Generator Bearing Deterioration Evaluation Method and System Based on Metal Self-Repair Technology

Abstract

This application relates to the technical field of bearing deterioration analysis, and discloses a method and system for evaluating the deterioration of a generator bearing based on metal self-repair technology. The method includes: collecting bearing vibration, balance, and temperature data; extracting balance and wear characteristics, and establishing a deterioration index system; injecting self-repair materials; monitoring the repair effect and obtaining improvement data; predicting the bearing health status through a memory network; generating an evaluation report, optimizing the repair strategy, and predicting the remaining life. This application can accurately evaluate the bearing health status, realize the personalized optimization of the metal self-repair material formula, monitor the entire repair process and evaluate the effect, and finally achieve the accurate prediction of the remaining service life of the bearing, improving the scientificity and efficiency of the bearing maintenance of wind turbines.

Description

Technical Field

[0001] This application relates to the technical field of bearing deterioration analysis, and particularly to a method and system for evaluating the deterioration of a generator bearing based on metal self-repair technology. Background Art

[0002] With the rapid development of the wind power industry, the reliability problems of the main shaft bearings and generator bearings of wind turbines have become increasingly prominent. Wind turbine bearings operate under high loads and variable speed conditions for a long time, and are prone to problems such as wear, fatigue, and dynamic balance damage. According to the research of the World Tribology Congress, the energy loss caused by friction during equipment operation exceeds 30%. As key components, the damage of the main shaft bearings and generator bearings of wind turbines directly affects the power generation efficiency and equipment reliability. Traditional bearing maintenance methods mainly rely on regularly replacing lubricating grease or disassembling the bearings for repair, which not only costs a lot but also requires shutdown maintenance, seriously affecting the economic benefits of wind farms. In recent years, as a new type of in-situ repair technology for friction surfaces, metal self-repair technology has received extensive attention. By adding nanomaterials with special properties to the lubricating grease, a metal ceramic protective layer is formed under the action of friction heat and pressure to achieve self-repair of the bearing surface.

[0003] However, the existing metal self-repair technologies for wind power bearings have obvious defects and limitations. First, there is a lack of scientific and systematic methods for evaluating the health status of bearings, and it is impossible to accurately judge the degree of bearing damage and repair requirements. Second, the formulation ratios of metal self-repair materials often adopt a unified standard and are not adjusted individually according to the damage characteristics of different bearings. Third, the repair process lacks effective monitoring, making it difficult to judge the repair effect and progress. Finally, the prediction method for the remaining service life of the repaired bearing is imperfect, resulting in a lack of scientific basis for formulating maintenance plans. These problems lead to uneven application effects of metal self-repair technology in wind power bearing maintenance, unable to fully exert its technical potential, and restricting the further reduction of wind farm operation and maintenance costs. Summary of the Invention

[0004] This application provides a method and system for evaluating the deterioration of a generator bearing based on metal self-repair technology, which is used to accurately evaluate the health status of the bearing, realize personalized optimization of the metal self-repair material formula, monitor the entire repair process and evaluate the effect, and finally achieve accurate prediction of the remaining service life of the bearing, improving the scientificity and efficiency of wind turbine bearing maintenance.

[0005] In a first aspect, the present application provides a method for evaluating the deterioration of a generator bearing based on a metal self-healing technology. The method for evaluating the deterioration of a generator bearing based on a metal self-healing technology includes: collecting vibration signals, dynamic balance parameters, and temperature data of the generator bearing to obtain a time series of comprehensive bearing operating state parameters; extracting bearing balance characteristics and wear characteristics using a multi-dimensional feature extraction algorithm based on the time series of comprehensive bearing operating state parameters to obtain a bearing deterioration index system; injecting a metal self-healing material into the generator bearing according to the bearing deterioration index system to obtain a bearing after repair treatment; performing a dynamic balance test and monitoring the working state of the bearing after repair treatment to obtain a discrimination result in the repair stage and balance improvement data; predicting and analyzing the balance stability and deterioration trend of the bearing through a long short-term memory network based on the discrimination result in the repair stage and the balance improvement data to obtain a bearing health state evaluation index; generating a bearing balance and deterioration evaluation report and optimizing the metal self-healing strategy based on the bearing health state evaluation index to obtain a prediction result of the remaining service life of the bearing.

[0006] In a second aspect, the present application provides a system for evaluating the deterioration of a generator bearing based on a metal self-healing technology. The system for evaluating the deterioration of a generator bearing based on a metal self-healing technology includes:

[0007] A collection module for collecting vibration signals, dynamic balance parameters, and temperature data of the generator bearing to obtain a time series of comprehensive bearing operating state parameters;

[0008] An extraction module for extracting bearing balance characteristics and wear characteristics using a multi-dimensional feature extraction algorithm based on the time series of comprehensive bearing operating state parameters to obtain a bearing deterioration index system;

[0009] A repair module for injecting a metal self-healing material into the generator bearing according to the bearing deterioration index system to obtain a bearing after repair treatment;

[0010] A monitoring module for performing a dynamic balance test and monitoring the working state of the bearing after repair treatment to obtain a discrimination result in the repair stage and balance improvement data;

[0011] A prediction module for predicting and analyzing the balance stability and deterioration trend of the bearing through a long short-term memory network based on the discrimination result in the repair stage and the balance improvement data to obtain a bearing health state evaluation index;

[0012] A generation module for generating a bearing balance and deterioration evaluation report and optimizing the metal self-healing strategy based on the bearing health state evaluation index to obtain a prediction result of the remaining service life of the bearing.

[0013] In a third aspect, a computer device is provided, including: a memory and at least one processor, wherein instructions are stored in the memory; the at least one processor invokes the instructions in the memory to cause the computer device to execute the above-mentioned method for evaluating the deterioration of a generator bearing based on the metal self-healing technology.

[0014] In a fourth aspect, a computer-readable storage medium is provided, in which instructions are stored. When it runs on a computer, it causes the computer to execute the above-mentioned method for evaluating the deterioration of a generator bearing based on the metal self-healing technology.

[0015] In the technical solution provided by this application, multi-dimensional parameter data of the bearing is collected through vibration sensors, balance testers, and temperature sensors to form a time series of comprehensive parameters of the bearing operating state, providing a comprehensive data basis for subsequent analysis and avoiding the problem of incomplete diagnosis caused by single parameters in traditional monitoring methods; the multi-dimensional feature extraction algorithm is used to deeply mine the balance characteristics and wear characteristics of the bearing, construct an index system for bearing deterioration, and realize the precise quantification of the bearing health state, providing a scientific basis for the optimization of the metal self-healing material formula; the metal self-healing material formula is customized according to the bearing deterioration index system and injected into the bearing, solving the limitation that the traditional unified formula does not adapt to different damage characteristics and improving the repair efficiency; the dynamic balance test and working state monitoring of the bearing after the repair treatment are carried out to obtain the discrimination results in the repair stage and the data of the balance improvement, realizing the full-process monitoring of the repair process and enabling the timely discovery and adjustment of abnormal situations; the long short-term memory network is applied to predict and analyze the balance stability and deterioration trend of the bearing, making full use of the advantages of this deep learning algorithm in processing time series data, significantly improving the prediction accuracy. This algorithm can capture the complex non-linear relationships in the long time series of bearing state parameters and is particularly suitable for the application scenarios where the bearing working conditions are changeable and the data noise is large in the wind farm environment. The contribution of the algorithm is reflected in upgrading the traditional prediction method based on experience or simple statistics to the intelligent prediction level based on deep learning, making the prediction results closer to the actual deterioration law; finally, the bearing balance and deterioration evaluation report is generated according to the bearing health state evaluation index and the metal self-healing strategy is optimized, and at the same time, the remaining service life of the bearing is predicted, providing a scientific basis for the operation and maintenance decision-making of the wind farm, realizing the transformation from after-sales maintenance to predictive maintenance, greatly reducing the unplanned downtime and maintenance costs caused by bearing failures in wind turbines, and improving the operation efficiency and economic benefits of the wind farm; in addition, this method can achieve repair and evaluation without disassembling the bearing, reducing the downtime for maintenance and ensuring the continuous and stable power generation of the wind farm. Description of the Drawings

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can also be obtained based on these drawings.

