligence in industry, transmission system sensors play a key role in modern industry. They are widely used to monitor and control the operating status of transmission systems, ensuring equipment safety and reliability. In recent years, their development has trended towards higher precision, higher sensitivity, multi-sensor data fusion, and automated processing.

[0054] Currently, there are many solutions for fault diagnosis of transmission systems, most of which use vibration probes as sensors. Although they are widely applicable and can respond to various faults, their vibration signals are mixed, the causes of vibration are complex and difficult to identify, and the difference between normal signals and fault signals is very small. Therefore, it is difficult to analyze the fault type through its vibration signals.

[0055] To address the aforementioned technical problems, this invention provides a fault diagnosis method and system for a transmission system. The fault diagnosis method includes: installing a triboelectric sensor on the coupling of the transmission system; collecting the output signals of the triboelectric sensor under different operating conditions; extracting and calibrating features from the output signals to obtain a dataset; training and testing a support vector machine using the dataset to obtain a fault diagnosis model; and collecting the target output signal of the triboelectric sensor in the transmission system to be diagnosed in real time, and inputting the target output signal into the fault diagnosis model for transmission system fault diagnosis. The fault diagnosis system for the transmission system includes a transmission system test bench and a triboelectric sensor. The transmission system test bench includes a base and a variable frequency motor, a speed sensor, a gearbox, a torque sensor, a conveyor pulley, a conveyor belt, a magnetic powder brake, and a control console mounted on the base. The motor shaft of the variable frequency motor is connected to the input shaft of the gearbox via a coupling. The coupling includes two steel claws and a plum blossom-shaped elastic body serving as an intermediate body connecting the two steel claws. The triboelectric sensor is embedded in the gap between the plum blossom-shaped elastic body and the steel claws. The triboelectric sensor is installed on any of the 12 extrusion surfaces of the plum blossom-shaped elastic body in a specific installation manner. This invention mounts a triboelectric sensor on the plum-shaped elastomer of a coupling using a specific installation method. The triboelectric sensor collects vibration signals between the elastomer and the steel claw and converts them into voltage signals. These voltage signals reflect the operating state of the transmission system. Machine learning algorithms are then used to process and analyze these voltage signals to automatically identify corresponding fault types. This application employs a specific installation method to mount the triboelectric sensor on the coupling's elastomer, effectively increasing the amplitude of the sensor's output signal. This results in significant differences in output signal characteristics under different operating conditions, effectively improving the accuracy of subsequent model fault type identification. Furthermore, the structure is compact, easy to install, and has a wide range of applications. This application, combined with machine learning algorithms, achieves automatic identification of transmission system fault types, which is beneficial for improving the intelligence level of industrial automation.

[0056] The application content will be further explained below with reference to the accompanying drawings and the description of the embodiments.

[0057] A preferred embodiment of the fault diagnosis method for the transmission system of the present invention is as follows: Figure 1 As shown, it specifically includes:

[0058] S1. Install a triboelectric sensor on the coupling of the transmission system.

[0059] In one implementation of this embodiment, such as Figure 2 As shown, the coupling 4 includes two steel claws 1 and a plum blossom-shaped elastic body 2 serving as an intermediate body connecting the two steel claws 1. The triboelectric sensor 3 consists of a three-layer structure of a triboelectric plate 301, a conductive copper foil 302, and a PI insulating tape 303. The triboelectric sensor is installed on the coupling of the transmission system by embedding it into the gap between the plum blossom-shaped elastic body and the steel claws, and the triboelectric sensor is installed on any surface of several extrusion surfaces of the plum blossom-shaped elastic body (the elastic body in this embodiment has 12 extrusion surfaces) in a specific installation manner. The specific installation manner refers to the installation manner in which the amplitude of the output signal of the triboelectric sensor is greater than a preset value (the preset value is set to 5V in this embodiment) and the difference in the output signal characteristics of the transmission system under different working conditions is greater than a threshold, that is, the difference in the output signal characteristics is obvious.

