Gearbox noise monitoring method

Abstract

The invention discloses a gearbox noise monitoring method, and mainly relates to the field of noise monitoring. Comprising the steps that a plurality of acoustic sensors are installed at preset positions around the gearbox, the acoustic sensors are arranged in the mode that the main noise radiation area of the gearbox can be covered, the sensors have the broadband response characteristic and can collect noise signals with the frequency range from [X] Hz to [Y] Hz, and X and Y are specific numerical values determined according to the working frequency range of the gearbox. The method has the beneficial effects that the high-sensitivity acoustic sensors are reasonably arranged around the gearbox, and the advanced signal processing and feature extraction technology is applied, so that tiny changes in noise signals can be caught acutely. The changes are often closely related to potential faults such as initial wear of a gear, early fatigue of a bearing, poor meshing and the like.

Description

Technical Field

[0001] The invention relates to the field of noise monitoring, and in particular to a gear box noise monitoring method. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] With the continuous development of industrial automation and intelligence, traditional manual inspection and regular maintenance methods are gradually unable to meet the needs of modern production. This is because problems such as missed inspections, misjudgments, and failure to respond in a timely manner caused by human factors are becoming increasingly apparent, increasing the risk of equipment failure and maintenance costs. Therefore, the use of advanced monitoring technology, especially monitoring based on sound signals, has become an effective way to solve this problem.

[0004] During the operation of the gearbox, the interaction between the various components will generate different types of noise, the main sources of which include gear meshing, bearing friction, and the flow of lubricating oil. The characteristics of these noises can be extracted and analyzed through methods such as spectrum analysis and time domain analysis, thus providing a strong basis for the health status of the gearbox. Studies have shown that gearbox failures, especially gear wear, poor gear meshing, bearing damage, etc., usually lead to significant changes in acoustic signals. The characteristics of these changes usually have a certain regularity and can be effectively extracted and used for fault diagnosis.

[0005] In recent years, with the advancement of sensor technology and signal processing technology, sound-based monitoring methods have been widely used. By installing acoustic sensors on the surface of the gearbox, the noise signals generated during operation can be collected in real time. Modern signal processing technologies such as fast Fourier transform (FFT) and wavelet transform are used to analyze the signals and extract characteristic parameters related to the gearbox status. These parameters can be used to identify fault modes, evaluate fault severity, and monitor the operating status of the equipment.

[0006] Compared with traditional vibration monitoring methods, sound signal monitoring has many advantages. First, the acquisition of sound signals is relatively simple and easy to implement, and only requires setting up a recording device or microphone near the equipment; second, sound signals can reflect the overall status of the equipment operation to a certain extent, rather than just the status of a specific component, and thus can provide more comprehensive fault detection information; finally, through in-depth analysis of sound signals, early warning of potential faults can be achieved, which helps reduce downtime and maintenance costs and improve equipment reliability and safety.

[0007] With the development of artificial intelligence and big data technology, fault diagnosis methods based on noise monitoring will become more intelligent and automated. Through intelligent algorithms such as machine learning, the correlation between noise characteristics and faults can be automatically mined and identified in massive data, thereby achieving efficient fault prediction and health management. Through these technical means, gearbox noise monitoring will not only be limited to fault diagnosis, but will also provide decision support for equipment performance optimization and resource management in the future, driving the entire industrial field towards a more intelligent, green and sustainable direction. Summary of the invention

[0008] The purpose of the present invention is to provide a gearbox noise monitoring method, which can keenly capture subtle changes in noise signals by reasonably arranging high-sensitivity acoustic sensors around the gearbox and using advanced signal processing and feature extraction technology. These changes are often closely related to potential faults such as initial wear of gears, early fatigue of bearings, and poor meshing. For example, when the gear tooth surface begins to show slight wear, the spectral characteristics of the noise will show slight energy changes and slight frequency shifts in specific frequency bands. This monitoring method can accurately identify these early signs of faults, effectively avoiding further deterioration of the fault and sudden shutdown of the equipment, greatly reducing maintenance costs and production losses caused by shutdowns.