[0017] Figure 1 It is a schematic diagram of an embodiment of the generator bearing deterioration evaluation method based on the metal self-repair technology in the embodiments of the present application;

[0018] Figure 2 It is a schematic diagram of an embodiment of the generator bearing deterioration evaluation system based on the metal self-repair technology in the embodiments of the present application;

[0019] Figure 3 It is a schematic block diagram of the structure of the computer device in the embodiments of the present invention. Specific embodiments

[0020] The embodiments of the present application provide a generator bearing deterioration evaluation method and system based on the metal self-repair technology. The terms "first", "second", "third", "fourth", etc. (if any) in the specification, claims and the above accompanying drawings of the present application are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments described herein can be implemented in an order other than those illustrated or described herein. In addition, the term "comprising" or "having" and any deformation thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units does not necessarily limit to those clearly listed steps or units, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0021] For easy understanding, the following describes the specific process of the embodiments of the present application. Please refer to Figure 1 An embodiment of the generator bearing deterioration evaluation method based on the metal self-repair technology in the embodiments of the present application includes:

[0022] Step S101, collect the vibration signal, dynamic balance parameter and temperature data of the generator bearing to obtain the time series of the comprehensive bearing operation state parameters;

[0023] Step S102, according to the time series of the comprehensive bearing operation state parameters, use the multi-dimensional feature extraction algorithm to extract the bearing balance feature and wear feature to obtain the bearing deterioration index system;

[0024] Step S103: Inject metal self - repair material into the generator bearing according to the bearing deterioration index system to obtain the bearing after repair treatment;

[0025] Step S104: Conduct dynamic balance testing and working condition monitoring on the bearing after repair treatment to obtain the discrimination result in the repair stage and the data of balance improvement;

[0026] Step S105: According to the discrimination result in the repair stage and the data of balance improvement, predict and analyze the balance stability and deterioration trend of the bearing through a long - short - term memory network to obtain the bearing health status evaluation index;

[0027] Step S106: Generate a bearing balance and deterioration evaluation report according to the bearing health status evaluation index and optimize the metal self - repair strategy to obtain the prediction result of the remaining service life of the bearing.

[0028] It can be understood that the execution subject of this application can be a generator bearing deterioration evaluation system based on metal self - repair technology, or a terminal or a server. Specifically, it is not limited here. In this embodiment of the application, the server is taken as the execution subject for illustration.

[0029] Specifically, install three - axial acceleration vibration sensors on the bearing seat of the wind turbine generator, set the sampling frequency to 10 kHz to capture the early fault characteristics of the bearing; at the same time, measure the unbalance amount and phase angle during the rotation of the bearing through a balance tester; in addition, arrange embedded temperature sensors on the surface of the bearing seat for continuous temperature monitoring. During the acquisition process, the vibration sensor records the operation of the bearing for one week under the condition of the rated speed of 1500 r / min, the balance tester measures the dynamic balance data three times a day, and the temperature sensor collects the temperature value once an hour. After the data acquisition is completed, synchronize and integrate the vibration signal, dynamic balance parameters, and temperature data according to the same time stamp, so as to obtain the time series of the comprehensive parameters of the bearing operation state, and this time series contains comprehensive information about the bearing operation state.

[0030] Extract multi - dimensional features from the time series of the comprehensive parameters of the bearing operation state. Perform multi - scale decomposition on the vibration signal through wavelet packet transform to extract the energy distribution characteristics of different frequency bands; then, extract the mean value, variance, peak factor, and waveform factor from the dynamic balance parameters to obtain quantitative evaluation parameters for balance; then, conduct correlation analysis on the temperature data and the speed data, and calculate the unit speed temperature rise rate as the thermodynamic wear index; then, perform envelope demodulation on the vibration energy distribution characteristics through Hilbert transform to extract the characteristic frequency energy of the inner ring, outer ring, rolling elements, and cage of the bearing; finally, construct the quantitative evaluation parameters for balance into the bearing balance index, and construct the thermodynamic wear index and the bearing fault characteristic spectrum into the bearing wear index, and the two together form the bearing deterioration index system.

[0031] Determine the formulation ratio of the metal self-repairing material based on the bearing deterioration index system. First, select magnesium hydroxy silicate as the main component, add boron and tin as auxiliary components, grind the material to a particle size of less than 100 nanometers through a ball milling process, and treat it with a surface modifier. Then, add the treated metal self-repairing material to the extreme pressure lithium-based grease at a ratio of 5%, and form a lubricating grease composite for repair through high-speed stirring. During the shutdown maintenance of the wind turbine, remove the oil filling port cover of the bearing housing, inject the lubricating grease composite for repair, and the injection volume is 70%-80% of the bearing cavity volume. Manually rotate the bearing several times to ensure uniform distribution of the grease, and then start the wind turbine to operate at a low speed for 4 hours to form a repair transition period. Dynamically monitor the repaired bearing. Periodically collect vibration signals and calculate the change rate of the root mean square value of vibration to form a vibration attenuation curve; measure the bearing unbalance every 4 hours and record the data to form a balance degree change sequence; continuously monitor the difference between the bearing temperature and the ambient temperature, calculate the unit load temperature rise ratio, and generate a temperature rise change curve. Based on these data, construct a three-dimensional feature space, project the data points to form a repair trajectory, identify the turning points of the trajectory through piecewise fitting, divide the repair process into four stages: fine grinding, adsorption cleaning, repair layer generation, and repair layer maintenance, and obtain the repair stage discrimination result. At the same time, by calculating the difference in unbalance before and after repair, the change in phase angle, and the balance stability time, obtain the data for improving balance.

[0032] Extract the duration ratio and conversion rate of each stage from the repair stage discrimination result to construct a repair dynamic feature vector; divide the balance improvement data by time window, calculate the balance degree change gradient and stability within each window to form a balance stability time series feature. Use the sliding window technique to resample these features in time to generate a time series sample set for training. Then, use a long short-term memory network to train the sample set, establish a bearing balance stability predictor, and input future working condition parameters to predict the change trend of the bearing balance state. According to the predicted trend and historical deterioration data, calculate the balance weight coefficient and wear weight coefficient through two-way weight assignment to form a comprehensive evaluation weight matrix, perform a weighted product operation on this matrix and the current state parameters of the bearing and normalize it to obtain the bearing health state evaluation index.

[0033] Finally, set the health status scoring criteria according to the bearing health status assessment index, map the index to the range of 0 - 100 score system, and establish a bearing health status classification table. Extract bearing failure cases with similar degradation patterns from the historical database, and construct a life mapping function. Through this function, convert the current bearing health status assessment index, correct the parameters in combination with the operating conditions, and calculate the theoretical value of the remaining service life. Analyze the action efficiency of the metal self-healing material according to the repair stage discrimination result, and adjust the proportions of magnesium hydroxy silicate, boron, and tin in the material formula to form an optimized self-healing strategy. Integrate the bearing health status score, balance improvement data, degradation trend prediction result, and optimized metal self-healing strategy into an assessment report. Finally, according to the theoretical value of the remaining service life and the change rate of the health status score, perform correction through the Weibull distribution fitting method to obtain the bearing remaining service life prediction result.

[0034] In the embodiment of the present application, multi-dimensional parameter data of the bearing is collected through vibration sensors, balance testers, and temperature sensors to form a comprehensive parameter time series of the bearing operating state, providing a comprehensive data basis for subsequent analysis and avoiding the problem of incomplete diagnosis caused by single parameters in traditional monitoring methods; the multi-dimensional feature extraction algorithm is used to deeply mine the balance characteristics and wear characteristics of the bearing, construct a bearing degradation index system, realize the accurate quantification of the bearing health status, and provide a scientific basis for the optimization of the metal self-healing material formula; customize the metal self-healing material formula according to the bearing degradation index system and inject it into the bearing, solving the limitation that the traditional unified formula does not adapt to different damage characteristics and improving the repair efficiency; perform dynamic balance testing and working state monitoring on the bearing after repair treatment to obtain the repair stage discrimination result and balance improvement data, realizing the full-process monitoring of the repair process and enabling timely discovery and adjustment of abnormal situations; apply the long short-term memory network to predict and analyze the balance stability and degradation trend of the bearing, making full use of the advantages of this deep learning algorithm in processing time series data, significantly improving the prediction accuracy. This algorithm can capture the complex non-linear relationships in the long time series of bearing state parameters, and is particularly suitable for the application scenarios with variable bearing working conditions and large data noise in the wind farm environment. The contribution of the algorithm is reflected in upgrading the traditional prediction method based on experience or simple statistics to the intelligent prediction level based on deep learning, making the prediction result closer to the actual degradation law; finally, generate a bearing balance and degradation assessment report according to the bearing health status assessment index and optimize the metal self-healing strategy, while predicting the remaining service life of the bearing, providing a scientific basis for the operation and maintenance decision-making of the wind farm, realizing the transformation from after-sales maintenance to predictive maintenance, greatly reducing the unplanned downtime and maintenance costs caused by bearing failures in wind turbines, and improving the operation efficiency and economic benefits of the wind farm; in addition, this method can achieve repair and assessment without disassembling the bearing, reducing the downtime for maintenance and ensuring the continuous and stable power generation of the wind farm.

[0035] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0036] Install a three-axial acceleration sensor on the generator of the wind turbine to collect bearing vibration signals, set the sampling frequency to 10 kHz, and obtain the original bearing vibration data;

[0037] Collect the generator speed data and power output data through the main control system of the wind turbine to obtain the generator operating condition parameters;

[0038] Use a balance tester to collect dynamic balance data of the generator bearing, measure the unbalance amount and unbalance phase angle when the bearing rotates, and obtain the bearing dynamic balance characteristic parameters;

[0039] Continuously monitor the bearing housing temperature through an embedded temperature sensor, and record the ambient temperature at the same time to obtain the bearing temperature rise data;

[0040] Synchronously integrate the original bearing vibration data, generator operating condition parameters, bearing dynamic balance characteristic parameters, and bearing temperature rise data according to the time stamp to form a time series database containing multi-dimensional parameters, and obtain the time series of the comprehensive bearing operating state parameters.