[0060] Specifically, the coupling connects the motor shaft of the variable frequency motor to the input shaft of the gearbox. It includes a plum-shaped elastomer and two steel claws. A triboelectric sensor is embedded in the gap between the elastomer and the steel claws. This triboelectric sensor consists of multiple units connected in parallel and installed at different positions on the 12 pressing surfaces of the elastomer. The elastomer, acting as an intermediary connecting the two steel claws, transmits the motor shaft's rotational speed and torque to the input shaft of the gearbox. Therefore, the embedded triboelectric sensor converts the vibration signal on the motor shaft into a voltage signal. It is understood that by installing the triboelectric sensor on the coupling elastomer in a specific arrangement, when the transmission system test bench is working, minute contact separations and friction occur between the steel claws and the elastomer of the coupling, causing the triboelectric sensor to generate a triboelectric signal. This signal can reflect the working state of the transmission system. Based on the principle of triboelectric generation, this invention achieves self-driven triboelectric sensing through a special assembly method, effectively improving the amplitude of the output signal and significantly enhancing the ability to distinguish fault types.

[0061] Different installation methods for triboelectric sensors result in varying fault responses. Through experimental exploration, specific installation methods were identified that caused significant differences in amplitude, frequency, and other characteristics of the output signals under healthy and faulty conditions, thereby improving the fault identification rate.

[0062] Specifically, different installation methods for triboelectric sensors produce different output signals. To investigate the optimal installation method for triboelectric sensors on couplings, this invention conducted batch experiments. For example... Figure 3 As shown, the elastic body 2 of the coupling has 6 blade branches, each blade branch has two mounting surfaces (left and right, represented by L and R). After permutations and combinations and eliminating equivalent duplicates, there are a total of 75 different mounting methods. For example... Figure 4 As shown, Figure 4This is a schematic diagram of the installation of the double-sided electrodes 1-2-4. A triboelectric sensor 3 is installed on each of the left and right sides of the 1, 2, and 4 blade branches of the elastomer 2, as shown below. Figure 5 As shown, Figure 5 The diagram shows the installation of a single-sided electrode 1-2-3-5-LRLR. A triboelectric sensor 3 is installed on the left side of the 1-blade branch, the right side of the 2-blade branch, the left side of the 3-blade branch, and the right side of the 5-blade branch of the elastomer 2.

[0063] Before each round of experiments, triboelectric sensors need to be installed on the coupling. There are 75 different installation methods, as described above. The number and location of the triboelectric sensors will affect the signal quality. This invention compares the sensor output signals under 75 different triboelectric sensor installation methods and determines the optimal signal installation method.

