Multi-signal combined fault diagnosis method and system for wind power variable pitch bearing

Abstract

The invention discloses a multi-signal joint fault diagnosis method and system for a wind power variable pitch bearing. The multi-signal combined fault diagnosis method for the wind power variable pitch bearing comprises the following steps: signal acquisition: acquiring a vibration signal, a strain signal, a temperature signal and a sound signal during operation of the variable pitch bearing; signal denoising: transmitting the collected vibration signal, strain signal, temperature signal and sound signal to a signal denoising module for variational mode decomposition; signal processing: transmitting at least one of the denoised vibration signal, strain signal, temperature signal and sound signal to a signal processing module to obtain a corresponding two-dimensional time-frequency image; and multi-signal combined fault diagnosis: transmitting the plurality of one-dimensional time-domain signals and the at least one two-dimensional time-frequency image to a combined fault diagnosis module, and carrying out combined fault diagnosis through a plurality of one-dimensional convolutional neural networks and a two-dimensional convolutional neural network, thereby realizing multi-signal combined fault diagnosis and improving the recognition rate of fault diagnosis.

Description

Technical Field

[0001] The present invention relates to the technical field of wind turbine pitch bearing fault diagnosis, and in particular to a multi-signal joint fault diagnosis method and system for a wind turbine pitch bearing. Background Art

[0002] With the development and progress of science and technology, wind power generation has been popularized and applied on a large scale. Wind turbines are important equipment for wind power generation and play an indispensable role in the wind power industry. Wind turbines contain multiple systems such as blade rotor system, pitch system, communication system, transmission system, etc. Among them, pitch bearings are one of the core components of wind turbines and play a very important role in the stable and efficient operation of wind turbines. In the situation where new technologies and industries continue to emerge, in order to respond to the national call, cater to market demand, keep up with the trend, and keep pace with the times, wind turbines are developing in the direction of intelligence and unmanned operation to improve their safety and achieve higher production efficiency.

[0003] Wind turbines are generally installed in high mountains, grasslands, coastal areas and other areas. The working environment is sometimes very harsh, and they may face harsh environments such as lightning, sandstorms, strong winds, and heavy rains. At the same time, the large temperature difference between day and night and extreme climate conditions will also have a bad impact on the stability of wind turbine operation. The weight of the blades of wind turbines ranges from several tons to dozens of tons. During operation, the load on the wind turbine is very large, and the wind turbine often faces strong vibration and impact during operation, which means that the drive system, load-bearing capacity, impact resistance and related components of the mining truck need to meet strict requirements. During the operation of the wind turbine, it will experience extreme environments, temperature differences, high loads, severe vibrations and impacts, and long-term operation. The above working environment will affect the normal operation of the wind turbine. The variable pitch bearing of the wind turbine will also be greatly affected. According to statistics, the failure rate of the variable pitch bearing is as high as 21%, and the failure of the variable pitch bearing will cause chain failures in other components, resulting in great economic losses. Therefore, it is very necessary to obtain the damage status of the variable pitch bearing and perform fault diagnosis. Summary of the invention

[0004] The purpose of the present invention is to provide a multi-signal joint fault diagnosis method and system for a wind turbine pitch bearing, so as to capture and analyze the vibration, strain, sound, temperature and other signals of the pitch bearing in real time, perform multi-signal joint fault diagnosis of the pitch bearing, and improve the recognition rate of fault diagnosis.

[0005] In order to solve the above technical problems, an embodiment of the present invention provides a multi-signal joint fault diagnosis method for a wind turbine pitch bearing, comprising the following steps:

[0006] Signal collection: collect vibration signals, strain signals, temperature signals and sound signals during the operation of the pitch bearing;

[0007] Signal denoising: The collected vibration signal, strain signal, temperature signal and sound signal are transmitted to the signal denoising module for variational mode decomposition;

[0008] Signal processing: transmitting at least one of the denoised vibration signal, strain signal, temperature signal and sound signal to a signal processing module to perform wavelet transform on the one-dimensional time domain signal to obtain a corresponding two-dimensional time-frequency image;

[0009] Multi-signal joint fault diagnosis: The one-dimensional time domain signals of vibration signals, strain signals, temperature signals and sound signals and at least one two-dimensional time-frequency image are transmitted to the joint fault diagnosis module for joint fault diagnosis through multiple one-dimensional convolutional neural networks and two-dimensional convolutional neural networks.