[0041] Specifically, installing a three-axial acceleration sensor on the generator of the wind turbine is to comprehensively capture the vibration characteristics of the bearing in three directions (radial, axial, and tangential). The three-axial acceleration sensor usually uses a piezoelectric sensor to convert mechanical vibration into an electrical signal. The sensor installation position is selected on the generator bearing housing, as close as possible to the bearing body without affecting the normal operation of the bearing. The sampling frequency is set to 10 kHz based on the Nyquist sampling theorem to ensure that the bearing fault characteristic frequencies can be captured, and these frequencies are usually in the range of several kilohertz. The collected data is stored as a time-domain vibration waveform to form the original bearing vibration data, which is an information matrix containing two dimensions of amplitude and time. The main control system of the wind turbine is integrated with a generator speed sensor and a power monitoring device. The speed sensor is usually based on the Hall effect principle or the optoelectronic principle and is installed at the end of the generator shaft. It generates one or more pulse signals per revolution, and the accurate speed value is obtained by calculating the number of pulses per unit time. The power output data is collected through current and voltage sensors and the real-time power value is obtained through multiplication. These two types of data are sampled by the data acquisition card of the main control system at a frequency not lower than 1 Hz and recorded in the form of a time series, constituting the generator operating condition parameters. These parameters are crucial for subsequent analysis because the vibration characteristics and temperature performance of the bearing are directly related to the speed and load.

[0042] The balance tester is a device specifically used to measure the dynamic balance state of rotating components. To collect dynamic balance data of the generator bearings, a balance tester with phase measurement function is required. During the test, a phase marking point is fixed on the generator shaft. The tester captures the reference position for each revolution through an optoelectronic or magnetic sensor, and at the same time measures the vibration force at the support point through a force sensor. The unbalance amount represents the degree of uneven mass distribution of the rotating body, with the unit of gram-millimeter (g·mm), and the unbalance phase angle represents the angular position of the maximum unbalance mass point relative to the reference point, with the unit of degree (°). The measured unbalance amount and unbalance phase angle constitute the dynamic balance characteristic parameters of the bearings, which directly reflect the balance state of the bearings and are important indicators for evaluating the bearing quality and health status.

[0043] Embedded temperature sensors use thermocouples or PT100 temperature sensors and are installed on the outer surface of the bearing housing, as close as possible to the bearing working area. At the same time, an ambient temperature sensor is set in the engine room to record the surrounding ambient temperature. The temperature data acquisition frequency is usually set to once per minute to form a continuous temperature time series. By calculating the difference between the bearing housing temperature and the ambient temperature, the bearing temperature rise data is obtained, eliminating the influence of ambient temperature fluctuations on the measurement results and more accurately reflecting the heat generation due to internal friction of the bearing. The bearing temperature rise is an important indicator of the bearing health state. An abnormal temperature rise often indicates problems such as abnormal friction, poor lubrication, or overload in the bearing. After the above four types of data (original bearing vibration data, generator operating parameters, bearing dynamic balance characteristic parameters, and bearing temperature rise data) are collected, they need to be synchronously integrated. First, a time stamp in a unified format is added to each data point, accurate to the millisecond level. Then, through data interpolation processing, data with different sampling frequencies are resampled to a unified time point to align all data in the time dimension. The integrated dataset contains multi-dimensional parameters, forming a time series database, that is, the time series of comprehensive parameters of the bearing operating state. This time series is the data basis for subsequent bearing health state assessment. By correlating parameters in different dimensions, it can comprehensively reflect the operating state and deterioration degree of the bearing.

[0044] Taking a 1.5MW wind turbine in a certain wind farm as an example, the data acquisition process of its main shaft bearing (model FD-239 / 750CA / W33) is described. Triaxial acceleration sensors are installed on the top, side and end face of the bearing housing. The sampling frequency is set to 10kHz, and 10 seconds of vibration data is collected every hour. The main control system records the generator speed and power output at a frequency of 1Hz. The balance tester measures the bearing during regular maintenance and records the unbalance amount and phase angle. The temperature sensor collects the bearing housing temperature and ambient temperature once every minute. These data are synchronized and integrated according to the time stamp to form a comprehensive time series containing multi-dimensional parameters such as vibration amplitude, spectral characteristics, speed, power, unbalance amount, phase angle, temperature, etc., recording the operating state of the bearing under different working conditions.

[0045] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0046] Perform multi-scale decomposition on the vibration signal in the comprehensive parameter time series of the bearing operating state through wavelet packet transform to obtain the vibration energy distribution characteristics;

[0047] Extract the mean value, variance, peak factor and waveform factor of the dynamic balance parameters from the comprehensive parameter time series of the bearing operating state to obtain the quantitative evaluation parameters of balance;

[0048] Perform correlation analysis on the temperature data and generator speed data in the comprehensive parameter time series of the bearing operating state, calculate the temperature rise rate per unit speed, and obtain the thermodynamic wear index;

[0049] Perform envelope demodulation processing on the vibration energy distribution characteristics through Hilbert transform, extract the characteristic frequency energy of the inner ring, outer ring, rolling elements and cage of the bearing, and obtain the bearing fault characteristic spectrum;

[0050] Construct a bearing balance index based on the quantitative evaluation parameters of balance, and construct a bearing wear index based on the thermodynamic wear index and the bearing fault characteristic spectrum to obtain the bearing double deterioration evaluation criterion;

[0051] Normalize the bearing balance index and the bearing wear index and set the weight coefficient, and generate a comprehensive evaluation standard through weighted fusion to obtain the bearing deterioration index system.

[0052] Specifically, the vibration signal in the time series of comprehensive parameters of bearing operating state is processed. Wavelet packet transform is a time-frequency domain analysis method. Compared with the traditional Fourier transform, it can have good resolution in both the time domain and the frequency domain. The process of wavelet packet transform for the vibration signal includes: selecting a suitable wavelet basis function (such as Daubechies wavelet), determining the decomposition level (usually 3 - 5 levels), and then decomposing the vibration signal through high-pass and low-pass filters in sequence to form a tree structure. Each node represents the signal of a specific frequency band, calculating the energy values of each node, and constituting the vibration energy distribution characteristics. This method can effectively extract the vibration characteristics of the bearing in different frequency bands and provide more detailed information for fault diagnosis. Extracting the statistical characteristics of dynamic balance parameters from the time series of comprehensive parameters of bearing operating state is an important means to evaluate the balance of the bearing. The mean value reflects the average level of the unbalance amount, and the calculation method is to calculate the arithmetic mean of the unbalance amount data within a certain time window. The variance describes the fluctuation degree of the unbalance amount, and the calculation method is the sum of the squares of the differences between each measured value and the mean value divided by the number of samples. The peak factor is the ratio of the maximum unbalance amount to the root mean square value, reflecting the peak characteristics of the unbalance degree. The waveform factor is the ratio of the root mean square value to the average value, characterizing the complexity of the unbalance amount waveform. These four parameters together constitute the quantitative evaluation parameters of balance, comprehensively describing the dynamic balance state of the bearing.

[0053] The calculation of the thermodynamic wear index depends on the correlation analysis of temperature data and rotational speed data. The unit rotational speed temperature rise rate is calculated by the formula:

[0054]

[0055] where, represents the temperature rise of the bearing (the difference between the bearing temperature and the ambient temperature), is the bearing temperature, is the ambient temperature, is the rotational speed of the generator. When calculating, first pair the temperature data and the rotational speed data, then calculate the ratio of the temperature rise to the rotational speed at each time point, and finally take the average value within a certain time window as the unit rotational speed temperature rise rate. This index can reflect the intensity of heat generation due to bearing friction and is an important parameter for evaluating the wear state of the bearing.

[0056] The Hilbert transform is an effective tool for extracting bearing fault characteristics. It converts a real signal into an analytic signal to obtain the instantaneous amplitude of the signal. Perform the Hilbert transform on the vibration energy distribution characteristics to obtain its envelope signal, and then perform the Fourier transform on the envelope signal to obtain the envelope spectrum. The characteristic frequencies and their amplitudes of the inner ring, outer ring, rolling elements, and cage of the bearing can be identified from the envelope spectrum. These characteristic frequencies are related to the bearing geometric parameters and rotational speed, and the calculation formulas are as follows:

[0057]

[0058]

[0059]

[0060]

[0061] Among them, and and and are the characteristic frequencies of the inner ring, outer ring, rolling elements and cage respectively, is the number of rolling elements, is the relative rotational speed, is the diameter of the rolling element, is the pitch diameter, is the contact angle. Extract the energy values corresponding to these characteristic frequencies to form the bearing fault characteristic spectrum.

[0062] Bearing balance index is constructed based on the balance quantitative evaluation parameters, and the calculation formula is:

[0063]

[0064] Among them, is the mean value of the unbalance, is the variance of the unbalance, is the peak factor, is the waveform factor, and and and are the reference values of the corresponding parameters (usually taken from the data of healthy bearings), and and and are the weight coefficients of each parameter.

