Wind generating set data processing method and system, electronic equipment and storage medium

Abstract

The invention relates to a wind generating set data processing method and system, electronic equipment and a storage medium, and the method comprises the steps: collecting and uploading multi-dimensional operation data of a wind generating set in real time; preprocessing and normalizing the multi-dimensional operation data, and extracting key features from the processed data through a convolutional neural network to obtain key feature data; acquiring a historical data set and a Logistic regression model, and training the Logistic regression model according to the historical data set to obtain the fault early warning model; and calculating the key feature data through the fault early warning model to obtain a fault occurrence probability, and when the fault occurrence probability is greater than a preset threshold value, sending out an early warning signal. Through accurate data processing and model training, it is ensured that the model fitting degree is higher, the prediction error is smaller, the fault of the transmission chain of the wind generating set can be predicted more accurately, early warning is achieved, and the accuracy and reliability of fault early warning are improved.

Description

Technical Field

[0001] The present application relates to the field of power data processing, and in particular to a method, system, electronic device and storage medium for processing data of a wind turbine generator set. Background Art

[0002] Wind turbines may encounter various faults during operation, among which transmission chain failure is one of the most common faults. Traditional fault diagnosis methods usually rely on human experience and offline monitoring, and have problems such as low diagnostic accuracy and slow response speed. The wind turbine transmission chain is the core power generation component of the wind turbine. Once it fails, it will bring huge maintenance and replacement costs.

[0003] Therefore, a new wind turbine data processing solution is urgently needed to solve the above technical problems. Summary of the invention

[0004] The embodiments of the present application provide a wind turbine data processing method, system, electronic device and storage medium to at least solve the problems of low diagnostic accuracy, slow response speed and the like in the traditional fault diagnosis method in the related art.

[0005] In a first aspect, an embodiment of the present application provides a method for processing data of a wind turbine generator set, comprising:

[0006] Collect and upload multi-dimensional operating data of wind turbines in real time;

[0007] Preprocessing and normalizing the multidimensional operation data, and extracting key features from the processed data through a convolutional neural network to obtain key feature data;

[0008] Acquire a historical data set and a Logistic regression model, and train the Logistic regression model according to the historical data set to obtain the fault warning model, wherein the historical data set includes multi-dimensional operation data collected historically;

[0009] The key characteristic data is calculated by the fault warning model to obtain the probability of fault occurrence, and when the probability of fault occurrence is greater than a preset threshold, a warning signal is issued.

[0010] In one embodiment, the preprocessing and normalizing the multidimensional operation data includes:

[0011] Decompressing the multidimensional operation data, and interpolating missing values ​​of the decompressed multidimensional operation data according to physical characteristics of different measuring points to obtain complete multidimensional operation data;

[0012] Calculate the quartiles in the complete multidimensional operation data using the quartile method, determine the upper and lower limits of abnormal values ​​in the data according to the quartiles, and remove the complete multidimensional operation data that are not within the upper and lower limits to obtain normal multidimensional operation data;

[0013] The normal multi-dimensional operation data is normalized to obtain data to be characterized.

[0014] In one embodiment, the quartile method is used to calculate the quartiles in the complete multidimensional operation data, the upper and lower limits of the abnormal values ​​in the data are determined according to the quartiles, and the complete multidimensional operation data that is not within the upper and lower limits is processed to obtain normal multidimensional operation data, including:

[0015] Calculating the first quartile and the third quartile of the multidimensional running data;

[0016] Calculate the interquartile range based on the first quartile and the third quartile;

[0017] Determine the upper and lower limits of the outlier value according to the interquartile range;

[0018] Identifying and marking abnormal values ​​from the multidimensional operation data according to the upper and lower limit ranges;

[0019] The marked outliers are processed by deletion, replacement or format conversion.