[0010] Compared with the prior art, the embodiments of the present invention obtain the vibration signal, strain signal, temperature signal and sound signal of the variable pitch bearing in real time through vibration sensors, strain sensors, temperature sensors and sound sensors, and perform real-time analysis on the operating status of the variable pitch bearing. The method adopted is reasonable and reliable, and the authenticity of the data is guaranteed. The vibration, strain, temperature and sound signal monitoring and fault diagnosis are integrated into one, and abnormalities or potential faults can be discovered in time. Moreover, the combined judgment of multiple signals can effectively improve the recognition rate of fault diagnosis.

[0011] In one embodiment, the step transmits at least one of the denoised vibration signal, strain signal, temperature signal and sound signal to a signal processing module, performs wavelet transform on the one-dimensional time domain signal to obtain a corresponding two-dimensional time-frequency image, performs wavelet transform on the vibration signal and strain signal to obtain a corresponding two-dimensional vibration signal and two-dimensional strain signal.

[0012] In one embodiment, the step of performing joint fault diagnosis on multiple one-dimensional time domain signals and at least one two-dimensional time-frequency image after processing vibration signals, strain signals, temperature signals and sound signals comprises a network architecture including four one-dimensional convolutional neural networks and two two-dimensional convolutional neural networks.

[0013] In one embodiment, the step of signal denoising specifically includes:

[0014] Initialize the parameters of variational mode decomposition;

[0015] Decomposing the vibration signal, strain signal, temperature signal and sound signal into a number of intrinsic mode functions, each of which represents a characteristic mode of the vibration signal, strain signal, temperature signal and sound signal;

[0016] By using variational inference, the frequency and bandwidth of each intrinsic mode function are optimized so that the frequency distribution of each intrinsic mode function is concentrated, thereby separating the different components of vibration signals, strain signals, temperature signals and sound signals;

[0017] The vibration signal, strain signal, temperature signal and sound signal are reconstructed according to the intrinsic mode function to obtain the vibration signal, strain signal, temperature signal and sound signal after noise reduction.

[0018] In one embodiment, the wavelet transform formula is as follows:

[0019]

[0020]

[0021] Where a is the scale factor; τ is the translation factor; is the wavelet basis function.

[0022] In one embodiment, the joint fault diagnosis module uses the loss function after the training set training model reaches the standard to define the weight of the output value of each model output result, and the formula is as follows:

[0023]

[0024] In the formula, ω k is the weight corresponding to the kth signal, loss i is the loss value of the network corresponding to the i-th signal.

[0025] In one embodiment, the output result formula of the joint fault diagnosis module is as follows:

[0026] result=∑ω k y k

[0027] In the formula, y k is the output of the network corresponding to the kth signal.

[0028] In one embodiment, a vibration sensor is provided on the outer ring of the wind turbine pitch bearing to collect the vibration signal of the bearing;

[0029] The outer ring of the wind turbine pitch bearing is provided with a fiber grating sensor to collect the strain signal of the bearing;

[0030] A sound sensor is arranged on the outer ring of the wind turbine pitch bearing to collect sound signals;

[0031] A temperature sensor is arranged on the outer ring of the wind turbine pitch bearing to collect temperature signals.

[0032] The embodiment of the present invention further provides a multi-signal joint fault diagnosis system for a wind turbine pitch bearing, comprising: a vibration sensor for collecting vibration signals, a fiber grating sensor for collecting strain signals, a sound sensor for collecting sound signals, a temperature sensor for collecting temperature signals, a vibration signal collector for collecting vibration signals, a strain signal collector for collecting strain signals, a sound signal collector for collecting sound signals, a temperature signal collector for collecting temperature signals, a signal noise reduction module, a signal processing module and a joint fault diagnosis module;

[0033] The vibration sensor, the fiber grating sensor, the sound sensor and the temperature sensor are arranged on the outer ring of the wind turbine pitch bearing;

[0034] The vibration sensor is electrically connected to the vibration signal collector, the signal noise reduction module, the signal processing module and the joint fault diagnosis module;

[0035] The fiber grating sensor is electrically connected to the strain signal acquisition instrument, the signal noise reduction module, the signal processing module and the joint fault diagnosis module;

[0036] The sound sensor is electrically connected to the sound signal collector, the signal noise reduction module, the signal processing module and the joint fault diagnosis module;