[0064] Specifically, after collecting health and fault signals under various installation methods, to select the best installation method, a "good signal" was defined as follows: the signal amplitude is greater than 5V and fluctuates stably within 6 seconds; the fault signal differs significantly from the health signal, for example, their peak-to-peak value, frequency, and waveform factor should be significantly different to facilitate subsequent fault identification. Based on this standard, some installation methods with better performance were selected. Taking simulated belt drive slippage as an example, such as... Figure 6 and Figure 7 The image shows a comparison of the average voltage of various fault and health signals under 12 different installation methods for a single-sided electrode 5-piece triboelectric sensor. Figure 6 and Figure 7The horizontal axis represents the health signal or fault signal number under different installation methods. 1a represents the health signal under the "single-sided electrode 1-2-3-4-5-RLRLL" installation method, and 1b represents the fault signal under the "single-sided electrode 1-2-3-4-5-RLRLL" installation method; 2a represents the health signal under the "single-sided electrode 1-2-3-4-5-RLRLR" installation method, and 2b represents the fault signal under the "single-sided electrode 1-2-3-4-5-RLRLR" installation method; 3a represents the health signal under the "single-sided electrode 1-2-3-4-5-RLLRL" installation method, and 3b represents the fault signal under the "single-sided electrode 1-2-3-4-5-RLLRL" installation method; 4 4a represents the health signal under the "single-sided electrode 1-2-3-4-5-RLLLL" installation method; 4b represents the fault signal under the "single-sided electrode 1-2-3-4-5-RLLLL" installation method; 5a represents the health signal under the "single-sided electrode 1-2-3-4-5-RLLLR" installation method; 5b represents the fault signal under the "single-sided electrode 1-2-3-4-5-RLLLR" installation method; 6a represents the health signal under the "single-sided electrode 1-2-3-4-5-RLLRR" installation method; 6b represents the fault signal under the "single-sided electrode 1-2-3-4-5-RLLRR" installation method; 7a represents the fault signal under the "single-sided electrode 1-2-3-4-5-LRLLL" installation method. The health signals under the following conditions are listed: 7b represents a fault signal under the "single-sided electrode 1-2-3-4-5-LRLLL" installation method; 8a represents a health signal under the "single-sided electrode 1-2-3-4-5-RRLLL" installation method; 8b represents a fault signal under the "single-sided electrode 1-2-3-4-5-RRLLL" installation method; 9a represents a health signal under the "single-sided electrode 1-2-3-4-5-RRLLR" installation method; 9b represents a fault signal under the "single-sided electrode 1-2-3-4-5-RRLLR" installation method; 10a represents a health signal under the "single-sided electrode 1-2-3-4-5-LLLLL" installation method; 10b represents a fault signal under the "single-sided electrode 1-2-3- The fault signal is for the “4-5-LLLLL” installation method; 11a represents the health signal for the “single-sided electrode 1-2-3-4-5-LLRRR” installation method, and 11b represents the fault signal for the “single-sided electrode 1-2-3-4-5-LLRRR” installation method; 12a represents the health signal for the “single-sided electrode 1-2-3-4-5-LRRRR” installation method, and 12b represents the fault signal for the “single-sided electrode 1-2-3-4-5-LRRRR” installation method; among them, the “single-sided electrode 1-2-3-4-5-RLLRL” and “single-sided electrode 1-2-3-4-5-RRLLR” installation methods have the best effect among all installation schemes and are designated as the best installation methods.

[0065] This invention optimizes the installation method of the triboelectric sensor, thereby increasing its signal amplitude from below 5V to above 10V, and the signal difference is significant under fault conditions such as belt drive slippage compared to normal conditions.

[0066] S2. Collect the output signal of the triboelectric sensor under different operating conditions of the transmission system.

[0067] In one implementation of this embodiment, the acquisition of the output signal of the triboelectric sensor under different operating conditions of the transmission system specifically includes:

[0068] S21. Collect the output signal of the triboelectric sensor under the healthy state of the transmission system and use it as the health signal;

[0069] S22. Collect the output signal of the triboelectric sensor under simulated fault conditions of the transmission system and use it as a fault signal;

[0070] The simulated fault conditions include belt drive slippage, drive shaft misalignment, and gearbox wear.

[0071] Specifically, the sensor embedded in the coupling, designed based on the principle of triboelectric generation, can respond differently to various operating conditions, such as normal operation, belt drive slippage, and drive shaft misalignment, outputting different signals. To achieve automatic fault diagnosis of the transmission system, it is necessary to first collect sensor signals under fault conditions. Therefore, a preset fault method is used to collect the output signals of the triboelectric sensor under simulated fault conditions of the transmission system, comparing them with the signals under healthy conditions to lay the foundation for subsequent fault diagnosis.

[0072] For example: Figure 8 As shown, Figure 8 This is a schematic diagram of a belt drive system slippage fault structure. The driving pulley 16 rotates, while the driven pulley 17 and the conveyor belt 14 slip. In this embodiment, a magnetic powder brake is used as the load of the transmission system. The braking power of the magnetic powder brake is adjusted to increase the load, causing the belt drive to slip, thus simulating a belt drive system slippage fault. The differential power of the magnetic powder brake is 2kW, and the input speed of the three-phase motor is 600rpm. Signals for both healthy and slippage states are measured. After the equipment is installed and the motor is turned on, the magnetic powder brake is reset to zero, and all drive shafts are working normally; at this point, the healthy state signal is recorded. After saving, the magnetic powder brake is adjusted. When the differential power is around 2kW, the belt drive experiences overload slippage. The output signal of the triboelectric sensor at this time is recorded as the slippage state signal (fault signal). Figure 9 As shown, Figure 9 (a) is the health signal collected under the "single-sided electrode 1-2-3-4-5-RLLRL" installation method. Figure 9 (b) is the slippage fault signal collected under the "single-sided electrode 1-2-3-4-5-RLLRL" installation method. Figure 9(c) Health signals collected under the "single-sided electrode 1-2-3-4-5-RLRLR" installation method. Figure 9 (d) is the slippage fault signal collected under the "single-sided electrode 1-2-3-4-5-RLRLR" installation method.