[0065] Bearing wear index is constructed depending on the thermodynamic wear index and the bearing fault characteristic spectrum, and the calculation formula is:

[0066]

[0067] Among them, is the temperature rise rate per unit rotational speed, and and and are the energy values of the characteristic frequencies of the inner ring, outer ring, rolling elements and cage respectively, and and and , is the reference value of the corresponding parameter, , , , , are the weight coefficients of each parameter. The bearing balance index and the bearing wear index together constitute the bearing double deterioration evaluation criterion.

[0068] Finally, the bearing balance index and the bearing wear index are normalized and weighted and fused to obtain the bearing deterioration index system , and the calculation formula is:

[0069]

[0070] where, and are respectively the minimum value and the maximum value of the bearing balance index, and are respectively the minimum value and the maximum value of the bearing wear index, and are the weight coefficients of the bearing balance index and the bearing wear index, satisfying + = 1.

[0071] Taking the main shaft bearing (FD-239 / 750CA / W33) of a 1.5MW wind turbine in a certain wind farm as an example, the data processing process is as follows: First, the collected 10kHz vibration signal is decomposed by wavelet packet, the db4 wavelet basis function is selected, and decomposed to 4 layers to obtain the energy distribution of 16 frequency bands. Then, the average unbalance amount is calculated as 32 g·mm, the variance is 15.6, the peak factor is 3.4, and the waveform factor is 1.8 from the dynamic balance test data. Next, the difference between the bearing temperature (average 65°C) and the ambient temperature (average 20°C), and the average generator speed (1200 r / min) are correlated and analyzed, and the unit speed temperature rise rate is calculated as 0.0375 °C / (r / min). Subsequently, the vibration signal is envelope demodulated by Hilbert transform, and the energy values of the inner ring characteristic frequency (98.6 Hz), the outer ring characteristic frequency (65.2 Hz), the rolling element characteristic frequency (32.8 Hz), and the cage characteristic frequency (5.2 Hz) are extracted. According to these data, the bearing balance index is calculated as 0.68, the bearing wear index is 0.72, and after normalization and weighted fusion (weights are 0.4 and 0.6 respectively), the bearing deterioration index system value is 0.704. This index intuitively reflects the health status of the bearing and provides a scientific basis for the injection strategy of the metal self-repair material.

[0072] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0073] Determine the formulation ratio of the metal self - repairing material according to the bearing balance index and bearing wear index in the bearing deterioration index system, and prepare nano - scale self - repairing particles containing components of magnesium hydroxy silicate, boron and tin;

[0074] Grind the self - repairing particles to a particle size less than one hundred nanometers through a ball - milling process and treat them with a surface modifier to prepare a highly dispersible metal self - repairing material;

[0075] Add the metal self - repairing material to the extreme - pressure lithium - based grease at a ratio of 5%, and mix evenly through a high - speed stirring device to form a grease composite for repair;

[0076] During the shutdown maintenance of the wind turbine, remove the oil filling port cover on the generator bearing housing, and inject the grease composite for repair into the bearing cavity, with the injection volume controlled at 70% - 80% of the bearing cavity;

[0077] Manually rotate the bearing several circles to promote the even distribution of the grease composite for repair on the surfaces of the bearing rolling elements and raceways, forming an initial repair layer;

[0078] Restore the bearing housing to its original state and start the wind turbine, and run the generator at a low speed for four hours to form a repair transition period, obtaining the repaired bearing.

[0079] Specifically, according to the bearing balance index and bearing wear index in the bearing deterioration index system, the formulation ratio of the metal self-repair material needs to be determined. This process first analyzes the bearing wear state: when the bearing wear index is high and the balance index is low, it indicates that the main problem of the bearing lies in surface wear, and at this time, the proportion of magnesium hydroxy silicate in the self-repair material should be increased; when the balance index is high and the wear index is low, it indicates that the main problem of the bearing lies in imbalance, and at this time, the proportions of boron and tin should be increased to improve the fluidity and adhesion of the material. Magnesium hydroxy silicate Mg3(Si2O5)(OH)4 is a silicate mineral with a layered crystal structure, which can undergo micro-sintering and micro-metallurgical processes under the action of frictional heat and pressure to form a metal-ceramic repair layer. Boron can improve the hardness and wear resistance of the material, while tin helps to improve the adhesion and film-forming properties of the material. The typical ratio of the three components is 75% magnesium hydroxy silicate, 15% boron, and 10% tin, but it will be adjusted according to the specific bearing deterioration conditions. The preparation of nano-scale self-repair particles uses the ball milling process, which is a mechanical crushing method that can effectively reduce the particle size. The specific operation is to mix the raw materials in a determined ratio and put them into a ball mill, add an appropriate amount of grinding media (usually steel balls or zirconia balls of different diameters), set the ball-to-material ratio (the mass ratio of balls to materials) to 5:1, the rotation speed to 300 - 400 revolutions per minute, and the grinding time to 4 - 6 hours. During the grinding process, the material is gradually crushed into nano-scale particles under the impact, friction, and extrusion of the grinding media. After grinding, the particle size distribution is detected by a laser particle size analyzer to ensure that the particle size of all particles is less than 100 nanometers. Then, it is treated with a surface modifier. The surface modifier usually selects alkyl silane compounds, which can form an organic coating layer on the particle surface to prevent particle agglomeration and improve its dispersibility and stability in grease. The treatment method is to mix the nano-particles and the surface modifier at a mass ratio of 1:0.05, stir and react in a solvent for 2 hours, and then filter, wash, and dry to obtain a highly dispersible metal self-repair material. Adding the metal self-repair material to the grease is the key step to achieve bearing self-repair. Selecting extreme pressure lithium-based grease as the carrier is because it has good compressive performance, water stability, and mechanical stability, which are suitable for the working conditions of the main shaft and generator bearings of wind turbines. The addition ratio is set at 5%, and this value is the optimal concentration verified through a large number of experiments, which can ensure sufficient repair ability while not affecting the basic performance of the grease. The mixing process uses a high-speed stirring device, with the rotation speed set at 1500 - 2000 revolutions per minute and the stirring time at 30 minutes to ensure that the self-repair material is evenly dispersed in the grease and form a stable lubricating grease composite for repair. After mixing, the consistency level of the composite is confirmed through a penetration test to ensure that it still meets the bearing lubrication requirements.

[0080] Replacing the bearing grease during the shutdown maintenance of the wind turbine is a necessary measure to ensure safety. First, clean the area around the bearing housing to prevent impurities from entering. Then, remove the grease injection port cover on the generator bearing housing, which is usually located at the top or side of the bearing housing. Check the condition of the original grease. If there is obvious discoloration, hardening, or contamination, the old grease needs to be removed first. Use a special grease gun to inject the repair grease compound into the bearing cavity, and control the injection volume within 70% to 80% of the bearing cavity. Controlling the injection volume is very important: less than 70% will result in insufficient lubrication, causing bearing overheating and early wear; more than 80% will increase the operating resistance of the bearing and generate additional heat. During the injection process, a uniform and slow speed should be maintained to avoid generating bubbles or uneven distribution.

[0081] After the injection is completed, it is necessary to manually rotate the bearing several circles, usually 10 - 15 circles. This process has two purposes: one is to promote the uniform distribution of the repair grease compound on the bearing rolling elements and raceway surfaces; the other is to make the self-repairing material initially contact the worn parts to form an initial repair layer. When manually rotating, it should be kept steady, and the rotation time for each circle is about 2 - 3 seconds, neither too fast (affecting the grease distribution) nor too slow (affecting the formation of the initial repair layer). After the rotation is completed, observe the operation of the bearing to confirm that there is no abnormal resistance or noise.

[0082] After restoring the bearing housing to its original state, it is necessary to run the bearing at a low speed to form a repair transition period. Running at a low speed means controlling the generator speed within the range of 30% - 50% of the rated speed. The purpose of this is to reduce the working load of the bearing and allow the self-repairing material to have enough time to act on the friction surface. The running time is set to 4 hours. During this period, under the action of friction heat and pressure, the self-repairing material will gradually form a stable repair layer. After the transition period ends, the bearing can resume normal operation, and the repair layer will be continuously improved during subsequent operation, ultimately achieving an ideal repair effect.

[0083] Taking the main shaft bearing of a 1.5 MW wind turbine in a certain wind farm as an example, the bearing model is FD-239 / 750CA / W33. Through bearing deterioration assessment, the bearing wear index is 0.72 and the balance index is 0.65. According to these indicators, the self-repair material formula is determined as 78% magnesium hydroxy silicate, 12% boron, and 10% tin. The material is ground for 8 hours through a ball milling process, and particle size detection shows that 85% of the particles are less than 80 nanometers. After treatment with an alkyl silane surface modifier, the treated self-repair material is added to No. 2 extreme pressure lithium-based grease at a ratio of 5%, and high-speed stirring is carried out for 40 minutes to form a repair grease composite. During the planned shutdown maintenance of the wind turbine, the oil filling port cover of the bearing seat is disassembled, and about 2.5 kg of the repair grease composite is injected, accounting for 75% of the bearing cavity. After manually rotating the bearing 12 times, the bearing seat is restored, and the fan is controlled to operate at a low speed of 500 revolutions per minute for 4 hours. Detection after one week of repair shows that the bearing temperature has decreased by 8°C, the vibration amplitude has decreased by 25%, and the balance state has been significantly improved, verifying the effectiveness of this method in the maintenance of wind turbine bearings.