[0037] The temperature sensor is electrically connected to the temperature signal acquisition instrument, the signal noise reduction module, the signal processing module and the joint fault diagnosis module;

[0038] The signal denoising module is used to perform variational modal decomposition on the collected vibration signals, strain signals, temperature signals and sound signals; the signal processing module is used to perform wavelet transform on at least one of the denoised vibration signals, strain signals, temperature signals and sound signals to obtain a corresponding two-dimensional time-frequency image; the joint fault diagnosis module is used to perform joint fault diagnosis on the one-dimensional time domain signals and at least one two-dimensional time-frequency image of the received vibration signals, strain signals, temperature signals and sound signals through multiple one-dimensional convolutional neural networks and two-dimensional convolutional neural networks.

[0039] An embodiment of the present invention also provides a computer-readable storage medium, which is a non-volatile storage medium or a non-transient storage medium, on which a computer program is stored. When the computer program is run by a processor, the multi-signal joint fault diagnosis method for a wind turbine pitch bearing as described in any one of the above is executed. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] One or more embodiments are exemplarily described by pictures in the corresponding drawings, and these exemplified descriptions do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings represent similar elements, and unless otherwise stated, the figures in the drawings do not constitute proportional limitations.

[0041] Figure 1 is a flow chart of a multi-signal joint fault diagnosis method according to a first embodiment of the present invention;

[0042] Figure 2 is a schematic diagram of signal transmission of a multi-signal joint fault diagnosis system according to a first embodiment of the present invention;

[0043] Figure 3 is a schematic diagram of a multi-signal joint fault diagnosis network structure according to a first embodiment of the present invention;

[0044] Figure 4 4 is a flowchart of the specific steps of multi-signal joint fault diagnosis according to the first embodiment of the present invention. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical scheme and advantages of the embodiments of the present invention clearer, the various embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings. However, it will be appreciated by those skilled in the art that in the various embodiments of the present invention, many technical details are provided in order to enable the reader to better understand the present application. However, even without these technical details and various changes and modifications based on the following embodiments, the technical scheme claimed in the present application can be implemented.

[0046] In the following description, certain specific details are set forth for the purpose of illustrating various disclosed embodiments to provide a thorough understanding of the various disclosed embodiments. However, those skilled in the relevant art will recognize that the embodiments may be practiced without one or more of these specific details. In other cases, well-known devices, structures, and techniques associated with the present application may not be shown or described in detail to avoid unnecessarily obscuring the description of the embodiments.

[0047] Unless the context requires otherwise, throughout the specification and claims, the word "comprise" and variations such as "include" and "have" should be construed in an open, inclusive sense, ie, should be interpreted as "including, but not limited to."

[0048] The following will be described in detail with reference to the accompanying drawings to provide a clearer understanding of the purpose, features and advantages of the present invention. It should be understood that the embodiments shown in the accompanying drawings are not intended to limit the scope of the present invention, but are only intended to illustrate the essential spirit of the technical solution of the present invention.

[0049] References throughout the specification to "one embodiment" or "an embodiment" indicate that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. Thus, the appearances of "in one embodiment" or "in an embodiment" in various places throughout the specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any manner in one or more embodiments.

[0050] As used in this specification and the appended claims, the singular forms "a," "an," and "the" include plural referents unless the context clearly dictates otherwise. It should be noted that the term "or" is generally employed in its sense including "and / or" unless the context clearly dictates otherwise.

[0051] In the following description, in order to clearly show the structure and working mode of the present invention, many directional words will be used for description, but the words "front", "back", "left", "right", "outside", "inside", "outward", "inward", "up", "down", etc. should be understood as convenient terms and should not be understood as restrictive terms.

[0052] At present, there are few publicly disclosed technologies related to multi-signal real-time monitoring and fault diagnosis of wind turbine pitch bearings. Patent No. CN118190419A discloses a method for fault diagnosis of wind turbine pitch bearings based on vibration signals. The method extracts fault features by reconstructing vibration signals, frequency band segmentation, etc. However, the method only uses the vibration signal of the pitch bearing, which is easily interfered and has certain limitations; Patent No. CN114235415B discloses a method and device for fault diagnosis of wind turbine pitch bearings based on neural networks. The method collects and processes vibration data, and deploys the trained neural network model to the PLC for real-time monitoring and fault diagnosis. However, the method only diagnoses faults through the vibration signal of the pitch bearing, and fault identification is difficult; Patent No. CN114017267B discloses a device for fault diagnosis of wind turbine pitch bearings. When the bearing is deformed and tilted, the monitoring unit of the device monitors and reflects it to the control center for alarm. However, the pitch bearing faults targeted by the method are relatively single, and the specific faults of the pitch bearing cannot be determined, and the specific operating status of the pitch bearing cannot be reflected.