[0009] In order to achieve the above object, the present invention is implemented through the following technical solutions:

[0010] A gearbox noise monitoring method:

[0011] S1. Install a plurality of acoustic sensors at predetermined locations around the gearbox, wherein the arrangement of the acoustic sensors is capable of covering the main noise radiation area of ​​the gearbox, and the sensors have a wide frequency response characteristic and are capable of collecting noise signals in the frequency range of [X] Hz to [Y] Hz, where X and Y are specific values ​​determined according to the operating frequency range of the gearbox;

[0012] S2. Using the acoustic sensor to collect the noise signal generated by the gearbox during operation in real time, and transmitting the simulated noise signal to the signal conditioning circuit;

[0013] S3. The signal conditioning circuit amplifies and filters the input analog noise signal, with an amplification factor of [Z] times. The filtering adopts a combination of low-pass, high-pass and band-pass filters to remove electromagnetic interference, environmental background noise and invalid frequency components beyond the normal working frequency range of the gearbox in the signal, and obtains a conditioned analog noise signal;

[0014] S4. Converting the conditioned analog noise signal into a digital noise signal through an analog-to-digital converter, with a sampling frequency of not less than [F] Hz, to ensure that detailed information in the noise signal can be accurately captured, and the digital noise signal is transmitted to a microprocessor;

[0015] S5. The microprocessor preprocesses the digital noise signal, uses a wavelet transform algorithm to remove residual noise in the signal, and performs frame processing on the signal. The duration of each frame is [M] ms, and the frame shift is [N] ms, to obtain a preprocessed noise frame signal;

[0016] S6. Extract characteristic parameters from the preprocessed noise frame signal, including time domain characteristics of the noise signal such as root mean square value, peak factor, pulse factor, frequency domain characteristics such as spectrum center of gravity, spectrum peak frequency, energy proportion of each frequency band, and time-frequency domain characteristics such as wavelet packet energy entropy and Hilbert-Huang transform marginal spectrum characteristics;

[0017] S7. Inputting the extracted characteristic parameters into a pre-trained fault diagnosis model, wherein the fault diagnosis model is constructed based on a machine learning algorithm or a deep learning algorithm, and is trained and learned on a large amount of noise characteristic data containing different gearbox operating conditions and fault types, so as to realize the identification and judgment of the gearbox fault type and fault degree;

[0018] S8. Determine the operating status of the gearbox based on the output results of the fault diagnosis model. If it is judged to be a fault state, send the fault information to the remote monitoring terminal through the communication module. The fault information includes the fault type, fault severity, fault occurrence time and corresponding noise characteristic parameters. The communication module adopts wired communication or wireless communication.

[0019] The acoustic sensor is a capacitive microphone or a piezoelectric sensor, which can work stably within an operating temperature range of [Ta]°C to [Tb]°C and has good vibration resistance, wherein Ta and Tb are specific parameter values ​​that meet the actual working environment requirements of the gearbox.

[0020] The training process of the fault diagnosis model includes dividing the collected historical noise feature data into a training set, a validation set and a test set according to a certain ratio, optimizing the hyperparameters of the model by a cross-validation method, and completing the model training when the accuracy of the model on the test set reaches a predetermined threshold [P]% or more, where P is an accuracy index value set according to actual application requirements.

[0021] Before the microprocessor preprocesses the digital noise signal, it also includes a data verification step, which uses a verification algorithm to check the integrity and rationality of the collected digital noise signal, eliminates abnormal data points, and supplements missing data using linear interpolation or based on the statistical laws of adjacent frame data to ensure data continuity and reliability.

[0022] The remote monitoring terminal is provided with a visual interface, which can display the noise spectrum diagram, operation status information, fault alarm information and trend chart of historical monitoring data of the gearbox in real time, so as to facilitate the operator to intuitively understand the working condition and fault occurrence law of the gearbox. The visual interface also supports data export and printing functions for further data analysis and report generation.

[0023] The microprocessor can also perform real-time evaluation of the operating status of the gearbox according to preset threshold conditions. When the extracted noise characteristic parameters exceed the normal threshold range, the local sound and light alarm device is immediately activated to remind the on-site operator to pay attention to the operation of the gearbox in time. The alarm volume of the sound and light alarm device is not less than [V]dB, and the alarm light color is eye-catching red or yellow, where V is the volume value that can attract the attention of the operator under the on-site environmental noise conditions.