[0084] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0085] Periodically collect vibration signals after injecting the metal self-repair material into the bearing, and calculate the change rate of the root mean square value of vibration through spectrum analysis to form a vibration attenuation curve;

[0086] Use a balance tester to measure the unbalance of the bearing every four hours, record the data to form a balance change sequence, and obtain the dynamic balance improvement trend;

[0087] Continuously monitor the temperature difference between the bearing temperature and the ambient temperature through a temperature sensor, calculate the unit load temperature rise ratio, and generate a temperature rise change curve;

[0088] According to the vibration attenuation curve, the dynamic balance improvement trend, and the temperature rise change curve, construct a three-dimensional feature space, and project the data points into the three-dimensional feature space to form a repair trajectory;

[0089] Perform piecewise fitting on the repair trajectory, identify the turning points in the trajectory, and divide the repair process into a fine grinding stage, an adsorption cleaning stage, a repair layer generation stage, and a repair layer maintenance stage according to the turning points to obtain the repair stage discrimination result;

[0090] Calculate the difference in unbalance, the change in phase angle, and the balance stability time before and after repair, and combine the stable section data in the dynamic balance improvement trend to obtain the balance improvement data.

[0091] Specifically, after injecting the metal self - repair material into the bearing, it is necessary to periodically collect vibration signals. The monitoring period is usually set to once every 4 hours, the sampling duration is 10 seconds, and the sampling frequency is 10 kHz. The collected vibration time - domain signal is converted into a spectrum through the fast Fourier transform (FFT), and then the root mean square (RMS) value of the vibration is calculated. This value reflects the magnitude of the vibration energy. The calculation method of the change rate of the vibration RMS value is to subtract the RMS value at the initial moment from the RMS value at the current moment, and then divide by the RMS value at the initial moment. The obtained data sequence forms a vibration attenuation curve. This curve intuitively reflects the change trend of the bearing vibration energy over time and is an important indicator for evaluating the repair effect of the metal self - repair material. The balance test is an important means to evaluate the dynamic balance state of the bearing. Use a balance tester to measure the unbalance amount of the bearing every four hours. The measurement needs to be carried out under the same rotational speed condition to ensure the comparability of the data. The tester measures the vibration force at the support point through a force sensor and calculates the unbalance amount and phase angle in combination with the phase information provided by the phase sensor. These data are arranged in chronological order to form a balance change sequence. Trend analysis is performed on this sequence to calculate the change rate and change direction between consecutive time points, obtaining the dynamic balance improvement trend. This trend can reflect the change of the bearing balance state over time and is an important basis for judging the influence of the metal self - repair material on the bearing balance performance.

[0092] Temperature monitoring is a direct means to evaluate the friction state of the bearing. Continuously monitor the bearing temperature and ambient temperature at 1 - minute intervals through a temperature sensor, and calculate the difference between the two to obtain the bearing temperature rise. At the same time, obtain the current load data (usually expressed in torque or power) from the main control system of the wind turbine and calculate the unit load temperature rise ratio, that is, the bearing temperature rise divided by the current load. This index eliminates the influence of load changes on the bearing temperature and more accurately reflects the friction state of the bearing. Arrange the calculated unit load temperature rise ratio in chronological order to generate a temperature rise change curve. This curve is an intuitive index for evaluating the influence of the metal self - repair material on the bearing frictional heat. The above three curves (vibration attenuation curve, dynamic balance improvement trend, and temperature rise change curve) respectively reflect different aspects of the bearing repair process. To comprehensively analyze the repair effect, it is necessary to construct a three - dimensional feature space with the change rate of the vibration RMS value as the X - axis, the balance improvement rate as the Y - axis, and the change rate of the unit load temperature rise ratio as the Z - axis. Project the data points at different times into this three - dimensional space in chronological order to form a continuous repair trajectory. This trajectory intuitively shows the coordinated changes of various performance indicators of the bearing during the repair process and provides a basis for subsequent stage division.

[0093] Segmented fitting of the repair trajectory is a key step in identifying different stages of repair. First, methods such as piecewise linear fitting or polynomial fitting are used to mathematically describe the trajectory, and then by calculating the derivative or curvature change of the fitted curve, the turning points in the trajectory are identified. These turning points correspond to the state changes during the repair process and are the basis for dividing the repair stages. According to the working mechanism of the metal self-repair technology, the repair process can be divided into four stages: the fine grinding stage (characterized by a slight increase in the root mean square value of vibration, a slight deterioration in balance, and an increase in temperature rise), the adsorption and cleaning stage (vibration starts to decline, balance tends to stabilize, and temperature rise continues to increase), the repair layer generation stage (vibration rapidly declines, balance is significantly improved, and temperature rise starts to decline), and the repair layer maintenance stage (all indicators tend to stabilize). By comparing the characteristics of each segment of the trajectory with the characteristics of the above stages, the time boundaries of the repair stages are determined, and the discrimination results of the repair stages are obtained.

[0094] The calculation of the balance improvement data involves multiple indicators and is mainly achieved through the following formula:

[0095]

[0096]

[0097]

[0098]

[0099] Among them, represents the difference in unbalance, and are the unbalances before and after repair respectively; represents the change in phase angle, and are the phase angles before and after repair respectively; represents the balance stability time, is the time point when the stable state is reached, is the time point when the repair starts; is the balance improvement index, 、 、 and are the weight coefficients, corresponding to the weights of unbalance improvement, phase angle change, stable time, and stability respectively; is the reference stable time; and are the balance stability indicators after and before repair respectively, usually represented by the standard deviation of the balance.

[0100] Taking the main shaft bearing of a 1.5 MW wind turbine as an example, monitoring was carried out for 7 days after injecting the metal self-repairing material. On the first day, the root mean square value of vibration slightly increased from 4.5 mm / s to 4.8 mm / s, the unbalance increased from 38 g·mm to 42 g·mm, and the unit load temperature rise ratio increased from 0.12 °C / kNm to 0.14 °C / kNm. These data changes indicate that the bearing is in the fine grinding stage. On the second day, the root mean square value of vibration began to decrease to 4.3 mm / s, the unbalance stabilized at about 40 g·mm, and the unit load temperature rise ratio continued to rise to 0.15 °C / kNm, indicating that it entered the adsorption cleaning stage. From the third to the fifth day, the root mean square value of vibration rapidly decreased to 3.2 mm / s, the unbalance decreased to 26 g·mm, the phase angle changed from the original 135° to 95°, and the unit load temperature rise ratio began to decrease to 0.11 °C / kNm, indicating that the bearing entered the repair layer generation stage. On the sixth and seventh days, all indicators tended to be stable. The root mean square value of vibration stabilized at about 3.0 mm / s, the unbalance fluctuated at about 25 g·mm, and the unit load temperature rise ratio stabilized at 0.10 °C / kNm, indicating that the bearing entered the repair layer maintenance stage. According to the formula calculation, the difference in unbalance was 13 g·mm, the change in phase angle was 40°, the balance stability time was 96 hours, the balance stability increased by 65%, and the balance improvement index calculated comprehensively was 0.72. This value intuitively reflects the significant improvement effect of the metal self-repairing material on the balance performance of the bearing.

[0101] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0102] Extract the proportion of the duration of each stage and the conversion rate from the repair stage discrimination result to construct a repair dynamic feature vector;

[0103] Divide the balance improvement data by time window, calculate the balance change gradient and stability degree within each time window to form a balance stability time series feature;

[0104] Perform time resampling on the repair dynamic feature vector and the balance stability time series feature through the sliding window technique to generate a training time series sample set;

[0105] Use the long short-term memory network to train the training time series sample set, establish a bearing balance stability predictor, and input future working condition parameters to predict the change trend of the bearing balance state;

[0106] According to the change trend of the bearing balance state and the historical deterioration data, calculate the balance weight coefficient and the wear weight coefficient through the two-way weight distribution method to form a comprehensive evaluation weight matrix;

[0107] Perform a weighted product operation on the comprehensive evaluation weight matrix and the current state parameters of the bearing, and obtain the bearing health state evaluation index after normalization processing.