[0053] Developing a wind turbine pitch bearing vibration, strain, sound and other multi-signal online monitoring and multi-signal fault diagnosis system and method to enable the wind turbine to monitor the operating status of its own pitch bearing and accurately identify pitch bearing faults can improve equipment safety and operating efficiency and save equipment maintenance costs, which is of great significance to the field of wind power generation.

[0054] Embodiments of the present invention are described below with reference to the accompanying drawings.

[0055] An embodiment of the present invention relates to a multi-signal joint fault diagnosis method for a wind turbine variable pitch bearing. A multi-signal joint fault diagnosis system for a wind turbine variable pitch bearing is used to perform real-time monitoring on the variable pitch bearing. The multi-signal joint fault diagnosis system includes a vibration sensor, a fiber grating sensor, a sound sensor, a temperature sensor, a connecting line, and a corresponding signal acquisition instrument. The vibration sensor is installed on the outer ring of the variable pitch bearing by magnetic attraction to collect the vibration signal of the bearing, the fiber grating sensor is installed on the outer ring of the bearing by gluing to collect the strain signal of the bearing, a microphone array is used to collect the sound signal, and a microphone and a temperature sensor are installed on the outer ring of the bearing to collect the sound and temperature signals. After the signal is collected, it is transmitted to the corresponding acquisition instrument through a connecting line and converted into corresponding data. In the multi-signal joint fault diagnosis system, signal processing, signal noise reduction, and multi-signal joint fault diagnosis need to be completed, and finally the diagnosis result is fed back to the wind power system. If a fault occurs, a timely warning is issued.

[0056] Specifically, Figure 1 As shown, the multi-signal joint fault diagnosis method for wind turbine pitch bearings includes the following steps:

[0057] Step 110, signal collection: collecting vibration signals, strain signals, temperature signals and sound signals during the operation of the variable pitch bearing;

[0058] Step 120, signal denoising: transmitting the collected vibration signal, strain signal, temperature signal and sound signal to a signal denoising module for variational mode decomposition;

[0059] Step 130, signal processing: transmitting at least one of the denoised vibration signal, strain signal, temperature signal and sound signal to a signal processing module to perform wavelet transform on the one-dimensional time domain signal to obtain a corresponding two-dimensional time-frequency image;

[0060] Step 140, multi-signal joint fault diagnosis: transmit the one-dimensional time domain signal of the vibration signal, strain signal, temperature signal and sound signal and at least one two-dimensional time-frequency image to the joint fault diagnosis module to perform joint fault diagnosis through multiple one-dimensional convolutional neural networks and two-dimensional convolutional neural networks.

[0061] Vibration sensors, strain sensors, temperature sensors and sound sensors are used to obtain the vibration signals, strain signals, temperature signals and sound signals of the variable pitch bearing in real time, and the operating status of the variable pitch bearing is analyzed in real time. The method adopted is reasonable and reliable, which ensures the authenticity of the data. The vibration, strain, temperature and sound signal monitoring and fault diagnosis are integrated into one, which can detect abnormalities or potential faults in time, and the combined judgment of multiple signals can effectively improve the recognition rate of fault diagnosis.

[0062] Furthermore, after completing the collection of vibration signals, strain signals, temperature signals and sound signals, the collected data is transmitted to the signal denoising module through a connecting line. The module includes a memory and a processor for storing programs and executing program operations. The signal denoising module performs denoising on the collected signals to reduce the noise components contained in the signals, thereby improving the accuracy of fault diagnosis. Step 120 signal denoising specifically includes:

[0063] Step 121, initializing parameters of variational mode decomposition (VMD);

[0064] Step 122, decomposing the vibration signal, strain signal, temperature signal and sound signal into a plurality of intrinsic mode functions (IMFs), each intrinsic mode function (IMF) representing a characteristic mode of the vibration signal, strain signal, temperature signal and sound signal;

[0065] Step 123, optimizing the frequency and bandwidth of each intrinsic mode function (IMF) by variational inference, so that the frequency distribution of each intrinsic mode function is concentrated, thereby separating different components of the vibration signal, strain signal, temperature signal and sound signal;

[0066] Step 124, reconstructing the vibration signal, strain signal, temperature signal and sound signal according to the intrinsic mode function (IMF) to obtain the vibration signal, strain signal, temperature signal and sound signal after noise reduction.