[0024] The amplifier circuit in the signal conditioning circuit adopts a programmable gain amplifier, which can automatically adjust the amplification factor according to the actual amplitude of the noise signal to ensure the accuracy and resolution of the signal during the analog-to-digital conversion process, while avoiding signal overload distortion; the filter parameters of the filtering circuit can be adjusted in real time by a microprocessor to adapt to the changes in the frequency characteristics of the gearbox noise signal under different working conditions.

[0025] The method also includes a self-diagnosis function for the gearbox noise monitoring system. During the operation of the system, hardware devices such as acoustic sensors, signal conditioning circuits, analog-to-digital converters, microprocessors and communication modules are regularly tested.

[0026] Compared with the prior art, the beneficial effects of the present invention are:

[0027] Accurate detection of early faults: By rationally arranging highly sensitive acoustic sensors around the gearbox and applying advanced signal processing and feature extraction techniques, it is possible to keenly capture subtle changes in noise signals. These changes are often closely related to potential faults such as initial wear of gears, early fatigue of bearings, and poor meshing. For example, when the gear tooth surface begins to show slight wear, the spectral characteristics of the noise will show slight energy changes and slight frequency shifts in specific frequency bands. This monitoring method can accurately identify these early signs of faults, effectively avoiding further deterioration of the fault and sudden shutdown of the equipment, greatly reducing maintenance costs and production losses caused by shutdowns.

[0028] Comprehensive fault diagnosis capability: Extract rich time domain, frequency domain and time-frequency domain feature parameters from noise signals and input them into the fault diagnosis model built based on machine learning or deep learning algorithms. The model has been trained with a large amount of data of different working conditions and fault types and has strong pattern recognition capabilities. It can not only accurately determine whether there is a fault in the gearbox, but also accurately identify a variety of specific fault types, such as broken teeth, tooth surface bonding, tooth surface peeling, inner ring cracking, outer ring wear, rolling element damage, as well as faults caused by insufficient lubrication and improper assembly. This comprehensive and accurate fault diagnosis capability provides maintenance personnel with detailed fault information, which helps them quickly determine the cause of the fault and develop targeted maintenance plans, greatly improving maintenance efficiency and quality, reducing maintenance time waste and incorrect replacement of parts due to inaccurate fault detection, further reducing equipment maintenance costs and downtime, and improving the reliability and stability of the entire production system.

[0029] Advantages of non-contact real-time monitoring: Compared with some traditional contact monitoring methods, noise monitoring is a non-contact monitoring technology. It does not require complex internal modifications to the gearbox and sensor installation and debugging, and will not interfere with the normal operation of the gearbox. It also avoids problems such as data loss or measurement errors caused by poor sensor contact or damage. Moreover, this method can continuously collect and analyze noise data in real time, monitor the operating status of the gearbox 24 hours a day, and immediately issue an alarm and perform fault diagnosis in the event of abnormal noise changes, so that operators and maintenance personnel can promptly grasp the health status of the equipment and take corresponding measures in a timely manner to ensure the continuity and safety of the production process. It is particularly suitable for industrial production scenarios that require extremely high equipment operation stability.

[0030] Reduce labor intensity and improve work efficiency: Due to the use of automated noise monitoring and intelligent fault diagnosis systems, operators do not need to conduct frequent manual inspections and manual tests, which greatly reduces the labor intensity of staff and enables them to devote more energy to other important production tasks and equipment maintenance management. At the same time, fast and accurate fault diagnosis and timely alarm notifications enable maintenance personnel to respond quickly and solve problems efficiently, avoiding production stagnation caused by long troubleshooting time, significantly improving the work efficiency of the entire production team, and enhancing the production benefits and competitiveness of the enterprise.

[0031] Optimize equipment maintenance strategies: Based on long-term accumulated noise monitoring data and accurate fault diagnosis results, enterprises can deeply analyze the failure patterns of gearboxes and the deterioration trends of equipment, so as to formulate more scientific and reasonable preventive maintenance plans and equipment maintenance strategies. Reasonable arrangement of equipment downtime and parts replacement cycle avoids waste of resources and unnecessary equipment downtime caused by excessive maintenance, and also prevents sudden equipment failures caused by insufficient maintenance. This data-driven equipment maintenance strategy optimization helps enterprises to achieve reasonable allocation of equipment maintenance resources, reduce the cost of the entire life cycle of equipment, improve the comprehensive utilization rate and production efficiency of equipment, and ensure the sustainable development of enterprise production.