[0108] Specifically, extracting the duration ratio and conversion rate of each stage from the repair stage discrimination result is the basic step for constructing the repair dynamic feature vector. The repair stage discrimination result includes the start and end time points of four stages: the fine grinding stage, the adsorption and cleaning stage, the repair layer generation stage, and the repair layer maintenance stage. The duration ratio of each stage refers to the proportion of the duration of each stage in the total repair time, and the calculation method is to divide the duration of each stage by the total repair time. The conversion rate describes the speed of transitioning from one stage to the next, and the calculation method is to divide the change amount of relevant parameters (such as the root mean square value of vibration, unbalance amount) before and after the stage conversion by the time required for the conversion. The combination of these two types of indicators forms the repair dynamic feature vector, which intuitively reflects the dynamic process of the metal self-repair material playing a role in the bearing and is an important basis for predicting future repair effects. The balance improvement data includes information such as the unbalance amount difference, phase angle change, and balance stability time, and needs to be segmented and processed according to the time window. The time window refers to dividing the entire monitoring period into several equal time periods, and each window contains a certain number of consecutive data points. For each time window, calculate the balance change gradient, that is, the change rate of the balance parameter within the window, and the calculation method is to subtract the balance parameter value at the start of the window from the balance parameter value at the end of the window, and then divide by the time span of the window. At the same time, calculate the standard deviation of the balance data within the window as a measure of the balance stability degree. These two types of indicators are arranged in chronological order to form the balance stability time series feature, which intuitively shows the change law of the bearing balance state over time and provides the time series data basis for subsequent prediction and analysis.

[0109] The sliding window technique is a commonly used time series processing method for resampling and feature extraction of time series data. In this method, the application of the sliding window is to convert the repaired dynamic feature vectors and balanced stability time series features into a data format suitable for the input of machine learning algorithms. Specifically, when implementing, the window size (e.g., 24 hours) and the sliding step (e.g., 4 hours) are set. The window starts from the starting position of the time series and slides backward by one step each time, extracting all the feature values within the window to form a sample. As the window slides continuously, multiple samples are generated, constituting the time series sample set for training. This sample set contains the dynamic change information of the bearing state during the repair process and is the data basis for training the prediction model. Long Short-Term Memory (LSTM) is a special type of recurrent neural network that is good at dealing with long-term dependencies in time series data. In the assessment of bearing health status, the input of the LSTM network is the time series sample set for training generated previously, and the output is the predicted value of the bearing balance state at future moments. The network structure includes an input layer, an LSTM layer, and an output layer. The number of neurons in the LSTM layer is usually set to 32 - 128 and adjusted according to the data complexity. The training process uses the backpropagation algorithm, the loss function selects the mean squared error (MSE), and the optimization algorithm adopts Adam. After training, future working condition parameters (such as the expected load level, operating speed, etc.) are input into the trained network to obtain the prediction results of the bearing balance state change trend, including the predicted values of key indicators such as the unbalance amount and phase angle within a future period (such as 7 days).

[0110] The two-way weight assignment method is an algorithm that dynamically adjusts the weights of evaluation indicators according to historical data and prediction trends. The health status of a bearing is affected by two main aspects: balance and wear, and the importance of the two will change with the working conditions and usage stages of the bearing. The two-way weight assignment method realizes the dynamic trade-off between the two aspects by calculating the balance weight coefficient and the wear weight coefficient. When calculating, first analyze the change amplitude and stability in the change trend of the bearing balance state. The indicator with a large change amplitude and poor stability should be assigned a higher weight. Then, combine the correlation between each indicator and bearing faults in the historical deterioration data. The indicator with a high correlation is assigned a higher weight. The calculation results of the two parts are combined to form the balance weight coefficient and the wear weight coefficient, constituting the comprehensive evaluation weight matrix. This matrix reflects the contribution size of different indicators to the bearing health status under the current working conditions and is the key to comprehensively evaluating the bearing health status. Perform a weighted product operation on the comprehensive evaluation weight matrix and the current state parameters of the bearing to obtain the original evaluation value, and then map this value to the interval of 0 - 100 through normalization processing to obtain the bearing health status evaluation index. The method of normalization processing is to subtract the historical minimum value from the original evaluation value, then divide by the difference between the historical maximum value and the minimum value, and then multiply by 100. This index intuitively reflects the health status of the bearing. The higher the value, the better the bearing state, which is an important basis for bearing maintenance decisions.

[0111] Taking the main shaft bearing of a 1.5 MW wind turbine in a certain wind farm as an example, through the analysis of the discrimination results in the repair stage, it is found that the fine grinding stage accounts for 15% of the total repair time, the adsorption and cleaning stage accounts for 20%, the repair layer generation stage accounts for 45%, and the repair layer maintenance stage accounts for 20%. The conversion rates of each stage are 0.06 / hour, 0.12 / hour, and 0.03 / hour respectively. These data are composed into a repair dynamic feature vector (0.15, 0.20, 0.45, 0.20, 0.06, 0.12, 0.03). The data on balance improvement are segmented according to a 24-hour time window, and the calculated balance change gradients of each window are -1.2 g·mm / day, -2.5 g·mm / day, -3.8 g·mm / day, -2.0 g·mm / day, -0.5 g·mm / day respectively, and the corresponding balance stability degrees (standard deviations) are 3.2, 2.8, 1.9, 1.5, 1.2. Using the sliding window technique with a window size of 72 hours and a step size of 24 hours, a training sample set is generated. Through the training of the LSTM network, a bearing balance stability predictor is obtained to predict the change trend of the balance state in the next 7 days. Combining with historical deterioration data, the balance weight coefficient is calculated to be 0.62, and the wear weight coefficient is 0.38, forming a comprehensive evaluation weight matrix. After weighted calculation and normalization, the bearing health state evaluation index is 82, indicating that the bearing is in good condition and is expected to operate normally for a long time, which verifies the effectiveness of the metal self-repair technology for the maintenance of wind turbine bearings.

[0112] In a specific embodiment, the process of executing step S106 may specifically include the following steps:

[0113] Set the health state scoring standard according to the bearing health state evaluation index, map the index value to the scoring range of zero to one hundred, and establish a bearing health state grading table;

[0114] Extract the bearing failure cases with similar deterioration patterns from the historical database, extract the corresponding relationship between the failure time point and the bearing health state evaluation index, and construct a life mapping function;

[0115] Convert the current bearing health state evaluation index through the life mapping function, and combine the operation condition correction parameters to calculate the theoretical value of the remaining service life of the bearing;

[0116] According to the analysis of the discrimination results in the repair stage, analyze the action efficiency of the metal self-repair material, and combine the data on balance improvement to adjust the proportions of magnesium hydroxy silicate, boron, and tin in the metal self-repair material formula to form an optimized metal self-repair strategy;

[0117] Integrate the bearing health status score, balance improvement data, deterioration trend prediction results, and the optimized metal self-repair strategy to generate a bearing balance and deterioration assessment report;

[0118] According to the theoretical value of the remaining service life and the change rate of the bearing health status score, perform correction calculations through the Weibull distribution fitting method to obtain the bearing remaining service life prediction result.

[0119] Specifically, set the health status scoring standard according to the bearing health status evaluation index, and map the index value to the zero to one hundred-point scoring interval. This mapping process is based on the bearing performance requirements and historical operation data in the wind power industry. Usually, 0 - 40 points are defined as the dangerous area (indicating that the bearing has severely deteriorated and needs to be replaced immediately), 41 - 60 points as the warning area (indicating that the bearing has started to deteriorate significantly and needs to be closely monitored), 61 - 80 points as the attention area (indicating that the bearing performance has declined but is still within the acceptable range), and 81 - 100 points as the healthy area (indicating that the bearing is in good condition). The scoring standard takes into account multiple factors such as the vibration characteristics, balance performance, temperature rise situation, and noise level of the bearing. The weights of each factor are adjusted according to the operating characteristics of the wind turbine. Organize these area divisions and corresponding treatment suggestions into a bearing health status grading table to provide an intuitive reference for operation and maintenance decisions. Extracting bearing failure cases with similar deterioration patterns from the historical database is a key step in establishing the life prediction model. The judgment criteria for similar deterioration patterns include the same or similar bearing models, similar working conditions, and similar deterioration characteristics (such as vibration spectrum characteristics, temperature rise patterns, etc.). For each failure case, extract the corresponding relationship between the failure time point and the health status evaluation index, and record the entire process change curve of the health status evaluation index from health to failure. By analyzing the data of multiple failure cases, identify the statistical relationship between the health status evaluation index and the remaining service life, and use regression analysis methods (such as polynomial regression, exponential regression, etc.) to construct a life mapping function. This function takes the health status evaluation index as the independent variable and the remaining service life as the dependent variable to establish the mathematical relationship between the two.

[0120] Convert the current bearing health status evaluation index through the life mapping function to obtain a preliminary estimate of the remaining service life. Then, adjust it by combining the operating condition correction parameters. The operating condition correction parameters consider the influence of factors such as the current and expected load levels, speed range, ambient temperature, start-stop frequency, etc. on the bearing life. The correction method is to multiply the preliminary estimate by the correction coefficients of each operating condition parameter to obtain the theoretical value of the bearing remaining service life considering the actual operating conditions. This theoretical value more accurately reflects the expected life of the bearing under actual operating conditions and provides a scientific basis for operation and maintenance planning.