[0067] In this embodiment, variational mode decomposition (VMD) is used for noise reduction. The VMD algorithm determines the center frequency and bandwidth of each modal component by iteratively searching for the optimal solution of the variational model to achieve effective signal separation. The algorithm can effectively decompose according to the actual characteristics of the signal, has strong adaptability, can effectively separate the noise and useful components in the signal, and can effectively improve the signal-to-noise ratio. The specific steps are as follows:

[0068] (1) Construct k modes so that the sum of the estimated bandwidth of each mode is minimized.

[0069]

[0070] (2) Introduce the quadratic penalty factor α and the Lagrangian penalty operator λ(t).

[0071]

[0072] (3) Iterative Update and Obtain optimal solutions to constrained variational models.

[0073]

[0074] (4) Convert to the frequency domain through Fourier isometric transform.

[0075]

[0076] (5) Solved using the same principle Minimum problem, convert this problem to the frequency domain and solve it, and get the center frequency as:

[0077]

[0078] (6) Calculate the normalized correlation coefficient between each component and the signal.

[0079]

[0080] Where: r i is the correlation coefficient between the i-th component and the signal s; is the average value of the i-th component, i=1,2,…,K.

[0081] (7) The standard deviation ξ of the correlation coefficient is used to distinguish between effective signals and noise.

[0082]

[0083] If r i >ξ, then the component is a valid signal.

[0084] Furthermore, after the signal noise reduction process is completed, the connecting line transmits the noise-reduced signal to the signal processing module. The signal processing module mainly performs preprocessing operations on the signal. The preprocessing operation uses wavelet transform to convert the one-dimensional signal into a two-dimensional time-frequency image, thereby enriching the signal characteristics. The wavelet transform formula is as follows:

[0085]

[0086] Where a is the scale factor; τ is the translation factor; is the wavelet basis function.

[0087] Specifically, the one-dimensional signal can provide time series dynamic change information, while the two-dimensional signal can provide more comprehensive spatial feature information. In order to more comprehensively understand the system status and improve the accuracy and reliability of fault diagnosis, this embodiment uses one-dimensional signals and two-dimensional signals for fault diagnosis. Therefore, the next joint fault diagnosis module will receive the one-dimensional time domain signal from the noise reduction module and the two-dimensional image signal from the signal processing module.

[0088] In addition, the signal processed data is transmitted to the joint fault diagnosis module, which consists of a processor and a memory. The memory is used to store the joint fault diagnosis program, and the processor is used to execute the program. This module uses a one-dimensional convolutional neural network architecture and a two-dimensional convolutional neural network for joint fault diagnosis. The convolution operation process is as follows:

[0089]

[0090] In the formula, is the j′th weight value of the i-th convolution kernel in the l-th layer; x l(j) is the jth convolved local area of ​​the lth layer; y l(i,j) is the activation value of the convolution output; f(*) is the activation function; W is the width of the convolution kernel.

[0091] like Figure 2 As shown, in order to reduce the complexity of the model, only the vibration signal and the strain signal are processed, so the multi-signal joint diagnosis module only receives one-dimensional vibration signals, two-dimensional vibration signals, one-dimensional strain signals, two-dimensional strain signals, one-dimensional sound signals and one-dimensional temperature signals. The fault diagnosis modules used need to be trained separately before they can be used. Therefore, it is necessary to collect the vibration signals, strain signals and other signals of the variable pitch bearing in the actual operation of the wind turbine to obtain data samples in the healthy state and various fault conditions for model training. After receiving the data processed by the processing module, the joint fault diagnosis module performs fault diagnosis and judgment, and transmits the diagnosis results to the system platform through the connecting line after the diagnosis is completed. In this module, in order to make each signal reasonably used to improve the accuracy of fault diagnosis, the weight of the output value of each model output result is defined according to the loss function after the training model is trained using the training set. The formula is as follows:

[0092]

[0093] In the formula, ω k is the weight corresponding to the kth signal, loss i is the loss value of the network corresponding to the i-th signal.