[0032] Improve the working environment and protect the health of personnel: The noise generated during the operation of the gearbox often causes serious noise pollution to the surrounding working environment. Long-term exposure to a high-noise environment will have adverse effects on the hearing and physical and mental health of operators. Through effective noise monitoring and timely fault repair, the operating noise of the gearbox can be reduced, the acoustic environment of the workplace can be improved, and a relatively quiet and comfortable working condition can be created for operators, which helps to improve their job satisfaction and work enthusiasm and reduce occupational health problems caused by noise pollution. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Attached Figure 1 The present invention is a flow chart of a gear box noise monitoring method. DETAILED DESCRIPTION

[0034] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the application equally.

[0035] The present invention is a gearbox noise monitoring method, the main structure of which includes: sensor selection and installation: a high-precision capacitive acoustic sensor is selected, the frequency response range of which is 20Hz-20kHz, and the sensitivity reaches 50mV / Pa, which can meet the demand for accurate collection of gearbox noise signals. Four acoustic sensors are evenly arranged on the top, side and key parts of the gearbox such as the bearing seat, and the sensors are fixed by magnetic attraction or bolts to ensure that the sensors are firmly installed and accurately positioned. The sensing surface of the sensor faces the noise radiation direction of the gearbox to fully capture noise signals propagating in different directions.

[0036] Signal conditioning circuit design: The signal conditioning circuit is mainly composed of a preamplifier, a filter, and a post-amplifier. The preamplifier preliminarily amplifies the weak electrical signal output by the acoustic sensor, and the gain is set to 10 times to improve the signal-to-noise ratio. The filter uses an active Butterworth filter, and the cutoff frequency of the low-pass filter is designed to be 10kHz, and the cutoff frequency of the high-pass filter is 30Hz, which is used to remove high-frequency noise and low-frequency interference signals in the signal, such as electromagnetic interference in the environment and low-frequency vibration signals of the equipment. The post-amplifier further amplifies the filtered signal by 5 times to make its amplitude range suitable for subsequent analog-to-digital conversion processing.

[0037] Data acquisition and transmission system: A high-speed data acquisition card with 16-bit resolution and a maximum sampling rate of 100kHz is selected to convert the analog noise signal after signal conditioning into a digital signal. The data acquisition card is connected to the industrial control computer through a USB interface to achieve high-speed data transmission. A specially developed data acquisition software is installed on the industrial control computer, which can control the sampling parameter setting, data storage and preliminary data visualization of the data acquisition card in real time.

[0038] Noise Monitoring Process

[0039] Data acquisition: After starting the data acquisition system, set the sampling frequency of the data acquisition card to 40kHz to ensure that the various frequency components in the gearbox noise signal can be fully acquired. The noise signal collected by each acoustic sensor is transmitted to the signal conditioning circuit through an independent signal channel for amplification and filtering, and then the data acquisition card performs analog-to-digital conversion synchronously, and transmits the digital signal to the industrial control computer for subsequent processing.

[0040] Data preprocessing: The data preprocessing program running on the industrial control computer first denoises the collected digital noise signal, using the wavelet threshold denoising algorithm, selecting the appropriate wavelet basis function and threshold rule according to the characteristics of the noise signal, removing the random noise in the signal and improving the signal quality. Then, the denoised signal is framed, with each frame length set to 50ms and the frame shift to 25ms, to facilitate subsequent feature extraction and analysis.

[0041] Feature extraction: Extract multiple feature parameters from the framed noise signal, including time domain features, frequency domain features, and time-frequency domain features.

[0042] Time domain characteristics: Calculate the root mean square value (RMS), crest factor (CF) and impulse factor (IF) of each frame signal. RMS reflects the energy of the signal, while CF and IF are more sensitive to the impact components in the signal and can effectively detect early faults of gears and bearings.

[0043] Frequency domain features: Perform fast Fourier transform (FFT) on each frame of signal to obtain its spectrum distribution, extract the spectrum center of gravity (SC), spectrum peak frequency (PF) and energy ratio (EBR) of each frequency band. These frequency domain features can reflect information such as the meshing frequency of the gear, the fault characteristic frequency of the bearing and other resonance frequencies, which helps to determine the type and location of the fault.