[0121] Analyzing the action efficiency of metal self-repair materials based on the discrimination results of the repair stages is the basis for optimizing the repair strategy. The evaluation of the action efficiency includes the duration of each repair stage, the rate and amplitude of the improvement in bearing performance within each stage. If the fine grinding stage and the adsorption and cleaning stage are relatively long, while the repair layer generation stage is relatively short, it indicates that the current self-repair material has strong grinding and cleaning capabilities but weak film-forming capabilities, and the proportion of film-forming components needs to be increased. Combining the balance improvement data, specifically analyzing indicators such as the improvement effect of the unbalance amount, the stability of the phase angle, and the balance stability time, to judge the impact of the self-repair material on the dynamic balance performance of the bearing. According to these analysis results, adjust the proportions of magnesium hydroxy silicate, boron, and tin in the formula of the metal self-repair material. Magnesium hydroxy silicate mainly provides the basic material for the repair layer, boron enhances the hardness and wear resistance of the material, and tin improves the adhesion and fluidity of the material. When the demand for balance improvement is greater than the demand for wear repair, increase the proportion of tin; when the demand for wear repair is more prominent, increase the proportion of magnesium hydroxy silicate; when it is necessary to increase the hardness of the repair layer, increase the proportion of boron. Through this targeted adjustment, an optimized metal self-repair strategy is formed to improve the repair effect.

[0122] Integrate the bearing health status score, balance improvement data, deterioration trend prediction results, and the optimized metal self-repair strategy to generate a bearing balance and deterioration assessment report. The report includes the basic information of the bearing (model, installation location, operating time, etc.), the current health status score and its position in the grading table, balance performance analysis (including the change trend of the unbalance amount, phase angle stability, etc.), deterioration trend prediction (including the prediction of the change in the health status index in the future for a period of time), repair effect evaluation (including the performance of each stage and the overall improvement effect), as well as the recommended self-repair material formula and maintenance suggestions. This report intuitively presents the health status and repair effect of the bearing, providing a comprehensive reference for subsequent maintenance decisions.

[0123] Finally, based on the theoretical value of the remaining service life and the change rate of the bearing health status score, through the Weibull distribution fitting method for correction calculation, a more accurate prediction result of the bearing remaining service life is obtained. The Weibull distribution is a classic probability distribution that describes the failure law of equipment and is applicable to the life prediction of mechanical components such as bearings. When calculating, first determine the shape parameter and scale parameter of the Weibull distribution according to historical data, and then substitute the current change rate of the health status score into the distribution function to calculate the time when the bearing reaches the preset failure threshold, obtaining the corrected predicted value of the remaining service life. This prediction result based on statistical methods takes into account the randomness of bearing deterioration and is closer to the actual situation than a simple deterministic prediction.

[0124] Taking the main shaft bearing of a 1.5 MW wind turbine in a certain wind farm as an example, after applying the metal self - repair technology to this bearing, the health status evaluation index is 78 points, which belongs to the upper layer of the attention area in the health status grading table. Five bearing failure cases of similar models and working conditions were extracted from the historical database. By analyzing the relationship between the health status index and the remaining life of these cases, a life mapping function was constructed. Substituting the health status evaluation index of this bearing into the function and considering the actual operating conditions of the current wind turbine (including average wind speed, proportion of full - load operation time, etc.), the theoretical value of the remaining service life is calculated to be 9500 hours. By analyzing the discrimination results in the repair stage, it is found that the generation stage of the repair layer of this bearing is short but the effect is obvious, indicating that the film - forming speed of the self - repair material is fast, but there is still room for improvement in the thickness of the repair layer. Combining the data on balance improvement (the unbalance amount is reduced by 35%, the phase angle tends to be stable, and the balance stability time is short), the formula of the self - repair material is adjusted. The proportion of magnesium hydroxy silicate is increased from 75% to 78%, the proportion of boron is reduced from 15% to 12%, and the proportion of tin remains unchanged at 10%. Integrating the above information to generate an evaluation report, and through Weibull distribution correction (considering that the health status score shows a steady downward trend), the predicted result of the remaining service life after correction is 8800 hours.

[0125] The above describes the generator bearing deterioration evaluation method based on the metal self - repair technology in the embodiments of the present application. Next, the generator bearing deterioration evaluation system based on the metal self - repair technology in the embodiments of the present application will be described. Please refer to Figure 2 , an embodiment of the generator bearing deterioration evaluation system based on the metal self - repair technology in the embodiments of the present application includes:

[0126] An acquisition module, configured to acquire the vibration signal, dynamic balance parameters, and temperature data of the generator bearing to obtain the time series of the comprehensive bearing operation state parameters;

[0127] An extraction module, configured to extract the bearing balance characteristics and wear characteristics according to the time series of the comprehensive bearing operation state parameters by using a multi - dimensional feature extraction algorithm to obtain the bearing deterioration index system;

[0128] A repair module, configured to inject the metal self - repair material into the generator bearing according to the bearing deterioration index system to obtain the bearing after repair treatment;

[0129] A monitoring module, configured to perform dynamic balance testing and working state monitoring on the bearing after repair treatment to obtain the discrimination result in the repair stage and the data on balance improvement;

[0130] A prediction module, configured to perform prediction analysis on the bearing balance stability and deterioration trend through a long - short - term memory network according to the discrimination result in the repair stage and the data on balance improvement to obtain the bearing health status evaluation index;

[0131] A generation module, configured to generate a bearing balance and deterioration assessment report and optimize the metal self-repair strategy according to the bearing health status assessment index, so as to obtain a prediction result of the remaining service life of the bearing.

[0132] Through the collaborative cooperation of the above-mentioned various components, multi-dimensional parameter data of the bearing is collected by a vibration sensor, a balance tester and a temperature sensor to form a time series of comprehensive parameters of the bearing operating state, providing a comprehensive data basis for subsequent analysis and avoiding the problem of incomplete diagnosis caused by a single parameter in traditional monitoring methods; using a multi-dimensional feature extraction algorithm to deeply mine the bearing balance characteristics and wear characteristics, constructing a bearing deterioration index system, realizing the precise quantification of the bearing health status, and providing a scientific basis for the optimization of the metal self-repair material formula; customizing the metal self-repair material formula according to the bearing deterioration index system and injecting it into the bearing, solving the limitation that the traditional unified formula does not adapt to different damage characteristics and improving the repair efficiency; performing dynamic balance testing and working state monitoring on the repaired bearing, obtaining the discrimination result in the repair stage and the data of the improved balance, realizing the whole-process monitoring of the repair process, and being able to detect and adjust abnormal situations in time; applying a long short-term memory network to predict and analyze the bearing balance stability and deterioration trend, making full use of the advantages of this deep learning algorithm in processing time series data, significantly improving the prediction accuracy. This algorithm can capture the complex non-linear relationships in the long time series of bearing state parameters, and is particularly suitable for the application scenarios where the bearing working conditions in a wind farm are changeable and the data noise is large. The contribution of the algorithm is reflected in promoting the traditional prediction method based on experience or simple statistics to the intelligent prediction level based on deep learning, making the prediction result closer to the actual deterioration law; finally, generating a bearing balance and deterioration assessment report according to the bearing health status assessment index and optimizing the metal self-repair strategy, and at the same time predicting the remaining service life of the bearing, providing a scientific basis for the operation and maintenance decision-making of the wind farm, realizing the transformation from after-sales maintenance to predictive maintenance, greatly reducing the unplanned downtime and maintenance cost caused by bearing failures in wind turbines, and improving the operation efficiency and economic benefits of the wind farm; in addition, this method can realize repair and evaluation without disassembling the bearing, reducing the downtime for maintenance and ensuring the continuous and stable power generation of the wind farm.

[0133] Referring to Figure 3 In the embodiment of the present invention, a computer device is further provided. The computer device may be a server, and its internal structure may be as Figure 3As shown in the figure. The computer device includes a processor, a memory, a display screen, an input device, a network interface, and a database connected through a system bus. Among them, the processor of the computer design is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. The computer program, when executed by the processor, implements the above method.

[0134] Those skilled in the art can understand that Figure 3 the structure shown in the figure is only a block diagram of some structures related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0135] An embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0136] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium provided by the present invention and used in the embodiments can include non-volatile and / or volatile memories. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or an external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM, etc.

[0137] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described systems, systems, and units can refer to the corresponding processes in the foregoing method embodiments and will not be repeated here.

[0138] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The foregoing storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs.

[0139] As described above, the above embodiments are only used to illustrate the technical solutions of the present application and are not intended to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments or equivalently replace some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present application.