[0073] S3. Extract and calibrate the features of the output signal to obtain the dataset.

[0074] In one implementation of this embodiment, feature extraction and calibration of the output signal are performed to obtain a dataset, specifically including:

[0075] S31. Perform a fast Fourier transform on the output signal to obtain the frequency domain signal corresponding to the output signal;

[0076] S32. Extract the time-domain feature value of the output signal and the frequency-domain feature value of the corresponding frequency-domain signal of the output signal respectively. The time-domain feature value and the frequency-domain feature value include standard deviation, root mean square, waveform factor and peak-to-peak value.

[0077] S33. Calibrate the health and fault signals in the output signal and obtain the tag of the output signal;

[0078] S34. Construct a dataset based on the feature values ​​and labels of the output signal.

[0079] In this embodiment, 10 sets of healthy signals and 10 sets of fault signals are collected respectively to avoid the influence of accidental situations. Feature extraction is performed on these 20 sets of signals, and four feature values ​​are selected: standard deviation (Std), root mean square (RMS), waveform factor (SF), and peak-to-peak value (PTP). By extracting the feature values ​​of the signals, the differences between different signals can be accurately analyzed. Among them, the standard deviation (Std) reflects the dispersion of the waveform signal; the larger the value, the higher the dispersion of the signal. The root mean square (RMS) reflects the effective value of the signal; the higher the amplitude and effective percentage of the signal, the larger its effective value. The waveform factor reflects the shape characteristics of the signal, such as the degree of waveform distortion, which is the ratio of the signal's root mean square value to the rectified average value. The peak-to-peak value (PTP) is the difference between the largest positive value and the largest negative value in the waveform, reflecting the range of signal amplitude variation; the larger the peak-to-peak value, the larger the range of signal variation.

[0080] To analyze the signals more comprehensively and understand their amplitude and frequency characteristics, a Fast Fourier Transform (FFT) was performed on the signals. Four feature values ​​each in the time and frequency domains of the 20 signal sets were extracted, forming a 20×9 dataset. The first eight columns represent the time and frequency domain feature values, and the ninth column represents the output value. Healthy data is labeled as output 1, and fault data is labeled as output 2.

[0081] S4. Use the dataset to train and test the support vector machine to obtain the fault diagnosis model.

[0082] In one implementation of this embodiment, a fault diagnosis model is obtained by training and testing a support vector machine using a dataset, specifically including:

[0083] S41. Divide the dataset into a training set and a test set;

[0084] S42. Based on the improved particle swarm optimization algorithm, determine the optimal penalty factor and optimal kernel function parameters for the support vector machine;

[0085] S43. Based on the optimal penalty factor and optimal kernel function parameters, train the support vector machine using the training set to obtain the classification model of the support vector machine.

[0086] S44. Test the classification model of the support vector machine using the test set to verify the performance of the support vector machine classification model. Use the verified support vector machine classification model as a fault diagnosis model for transmission system fault diagnosis.

[0087] In this embodiment, the Support Vector Machine (SVM) algorithm is used to classify and identify the dataset, enabling automatic identification and diagnosis of health and fault data. The basic idea is to map the data to a high-dimensional space and find an optimal hyperplane (a straight line in two-dimensional space) so that one side of this hyperplane represents one category and the other side another. The steps for automatic fault diagnosis using the SVM algorithm are: first, training, then prediction. First, a large amount of data is used for training to find an optimal hyperplane, i.e., the trained model. Then, this hyperplane is used to classify the test data. Finally, the deviation between the classification result and the manually calibrated result is compared, and the prediction error value is output.