[0044] Time-frequency domain features: Wavelet packet transform is used to perform time-frequency analysis on the signal, and the energy entropy (WPEE) of each wavelet packet node is calculated. By analyzing the change of energy entropy, the energy distribution characteristics of the signal at different times and frequencies can be captured, which has a good effect on detecting non-stationary fault signals. At the same time, the Hilbert-Huang transform (HHT) can also be used to calculate the marginal spectrum characteristics of the signal to further reveal the time-frequency characteristics and fault information of the signal.

[0045] Fault diagnosis: The extracted feature parameters are combined into feature vectors and input into a pre-trained fault diagnosis model based on deep learning. The model adopts a convolutional neural network (CNN) structure, which contains multiple convolutional layers, pooling layers, and fully connected layers. The convolutional layer extracts local features in the feature vector through the convolution kernel, the pooling layer performs downsampling to reduce the amount of data and improve the robustness of the features, and the fully connected layer is used to classify the features and output the diagnostic results of the fault type and fault degree.

[0046] The training process of the fault diagnosis model is as follows:

[0047] A large number of gearbox noise data samples under different working conditions are collected, and the above-mentioned data preprocessing and feature extraction operations are performed on these samples to form a training data set and a test data set. The ratio of the training data set to the test data set is 8:2.

[0048] The CNN model is trained using the training data set, and the stochastic gradient descent algorithm (SGD) is used to optimize the weight parameters of the model. The learning rate is set to 0.001, the momentum is set to 0.9, and the cross entropy loss function is used as the loss function of the model. During the training process, after each training cycle, the performance of the model is evaluated using the test data set, and evaluation indicators such as accuracy and recall are calculated. When the accuracy of the model on the test data set reaches more than 95% and the loss function value tends to be stable, stop training and save the trained model parameters.

[0049] Fault alarm and remote monitoring: Based on the output results of the fault diagnosis model, if the gearbox is judged to be faulty, the industrial control computer will immediately sound an alarm through the sound and light alarm device to alert the on-site operator. At the same time, the fault information will be sent to the remote monitoring center via Ethernet or wireless network.

[0050] The remote monitoring center is equipped with a server and a dedicated monitoring software platform. The server receives fault information from multiple gearbox monitoring systems and stores it in the database. The monitoring software platform can display the operating status, noise spectrum, fault alarm information and trend charts of historical monitoring data of each gearbox in real time, which is convenient for technicians to centrally manage and remotely diagnose gearboxes distributed in different locations. Based on the information provided by the remote monitoring platform, technicians can further analyze the cause of the fault, formulate a detailed maintenance plan, and promptly notify on-site maintenance personnel to repair the fault.

[0051] System self-diagnosis and maintenance: To ensure the reliability and stability of the entire noise monitoring system, the system has a self-diagnosis function. Regularly test hardware devices such as acoustic sensors, signal conditioning circuits, data acquisition cards, and industrial control computers. By sending specific test signals and checking the feedback of the signals, it is determined whether each device is working properly. If a device is found to be faulty, the backup device is immediately started, a system fault alarm is issued, and maintenance personnel are notified to repair it. At the same time, the software system is regularly updated and optimized to repair possible vulnerabilities and faults to ensure the accuracy and stability of data collection, processing, and transmission.