[0094] The final output formula of the fault diagnosis system is as follows:

[0095] result=∑ω k y k

[0096] In the formula, y k is the output of the network corresponding to the kth signal.

[0097] In other embodiments, the signal processing may be different from that in this embodiment, and any one or more of the vibration signal, strain signal, temperature signal and sound signal may be processed.

[0098] like Figure 3 As shown, the network architecture in this embodiment includes 4 one-dimensional convolutional neural networks and 2 two-dimensional convolutional neural networks. Each network architecture includes multiple convolutional layers, pooling layers, fully connected layers and softmax output layers.

[0099] Specifically, the number of convolution kernels of the one-dimensional convolutional neural network is set to 64, 128 and 256 respectively, the size of the convolution kernel is 20×1, 10×1 and 5×1 respectively, the step size is 8, 4, 2, the pooling layer uses the maximum pooling, the number of convolution kernels is 64, 128 and 256 respectively, the size is 4×1, 4×1 and 2×1, the step size is 4, 4 and 2, the activation function uses the Relu function, the cross entropy loss function is used to train the weight parameters and bias parameters, and the regularization selection of the drop rate is set to 0.5 to prevent the model from overfitting. The nodes of the two fully connected layers are set to 500 and 100 respectively, and finally the soft-max function is used for classification.

[0100] The number of convolution kernels of the two-dimensional convolutional neural network is set to 64, 128 and 256 respectively, the convolution kernel size is 7×7, 3×3 and 3×3 respectively, the step size is 2, the pooling layer uses the maximum pooling, the number of convolution kernels is 64, 128 and 256 respectively, the step size is 4, 4 and 2, the activation function uses the Relu function, the cross entropy loss function is used to train the weight parameters and bias parameters, and the drop rate selected for regularization is set to 0.5 to prevent the model from overfitting. After a series of convolution operations, continuous layers (including average pooling, fully connected layers and softmax functions) are used to determine the fault type.

[0101] Each network has a corresponding loss function value. The weight value of the output value of each model's output result is defined by calculating the loss function after the training set training model reaches the standard. After assigning corresponding weights to the output results of each network, the joint fault diagnosis results are output.

[0102] like Figure 4As shown in the figure, the model needs to be trained before it can be used normally. First, the vibration signal, strain signal, temperature signal and sound signal of the variable pitch bearing in the actual operation of the wind turbine are collected to obtain data samples in the healthy state and under the conditions of outer ring fault, inner ring fault, rolling element fault and multiple faults of the rolling bearing. After signal denoising, the denoised signal is transmitted to the signal processing module, and the one-dimensional time series signal is converted into a two-dimensional image signal. After the collected data is randomly divided into a training set and a validation set in a ratio of 7:3, the training set is sent to the fault diagnosis model for training. After the network initialization parameters are established, the multi-signal joint diagnosis module is trained through feedforward propagation to calculate whether its loss parameters converge. If converged, the diagnosis results are output, otherwise, the network parameters will be continuously adjusted through back propagation. The fault diagnosis model is optimized using the SGD algorithm, the initial learning rate is set to 0.01, the number of iterations is 500, the model parameters are randomly initialized, the network parameters are updated, and the network is optimized. Finally, the test set data is sent to the optimized model for testing. If the recognition performance of the model meets the standard, the final fault diagnosis model is obtained. The joint fault diagnosis module performs fault diagnosis after receiving various signal data and transmits the final diagnosis result to the system platform.