[0088] In this Support Vector Machine (SVM) algorithm, two important parameters need to be confirmed: the penalty factor C and the kernel function parameter g. The penalty factor C adjusts the classifier's tolerance for error. A smaller C value indicates a higher tolerance for error, but too small a value may lead to poor classification performance. A larger C value indicates stricter classification, but too large a value can make the classifier overly sensitive to the training data, leading to overfitting. The kernel function parameter g represents the width of the kernel function. A smaller g value results in a narrower kernel bandwidth, making the SVM model simpler, but too small a value may fail to capture complex structures in the data, affecting classification accuracy. A larger g value results in a wider kernel bandwidth, which can capture more subtle structures in the data, but also increases the risk of overfitting. Figure 10 As shown, through multiple experiments and comparison of the output prediction error, this invention obtained the optimal penalty factor C of 10 and kernel function parameter g of 0.1.

[0089] Based on the above, such as Figure 11As shown, the overall SVM algorithm implementation process for transmission system fault diagnosis includes: acquiring the initial signal and performing low-pass filtering to remove high-frequency interference; extracting the time-domain and frequency-domain feature values ​​of the signal to obtain the dataset; then inputting the aforementioned 20-column 8-dimensional dataset, dividing it into 80% as the training set and 20% as the test set, adjusting the penalty factor to 10, the kernel function parameter to 0.1, and running the training model; after running for a period of time, obtaining the output prediction results. In this experiment, as... Figure 12 As shown in the figure, the true values ​​and predicted values ​​of the training set and the test set are compared respectively. It can be seen from the figure that the true values ​​and predicted values ​​are highly consistent, and their root mean square errors are all less than 0.02, which shows a very high accuracy in fault diagnosis.

[0090] S5. Real-time acquisition of the output signal of the triboelectric sensor in the transmission system to be diagnosed, and input of the output signal into the fault diagnosis model to diagnose the transmission system fault.

[0091] To test the sensor's durability, this invention underwent a long-term experiment of 6000 cycles, such as... Figure 13 As shown in the figure, the experiment proves that the triboelectric sensor of the present invention has no significant voltage signal attenuation after running continuously and stably for 6000 cycles. The voltage signal changes from about 7V at the beginning to about 8.5V at the end. Although there are slight changes in the middle, the overall output voltage is stable at about 8V, which shows that the triboelectric sensor of the present invention has good durability and is suitable for long-term operation.

[0092] This embodiment also provides a fault diagnosis system for a transmission system, used to implement the steps in the above-described fault diagnosis method for a transmission system, such as... Figure 14 As shown, a preferred embodiment of the fault diagnosis system for the transmission system includes a transmission system test bench and a triboelectric sensor. The transmission system test bench includes a base 8 and a variable frequency motor 9, a speed sensor 10, a gearbox 11, a torque sensor 12, a pulley 13, a conveyor belt 14, a magnetic powder brake 15, and a control console mounted on the base 8. The motor shaft of the variable frequency motor 9 is connected to the input shaft of the gearbox 11 via a coupling 4. Figure 2 As shown, the coupling 4 includes two steel claws 1 and a plum blossom-shaped elastic body 2, which serves as an intermediate body connecting the two steel claws 1. The triboelectric sensor 3 is embedded in the gap between the plum blossom-shaped elastic body 2 and the steel claws 1. The triboelectric sensor 3 is installed on any surface of a plurality of extrusion surfaces (the elastic body in this embodiment has 12 extrusion surfaces) of the plum blossom-shaped elastic body 3 in a specific installation manner.