Claims

1. A gearbox noise monitoring method, characterized in that: S1. Install acoustic sensors at predetermined locations around the gearbox, the arrangement of the acoustic sensors being able to cover the main noise radiation area of ​​the gearbox, and the sensors having a wide frequency response characteristic, capable of collecting noise signals in the frequency range of [X] Hz to [Y] Hz, where X and Y are specific values ​​determined according to the operating frequency range of the gearbox; S2. Using the acoustic sensor to collect the noise signal generated by the gearbox during operation in real time, and transmitting the simulated noise signal to the signal conditioning circuit; S3. The signal conditioning circuit amplifies and filters the input analog noise signal, with an amplification factor of [Z] times. The filtering adopts a combination of low-pass, high-pass and band-pass filters to remove electromagnetic interference, environmental background noise and invalid frequency components beyond the normal working frequency range of the gearbox in the signal, and obtains a conditioned analog noise signal; S4. Converting the conditioned analog noise signal into a digital noise signal through an analog-to-digital converter, with a sampling frequency of not less than [F] Hz, to ensure that detailed information in the noise signal can be accurately captured, and the digital noise signal is transmitted to a microprocessor; S5. The microprocessor preprocesses the digital noise signal, uses a wavelet transform algorithm to remove residual noise in the signal, and performs frame processing on the signal. The duration of each frame is [M] ms, and the frame shift is [N] ms, to obtain a preprocessed noise frame signal; S6. Extract characteristic parameters from the preprocessed noise frame signal, including time domain characteristics of the noise signal such as root mean square value, peak factor, pulse factor, frequency domain characteristics such as spectrum center of gravity, spectrum peak frequency, energy proportion of each frequency band, and time-frequency domain characteristics such as wavelet packet energy entropy and Hilbert-Huang transform marginal spectrum characteristics; S7. Inputting the extracted characteristic parameters into a pre-trained fault diagnosis model, wherein the fault diagnosis model is constructed based on a machine learning algorithm or a deep learning algorithm, and is trained and learned on a large amount of noise characteristic data containing different gearbox operating conditions and fault types, so as to realize the identification and judgment of the gearbox fault type and fault degree; S8. Determine the operating status of the gearbox based on the output results of the fault diagnosis model. If it is judged to be a fault state, send the fault information to the remote monitoring terminal through the communication module. The fault information includes the fault type, fault severity, fault occurrence time and corresponding noise characteristic parameters. The communication module adopts wired communication or wireless communication.

2. A gearbox noise monitoring method according to claim 1, characterized in that: The acoustic sensor is a capacitive microphone or a piezoelectric sensor, which can work stably within an operating temperature range of [Ta]°C to [Tb]°C and has good vibration resistance, wherein Ta and Tb are specific parameter values ​​that meet the actual working environment requirements of the gearbox.

3. A gearbox noise monitoring method according to claim 1, characterized in that: The training process of the fault diagnosis model includes dividing the collected historical noise feature data into a training set, a validation set and a test set according to a certain ratio, optimizing the hyperparameters of the model by a cross-validation method, and completing the model training when the accuracy of the model on the test set reaches a predetermined threshold [P]% or more, where P is an accuracy index value set according to actual application requirements.

4. A gearbox noise monitoring method according to claim 1, characterized in that: Before the microprocessor preprocesses the digital noise signal, it also includes a data verification step, which uses a verification algorithm to check the integrity and rationality of the collected digital noise signal, eliminates abnormal data points, and supplements missing data using linear interpolation or based on the statistical laws of adjacent frame data to ensure data continuity and reliability.

5. The gearbox noise monitoring method according to claim 1, characterized in that: The remote monitoring terminal is provided with a visual interface, which can display the noise spectrum diagram, operation status information, fault alarm information and trend chart of historical monitoring data of the gearbox in real time, so as to facilitate the operator to intuitively understand the working condition and fault occurrence law of the gearbox. The visual interface also supports data export and printing functions for further data analysis and report generation.

6. A gearbox noise monitoring method according to claim 1, characterized in that: The microprocessor can also perform real-time evaluation of the operating status of the gearbox according to preset threshold conditions. When the extracted noise characteristic parameters exceed the normal threshold range, the local sound and light alarm device is immediately activated to remind the on-site operator to pay attention to the operation of the gearbox in time. The alarm volume of the sound and light alarm device is not less than [V]dB, and the alarm light color is eye-catching red or yellow, where V is the volume value that can attract the attention of the operator under the on-site environmental noise conditions.

7. A gearbox noise monitoring method according to claim 1, characterized in that: The amplifier circuit in the signal conditioning circuit adopts a programmable gain amplifier, which can automatically adjust the amplification factor according to the actual amplitude of the noise signal to ensure the accuracy and resolution of the signal during the analog-to-digital conversion process, while avoiding signal overload distortion; the filter parameters of the filtering circuit can be adjusted in real time by a microprocessor to adapt to the changes in the frequency characteristics of the gearbox noise signal under different working conditions.

8. The gearbox noise monitoring method according to claim 1, characterized in that: The method also includes a self-diagnosis function for the gearbox noise monitoring system. During the operation of the system, hardware devices such as acoustic sensors, signal conditioning circuits, analog-to-digital converters, microprocessors and communication modules are regularly tested.

Images