Claims

1. A method for evaluating the deterioration of a generator bearing based on metal self - repair technology, characterized in that The generator bearing deterioration evaluation method based on the metal self-repair technology includes: Collecting the vibration signals, dynamic balance parameters and temperature data of the generator bearing to obtain the time series of the comprehensive bearing operation state parameters; According to the time series of the comprehensive bearing operation state parameters, using the multi-dimensional feature extraction algorithm to extract the bearing balance characteristics and wear characteristics, and obtaining the bearing deterioration index system, including: Performing multi-scale decomposition on the vibration signals in the time series of the comprehensive bearing operation state parameters through wavelet packet transform to obtain the vibration energy distribution characteristics; extracting the mean value, variance, peak factor and waveform factor of the dynamic balance parameters from the time series of the comprehensive bearing operation state parameters to obtain the quantitative balance evaluation parameters; performing correlation analysis on the temperature data and the generator speed data in the time series of the comprehensive bearing operation state parameters to calculate the temperature rise rate per unit speed and obtain the thermodynamic wear index; performing envelope demodulation processing on the vibration energy distribution characteristics through Hilbert transform to extract the characteristic frequency energies of the inner ring, outer ring, rolling elements and cage of the bearing to obtain the bearing fault characteristic spectrum; constructing the bearing balance index based on the quantitative balance evaluation parameters, constructing the bearing wear index based on the thermodynamic wear index and the bearing fault characteristic spectrum, and obtaining the bearing double deterioration evaluation benchmark; normalizing the bearing balance index and the bearing wear index and setting the weight coefficients, and generating a comprehensive evaluation standard through weighted fusion to obtain the bearing deterioration index system; According to the bearing deterioration index system, injecting the metal self-repair material into the generator bearing to obtain the repaired bearing, including: determining the formulation ratio of the metal self-repair material according to the bearing balance index and the bearing wear index in the bearing deterioration index system, and preparing nano-scale self-repair particles containing components of magnesium hydroxy silicate, boron and tin; grinding the self-repair particles to a particle size less than one hundred nanometers through a ball milling process and treating them with a surface modifier to prepare a highly dispersible metal self-repair material; adding the metal self-repair material to the extreme pressure lithium-based grease at a ratio of five percent, and mixing evenly through a high-speed stirring device to form a grease composite for repair; during the shutdown maintenance of the wind turbine, removing the oil injection port cover on the generator bearing seat, injecting the grease composite for repair into the bearing cavity, and controlling the injection amount to be seventy to eighty percent of the bearing cavity; manually rotating the bearing several circles to promote the even distribution of the grease composite for repair on the surfaces of the bearing rolling elements and raceways to form an initial repair layer; restoring the bearing seat to its original state and starting the wind turbine to make the generator operate at a low speed for four hours to form a repair transition period, and obtaining the repaired bearing; Perform dynamic balance testing and working condition monitoring on the repaired bearing to obtain the discrimination results in the repair stage and the data on the improvement of balance, including: periodically collecting vibration signals after injecting the metal self-repairing material into the bearing, calculating the change rate of the root mean square value of vibration through spectrum analysis, and forming a vibration attenuation curve; measuring the unbalance of the bearing every four hours using a balance tester, recording the data to form a balance change sequence, and obtaining the trend of dynamic balance improvement; continuously monitoring the temperature difference between the bearing temperature and the ambient temperature through a temperature sensor, calculating the unit load temperature rise ratio, and generating a temperature rise change curve; constructing a three-dimensional feature space based on the vibration attenuation curve, the trend of dynamic balance improvement, and the temperature rise change curve, projecting the data points into the three-dimensional feature space to form a repair trajectory; performing piecewise fitting on the repair trajectory, identifying the turning points in the trajectory, and dividing the repair process into a fine grinding stage, an adsorption cleaning stage, a repair layer generation stage, and a repair layer maintenance stage according to the turning points to obtain the discrimination results in the repair stage; calculating the difference in unbalance, the change in phase angle, and the balance stability time before and after repair, and combining with the stable section data in the trend of dynamic balance improvement to obtain the data on the improvement of balance; According to the discrimination results in the repair stage and the data on the improvement of balance, perform predictive analysis on the balance stability and deterioration trend of the bearing through a long short-term memory network to obtain the bearing health status evaluation index; According to the bearing health status evaluation index, generate a bearing balance and deterioration evaluation report and optimize the metal self-repair strategy to obtain the prediction result of the remaining service life of the bearing.

2. The method for evaluating the deterioration of a generator bearing based on the metal self-repair technology according to claim 1, wherein The collection of the vibration signal, dynamic balance parameters, and temperature data of the generator bearing to obtain the time series of the comprehensive bearing operating state parameters includes: Install a three-axis acceleration sensor on the generator of the wind turbine to collect the bearing vibration signal, and set the sampling frequency to 10 kHz to obtain the original bearing vibration data; Collect the generator speed data and power output data through the main control system of the wind turbine to obtain the generator operating condition parameters; Use a balance tester to collect the dynamic balance data of the generator bearing, measure the unbalance and unbalance phase angle during the rotation of the bearing, and obtain the dynamic balance characteristic parameters of the bearing; Continuously monitor the bearing housing temperature through an embedded temperature sensor, and record the ambient temperature at the same time to obtain the bearing temperature rise data; Synchronously integrate the original bearing vibration data, generator operating condition parameters, bearing dynamic balance characteristic parameters, and bearing temperature rise data according to the time stamp to form a time series database containing multi-dimensional parameters, and obtain the time series of the comprehensive bearing operating state parameters.

3. The method for evaluating the deterioration of a generator bearing based on the metal self - repair technology according to claim 1, wherein The predictive analysis on the balance stability and deterioration trend of the bearing through a long short-term memory network according to the discrimination results in the repair stage and the data on the improvement of balance to obtain the bearing health status evaluation index includes: Extract the duration ratio and conversion rate of each stage from the discrimination results in the repair stage to construct a repair dynamic feature vector; Divide the data on the improvement of balance according to the time window, calculate the balance change gradient and stability degree within each time window, and form the time series characteristics of balance stability; Time resampling is performed on the repaired dynamic feature vector and the balanced stability time series feature through a sliding window technique to generate a time series sample set for training; A long short-term memory network is used to train the time series sample set for training, establish a bearing balance stability predictor, and input future operating condition parameters to predict the change trend of the bearing balance state; According to the bearing balance state change trend and historical deterioration data, the balance weight coefficient and the wear weight coefficient are calculated by a two-way weight distribution method to form a comprehensive evaluation weight matrix; The comprehensive evaluation weight matrix is subjected to a weighted product operation with the current state parameters of the bearing, and after normalization, the bearing health state evaluation index is obtained.

4. The method for evaluating the deterioration of a generator bearing based on the metal self-repair technology according to claim 1, wherein Based on the bearing health state evaluation index, a bearing balance and deterioration evaluation report is generated and the metal self-repair strategy is optimized to obtain the bearing remaining service life prediction result, including: A health state scoring standard is set according to the bearing health state evaluation index, the index value is mapped to a zero to one hundred-point scoring interval, and a bearing health state grading table is established; Bearing failure cases with similar deterioration patterns are extracted from the historical database, and the corresponding relationship between the failure time point and the health state evaluation index is extracted to construct a life mapping function; The current bearing health state evaluation index is converted through the life mapping function, and combined with the operating condition correction parameters, the theoretical value of the bearing remaining service life is calculated; According to the repair stage discrimination result, the action efficiency of the metal self-repair material is analyzed, and combined with the balance improvement data, the proportions of magnesium hydroxy silicate, boron and tin in the metal self-repair material formula are adjusted to form an optimized metal self-repair strategy; The bearing health state score, the balance improvement data, the deterioration trend prediction result and the optimized metal self-repair strategy are integrated to generate a bearing balance and deterioration evaluation report; According to the theoretical value of the remaining service life and the change rate of the bearing health state score, a Weibull distribution fitting method is used for correction calculation to obtain the bearing remaining service life prediction result.

5. A generator bearing deterioration evaluation system based on metal self - repair technology, which is used to implement the generator bearing deterioration evaluation method based on metal self - repair technology as described in any one of claims 1 to 4, characterized in that, The generator bearing deterioration evaluation system based on the metal self-repair technology includes: An acquisition module for acquiring the vibration signal, dynamic balance parameter and temperature data of the generator bearing to obtain the bearing operation state comprehensive parameter time series; An extraction module for extracting the bearing balance feature and wear feature by using a multi-dimensional feature extraction algorithm according to the bearing operation state comprehensive parameter time series to obtain a bearing deterioration index system; A repair module for injecting a metal self-repair material into the generator bearing according to the bearing deterioration index system to obtain a repaired bearing; A monitoring module for performing dynamic balance testing and working state monitoring on the repaired bearing to obtain a repair stage discrimination result and balance improvement data; A prediction module for predicting and analyzing the bearing balance stability and deterioration trend through a long short-term memory network according to the repair stage discrimination result and the balance improvement data to obtain a bearing health state evaluation index; A generation module, configured to generate a bearing balance and deterioration assessment report and optimize a metal self-repair strategy according to the bearing health status evaluation index, so as to obtain a prediction result of the remaining service life of the bearing.

6. A computer device, characterized in that, It includes a memory and a processor. The memory stores a computer program that can run on the processor. It is characterized in that when the processor executes the computer program, it implements the generator bearing deterioration assessment method based on the metal self-repair technology described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, A computer program is stored thereon. When the computer program runs on the processor, the processor is caused to execute the generator bearing deterioration assessment method based on the metal self-repair technology described in any one of claims 1 to 4.

Images