[0093] Specifically, the present invention includes a self-made triboelectric sensor 3, a coupling 4, a commercially available conductive slip ring 5, an oscilloscope 6, a host computer 7, and a transmission system test bench. The test bench includes a base 8, a variable frequency motor 9, a speed sensor 10, a gearbox 11, a torque sensor 12, a pulley 13, a conveyor belt 14, a magnetic powder brake 15, and a control console. The overall device is as follows: Figure 14 As shown. Figure 2 As shown, the triboelectric sensor 3 is installed in the gap between the elastic body 2 and the steel claw 1 of the coupling 4. The triboelectric sensor 3 is composed of a three-layer structure: a triboelectric plate 301, a conductive copper foil 302, and a PI insulating tape 303. The triboelectric material can be one or more of PTFE film, PVDF film, and PDMS film. The coupling elastic body 2, as an intermediate body connecting the two steel claws 1 (the other is not shown), can transmit the rotational speed and torque of the motor shaft to the input shaft of the rear gearbox. Therefore, the triboelectric sensor 3 embedded therein can convert the vibration signal on this shaft into an electrical signal. Furthermore, the plate-shaped triboelectric sensor can be installed on any of the 12 extruded surfaces of the clover-shaped elastic body 1. Multiple sensors can be connected in parallel to form a multi-sensor array by connecting the conductive copper foil to each triboelectric sensor. The clover-shaped coupling is a standard part, and the conductive slip ring and wires are commercially available. The oscilloscope model is RIGOL DS1000Z-E.

[0094] In this embodiment, triboelectric sensors are mounted on the elastic body of a plum-shaped coupling in a specific arrangement. When the transmission system test bench is working, minute contact separations and friction occur between the steel claws of the coupling and the elastic body, causing the triboelectric sensors mounted on them to generate triboelectric signals. These signals can reflect the working status of the transmission system. In other words, the triboelectric sensors of this invention are mounted on the coupling connecting the motor shaft and the gearbox shaft, and the working status of the entire transmission system is monitored in real time by detecting the working condition of the motor shaft.

[0095] Different triboelectric sensor installation methods result in varying fault responses. Through experimental exploration, a specific installation method was identified, causing significant differences in amplitude, frequency, and other characteristics of the output signal between healthy and faulty states, thus improving fault identification rates. Identifiable transmission system fault types include belt slippage, drive shaft misalignment, and gear wear. After using a specific triboelectric sensor installation method, experiments were conducted and data collected. Voltage signal data was acquired using an oscilloscope, and the signals were processed, calibrated, and classified using MATLAB software on a host computer (e.g., PC). A support vector machine (SVM) algorithm was used to train a model, ultimately enabling fault signal diagnosis and identification. This invention, based on the principle of triboelectric generation, achieves self-driven triboelectric sensing through a special assembly method, effectively improving the amplitude of the output signal and significantly enhancing the ability to distinguish fault types. Furthermore, by incorporating machine learning algorithms, it can achieve automatic diagnosis of real transmission system faults. Its advantages include high signal sensitivity, convenient installation, and ease of building automated sensor networks, which is beneficial for improving the intelligence level of industrial automation.

[0096] In one implementation of this embodiment, such as Figure 2 As shown, the triboelectric sensor 3 consists of a three-layer structure: a triboelectric element 301, a conductive copper foil 302, and a PI insulating tape 303. The triboelectric sensor 3 is used to collect vibration signals between the plum blossom-shaped elastomer 2 and the steel claw 1, and converts the vibration signals into voltage signals. This sensor design, based on the principle of triboelectric generation, can directly collect friction and vibration signals between the coupling elastomer and the steel claw, converting them into voltage signals through the triboelectric generation principle. It features rapid response, self-driving, and a compact structure.

[0097] In one implementation of this embodiment, the fault diagnosis system for the transmission system further includes a conductive slip ring, an oscilloscope, and a host computer. The conductive slip ring is mounted on the motor shaft where the coupling is located. A triboelectric sensor is connected to one end of the conductive slip ring, and the other end of the conductive slip ring is connected to one end of the oscilloscope. The other end of the oscilloscope is connected to the host computer.