[0103] The second embodiment of this embodiment relates to a multi-signal joint fault diagnosis system for a wind turbine variable pitch bearing. The system can adopt the method in the first embodiment, including: a vibration sensor for collecting vibration signals, a fiber Bragg grating sensor for collecting strain signals, a sound sensor for collecting sound signals, a temperature sensor for collecting temperature signals, a vibration signal collector for collecting vibration signals, a strain signal collector for collecting strain signals, a sound signal collector for collecting sound signals, a temperature signal collector for collecting temperature signals, a signal noise reduction module, a signal processing module and a joint fault diagnosis module. The vibration sensor, the fiber Bragg grating sensor, the sound sensor and the temperature sensor are arranged on the outer ring of the wind turbine variable pitch bearing. The vibration sensor is electrically connected to the vibration signal collector, the signal noise reduction module, the signal processing module and the joint fault diagnosis module; the fiber Bragg grating sensor is electrically connected to the strain signal collector, the signal noise reduction module, the signal processing module and the joint fault diagnosis module; the sound sensor is electrically connected to the sound signal collector, the signal noise reduction module, the signal processing module and the joint fault diagnosis module; the temperature sensor is electrically connected to the temperature signal collector, the signal noise reduction module, the signal processing module and the joint fault diagnosis module. The signal denoising module is used to perform variational modal decomposition on the collected vibration signals, strain signals, temperature signals and sound signals; the signal processing module is used to perform wavelet transform on at least one of the denoised vibration signals, strain signals, temperature signals and sound signals to obtain a corresponding two-dimensional time-frequency image; the joint fault diagnosis module is used to perform joint fault diagnosis on the one-dimensional time domain signals and at least one two-dimensional time-frequency image of the received vibration signals, strain signals, temperature signals and sound signals through multiple one-dimensional convolutional neural networks and two-dimensional convolutional neural networks.

[0104] It is not difficult to find that this embodiment is a system embodiment corresponding to the first embodiment, and this embodiment can be implemented in conjunction with the first embodiment. The relevant technical details mentioned in the first embodiment are still valid in this embodiment, and in order to reduce repetition, they are not repeated here. Accordingly, the relevant technical details mentioned in this embodiment can also be applied in the first embodiment.

[0105] The third embodiment of the present embodiment relates to a computer-readable storage medium, which is a non-volatile storage medium or a non-transient storage medium, on which a computer program is stored. When the computer program is run by a processor, the multi-signal joint fault diagnosis method for wind turbine pitch bearings in the first embodiment is executed.

[0106] That is, those skilled in the art can understand that all or part of the steps in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a program, and the program is stored in a storage medium, including several instructions to enable a device (which can be a single-chip microcomputer, chip, etc.) or a processor to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0107] Preferred embodiments of the present invention have been described above in detail, but it should be understood that aspects of the embodiments can be modified, if necessary, to employ aspects, features and concepts of the various patents, applications and publications to provide further embodiments.

[0108] These and other changes can be made to the embodiments in light of the above detailed description.In general, in the claims, the terms used should not be considered limited to the specific embodiments disclosed in the specification and the claims, but should be construed to include all possible embodiments along with the full scope of equivalents to which these claims are entitled.

[0109] Those skilled in the art will appreciate that the above embodiments are specific embodiments for implementing the present invention, and that in actual applications, various changes may be made in form and detail without departing from the spirit and scope of the present invention.

Claims

1. A multi-signal joint fault diagnosis method for wind turbine pitch bearings, characterized in that: The steps include: Signal collection: collect vibration signals, strain signals, temperature signals and sound signals during the operation of the pitch bearing; Signal denoising: The collected vibration signal, strain signal, temperature signal and sound signal are transmitted to the signal denoising module for variational mode decomposition; Signal processing: transmitting at least one of the denoised vibration signal, strain signal, temperature signal and sound signal to a signal processing module to perform wavelet transform on the one-dimensional time domain signal to obtain a corresponding two-dimensional time-frequency image; Multi-signal joint fault diagnosis: The one-dimensional time domain signals of vibration signals, strain signals, temperature signals and sound signals and at least one two-dimensional time-frequency image are transmitted to the joint fault diagnosis module for joint fault diagnosis through multiple one-dimensional convolutional neural networks and two-dimensional convolutional neural networks.

2. The multi-signal joint fault diagnosis method for wind turbine pitch bearing according to claim 1 is characterized in that: The step is to transmit at least one of the denoised vibration signal, strain signal, temperature signal and sound signal to a signal processing module to perform wavelet transform on the one-dimensional time domain signal to obtain a corresponding two-dimensional time-frequency image, and to perform wavelet transform on the vibration signal and strain signal to obtain a corresponding two-dimensional vibration signal and two-dimensional strain signal.

3. The multi-signal joint fault diagnosis method for wind turbine pitch bearing according to claim 2 is characterized in that: The step performs joint fault diagnosis on multiple one-dimensional time domain signals and at least one two-dimensional time-frequency image after processing vibration signals, strain signals, temperature signals and sound signals. The network architecture includes four one-dimensional convolutional neural networks and two two-dimensional convolutional neural networks.