[0098] Specifically, the triboelectric sensor is installed inside the coupling, with two signal wires leading out and connected to the inner loop wire of a conductive slip ring. The conductive slip ring is mounted on the motor shaft where the coupling is located, and the outer loop is fixed to the base. The inner loop wire is connected to the outer loop wire through a brush inside the conductive slip ring, ultimately connecting to the signal probe of an oscilloscope. The oscilloscope's output is connected to a host computer. The oscilloscope's sampling frequency is 2kHz, and the sampling duration is 6 seconds. Figure 15 As shown, the signal probe at one end of the oscilloscope 6 is connected to the conductive copper foil 302 of the triboelectric sensor 3 via a conductive slip ring 5, and the other end of the oscilloscope 6 is grounded along with the steel claw 1 of the coupling 4.

[0099] In one implementation of this embodiment, a conductive slip ring is used to derive the output signal of the triboelectric sensor, an oscilloscope is used to acquire the output signal, a host computer uses MATLAB software to extract features from the output signal and calibrate it to obtain a dataset, and the host computer also uses a support vector machine algorithm to classify and identify the dataset and output the fault type of the transmission system.

[0100] Specifically, the voltage signal generated by the triboelectric sensor is led out via a conductor through a conductive slip ring mounted on the drive shaft to avoid wire tangling caused by rotation. The led-out signal is connected to an oscilloscope via its signal probe for preview and saving. The oscilloscope is connected to a host computer to import the saved data, and MATLAB software is used to analyze, process, and calibrate the data. Finally, the dataset is trained using a Support Vector Machine (SVM) algorithm to achieve automatic identification and diagnosis of different sensor signals.

[0101] Given the existence of existing condition monitoring sensor solutions embedded in couplings, the difference in this invention lies in the use of different triboelectric sensor installation methods for different types of transmission system faults, resulting in a significant gain effect on the sensing signal compared to conventional solutions. It is understood that this invention achieves self-driven triboelectric sensing through a special assembly method, effectively increasing the amplitude of the output signal, significantly enhancing the ability to distinguish fault types, and, combined with machine learning algorithms, enabling automatic diagnosis of real transmission system faults.

[0102] In summary, the sensor embedded in the coupling based on the principle of triboelectric power generation can respond differently to various working conditions, such as normal operation, belt drive slippage, and drive shaft misalignment, outputting different signals. Using machine learning algorithms to process and analyze the output signals can automatically identify the corresponding fault types. Through experiments with different combinations of triboelectric sensors, combinations with strong output signals, sensitive fault responses, and significant differences in signal amplitude and frequency were explored. These combinations are highly beneficial for backend signal processing and identification, effectively improving the accuracy of fault identification. This invention uses sheet-like triboelectric material embedded in the gap between the coupling's elastomer and steel claws. It is small in size, has no impact on the coupling's performance, requires no additional power supply compared to traditional probe sensors, is easy to install, and can be widely applied to various transmission systems with couplings.

[0103] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal that includes that element.

[0104] It should be understood that the application of the present invention is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A method of diagnosing a failure of a power transmission system, characterized by, The fault diagnosis method of the transmission system comprises: The coupling comprises two steel claws and a plum blossom elastic body as an intermediate body connecting the two steel claws; A friction power generation sensor is mounted on the coupling of the transmission system, in particular: The friction power generation sensor is embedded in the gap between the plum blossom elastic body and the steel claw, and the friction power generation sensor is mounted on any surface of the plurality of extrusion surfaces of the plum blossom elastic body in a specific mounting manner; wherein the specific mounting manner refers to a mounting manner in which the amplitude of the output signal of the friction power generation sensor is greater than a preset value and the characteristic difference of the output signal of the friction power generation sensor under different working conditions of the transmission system is greater than a threshold value; The output signal of the friction power generation sensor under different working conditions of the transmission system is collected; The output signal is feature extracted and labeled to obtain a data set; A support vector machine is trained and tested using the data set to obtain a fault diagnosis model; The target output signal of the friction power generation sensor in the transmission system to be diagnosed is collected in real time, and the target output signal is input into the fault diagnosis model for transmission system fault diagnosis, and the fault type of the transmission system to be diagnosed is output.

2. The failure diagnostic method of a power transmission system according to claim 1, characterized by, The friction power generation sensor is composed of three layers of friction power generation sheets, conductive copper foils and PI insulating tapes, and is used to collect vibration signals between the plum blossom elastic body and the steel claw and convert the vibration signals into voltage signals.

3. The failure diagnostic method of a power transmission system according to claim 1, characterized by, The output signal of the friction power generation sensor under different working conditions of the transmission system is collected, in particular: The output signal of the friction power generation sensor under the healthy state of the transmission system is collected and used as a healthy signal; The output signal of the friction power generation sensor under the simulated fault state of the transmission system is collected and used as a fault signal; The simulated fault state includes belt drive slip, transmission shaft misalignment and gear wear.

4. The failure diagnostic method of a transmission system according to claim 3, characterized by, The output signal is feature extracted and labeled to obtain a data set, in particular: The output signal is subjected to fast Fourier transform to obtain a frequency domain signal corresponding to the output signal; The time domain characteristic values of the output signal and the frequency domain characteristic values of the frequency domain signal corresponding to the output signal are extracted, and the time domain characteristic values and the frequency domain characteristic values both include standard deviation, root mean square, waveform factor and peak-to-peak value; The healthy signal and the fault signal in the output signal are labeled to obtain labels of the output signal; Based on the characteristic values and the labels of the output signal, a data set is constructed.

5. The failure diagnostic method of a transmission system according to claim 1 or 4, characterized in that, The support vector machine is trained and tested using the data set to obtain a fault diagnosis model, in particular: The data set is divided into a training set and a test set; Based on an improved particle swarm algorithm, the best penalty factor and the best kernel function parameter of the support vector machine are determined; The training set is used to train the support vector machine according to the best penalty factor and the best kernel function parameter to obtain a classification model of the support vector machine; The classification model of the support vector machine is tested using the test set to verify the performance of the classification model of the support vector machine, and the classification model of the support vector machine that passes the verification is used as a fault diagnosis model for transmission system fault diagnosis.

6. A failure diagnosis system of a power transmission system for implementing the steps in the failure diagnosis method of the power transmission system according to any one of claims 1 to 5, characterized by The fault diagnosis system of the transmission system comprises a transmission system test bench and a friction power generation sensor; The transmission system test bench comprises a base, a variable frequency motor, a rotating speed sensor, a gear box, a torque sensor, a conveyor belt wheel, a conveyor belt, a magnetic powder brake and a control console arranged on the base, a motor shaft of the variable frequency motor is connected with an input shaft of the gear box through a coupling, the coupling comprises two steel claws and a plum-blossom-shaped elastic body as an intermediate body connecting the two steel claws, the friction power generation sensor is embedded in a gap between the plum-blossom-shaped elastic body and the steel claws, and the friction power generation sensor is installed on any surface of several extrusion surfaces of the plum-blossom-shaped elastic body in a specific installation mode.

7. The failure diagnostic system of a transmission system according to claim 6, characterized by, The friction power generation sensor comprises three layers of a friction power generation sheet, a conductive copper foil and a PI insulating tape, the friction power generation sensor is used for collecting vibration signals between the plum-blossom-shaped elastic body and the steel claws and converting the vibration signals into voltage signals.

8. The failure diagnostic system of a transmission system according to claim 6 or 7, characterized by, The fault diagnosis system of the transmission system further comprises a conductive slip ring, an oscilloscope and an upper computer, the conductive slip ring is installed on a motor shaft where the coupling is located, one end of the friction power generation sensor is connected with the conductive slip ring, the other end of the conductive slip ring is connected with one end of the oscilloscope, and the other end of the oscilloscope is connected with the upper computer.

9. The failure diagnostic system of a transmission system according to claim 8, characterized by, The conductive slip ring is used for leading out output signals of the friction power generation sensor, the oscilloscope is used for collecting the output signals, the upper computer is used for performing feature extraction and calibration on the output signals by using MATLAB software to obtain a data set, and the upper computer is further used for performing classification and identification on the data set by using a support vector machine algorithm to output a fault type of the transmission system.

Images