4. The multi-signal joint fault diagnosis method for wind turbine pitch bearing according to claim 1 is characterized in that: The steps in signal denoising specifically include: Initialize the parameters of variational mode decomposition; Decomposing the vibration signal, strain signal, temperature signal and sound signal into a number of intrinsic mode functions, each of which represents a characteristic mode of the vibration signal, strain signal, temperature signal and sound signal; By variational inference, the frequency and bandwidth of each intrinsic mode function are optimized so that the frequency distribution of each intrinsic mode function is concentrated, thereby separating the different components of vibration signals, strain signals, temperature signals and sound signals; The vibration signal, strain signal, temperature signal and sound signal are reconstructed according to the intrinsic mode function to obtain the vibration signal, strain signal, temperature signal and sound signal after noise reduction.

5. The multi-signal joint fault diagnosis method for wind turbine pitch bearing according to claim 1 is characterized in that: The wavelet transform formula is as follows: Where a is the scale factor; τ is the translation factor; is the wavelet basis function.

6. The multi-signal joint fault diagnosis method for wind turbine pitch bearing according to claim 1 is characterized in that: The joint fault diagnosis module uses the loss function after the training set training model reaches the standard to define the weight of the output value of each model output result. The formula is as follows: In the formula, ω k is the weight corresponding to the kth signal, loss i is the loss value of the network corresponding to the i-th signal.

7. The multi-signal joint fault diagnosis method for wind turbine pitch bearing according to claim 6 is characterized in that: The output result formula of the joint fault diagnosis module is as follows: result=∑ω k y k In the formula, y k is the output of the network corresponding to the kth signal.

8. The multi-signal joint fault diagnosis method for wind turbine pitch bearing according to claim 1 is characterized in that: A vibration sensor is provided on the outer ring of the wind turbine pitch bearing to collect the vibration signal of the bearing; The outer ring of the wind turbine pitch bearing is provided with a fiber grating sensor to collect the strain signal of the bearing; A sound sensor is arranged on the outer ring of the wind turbine pitch bearing to collect sound signals; A temperature sensor is arranged on the outer ring of the wind turbine pitch bearing to collect temperature signals.

9. A multi-signal joint fault diagnosis system for wind turbine pitch bearings, characterized in that: include: A vibration sensor for collecting vibration signals, a fiber grating sensor for collecting strain signals, a sound sensor for collecting sound signals, a temperature sensor for collecting temperature signals, a vibration signal collector for collecting vibration signals, a strain signal collector for collecting strain signals, a sound signal collector for collecting sound signals, a temperature signal collector for collecting temperature signals, a signal noise reduction module, a signal processing module and a joint fault diagnosis module; The vibration sensor, the fiber grating sensor, the sound sensor and the temperature sensor are arranged on the outer ring of the wind turbine pitch bearing; The vibration sensor is electrically connected to the vibration signal collector, the signal noise reduction module, the signal processing module and the joint fault diagnosis module; The fiber grating sensor is electrically connected to the strain signal acquisition instrument, the signal noise reduction module, the signal processing module and the joint fault diagnosis module; The sound sensor is electrically connected to the sound signal collector, the signal noise reduction module, the signal processing module and the joint fault diagnosis module; The temperature sensor is electrically connected to the temperature signal acquisition instrument, the signal noise reduction module, the signal processing module and the joint fault diagnosis module; The signal denoising module is used to perform variational modal decomposition on the collected vibration signals, strain signals, temperature signals and sound signals; the signal processing module is used to perform wavelet transform on at least one of the denoised vibration signals, strain signals, temperature signals and sound signals to obtain a corresponding two-dimensional time-frequency image; the joint fault diagnosis module is used to perform joint fault diagnosis on the one-dimensional time domain signals and at least one two-dimensional time-frequency image of the received vibration signals, strain signals, temperature signals and sound signals through multiple one-dimensional convolutional neural networks and two-dimensional convolutional neural networks.

10. A computer-readable storage medium, wherein the computer-readable storage medium is a non-volatile storage medium or a non-transient storage medium, and a computer program is stored thereon, wherein: When the computer program is executed by the processor, the multi-signal joint fault diagnosis method for a wind turbine pitch bearing as described in any one of claims 1 to 8 is executed.

Images