[0020] In one embodiment, extracting key features from the preprocessed data by a convolutional neural network to obtain key feature data includes:

[0021] Obtain a pre-built convolutional neural network architecture, wherein the convolutional neural network architecture includes an input layer, a convolution layer, a pooling layer, and a fully connected layer;

[0022] The data to be characterized is received through the input layer, and specific features in the data to be characterized are extracted through the convolution layer to obtain a key feature matrix; the key feature matrix is ​​subjected to dimensionality reduction processing through the pooling layer; and the key feature matrix after dimensionality reduction is mapped to the final output category through the fully connected layer to obtain the key feature data.

[0023] In one embodiment, the construction of the fault warning model includes:

[0024] Obtain risk factors for failure and corresponding risk factor coefficients, and construct an initial failure warning model based on the risk factors, risk factor coefficients and a Logistic regression model;

[0025] The coefficients of the initial fault warning model are solved by the maximum likelihood estimation method to obtain regression coefficients, and the regression coefficients are substituted into the initial fault warning model to obtain the fault warning model.

[0026] In one embodiment, the method further comprises:

[0027] Select a corresponding SVM kernel function according to user needs, and use grid search and cross-validation methods to adjust the parameters of the SVM to obtain an initial SVM model;

[0028] Obtain a labeled training data set to train the SVM model, and use a test data set to evaluate the performance of the SVM model;

[0029] The trained SVM model is used to predict the fault category of the key feature data, and a nonlinear kernel function is used to perform optimal calculation to obtain fault attribute classification of different key feature data.

[0030] In a second aspect, an embodiment of the present application provides a wind turbine generator set data processing system, which is used to run the wind turbine generator set data processing method described in any one of the above embodiments, and the system includes:

[0031] A data acquisition module is used to collect multi-dimensional operation data of the wind turbine generator set in real time and upload it, and send the multi-dimensional operation data to the data processing module;

[0032] A data processing module, used for preprocessing and normalizing the multi-dimensional operation data, and extracting key features from the preprocessed data through a convolutional neural network, obtaining key feature data and sending it to the fault warning module;

[0033] A data iteration module, used for acquiring a historical data set and a Logistic regression model, training the Logistic regression model according to the historical data set, obtaining the fault warning model and sending it to the fault warning module, wherein the historical data set includes multi-dimensional operation data collected historically;

[0034] The fault warning module is used to receive the key feature data and run the fault warning model, calculate the key feature data through the fault warning model to obtain the probability of fault occurrence, and issue a warning signal when the probability of fault occurrence is greater than a preset threshold.

[0035] In one embodiment, the data acquisition module includes a cache unit, and the cache unit is communicatively connected to a plurality of acquisition points of the wind turbine generator set;

[0036] The real-time collection and uploading of multi-dimensional operation data of the wind turbine generator set, and sending the multi-dimensional operation data to the data processing module, include:

[0037] The cache unit is monitored in real time, and when it is detected that new collected data is recorded, the new multi-dimensional operation data is compressed and encrypted to obtain the multi-dimensional operation data;

[0038] The data acquisition module transmits the multi-dimensional operation data from the acquisition point to the data processing module using a low-latency transmission protocol and a multiplexing method;

[0039] An abnormality detection mechanism is used to monitor the data transmission process, and when a data transmission failure is detected, the multi-dimensional operation data is resent.

[0040] In one embodiment, the data processing module includes a preprocessing unit and a feature extraction unit, wherein:

[0041] The preprocessing unit is used to perform preprocessing and normalization processing on the multidimensional operation data;

[0042] The feature extraction unit is used to extract key features from the preprocessed data through a convolutional neural network.

[0043] In a third aspect, an embodiment of the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the wind turbine data processing method as described in the first aspect above is implemented.

[0044] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the wind turbine generator set data processing method as described in the first aspect above.

[0045] The wind turbine data processing method, system, electronic device and storage medium provided by the embodiments of the present application have at least the following technical effects:

[0046] The quartile method is used to process abnormal data to ensure the quality and reliability of the data. The data is normalized to make the features within the same scale range, which is conducive to subsequent feature extraction and model training. The convolutional neural network is used to extract key features from the processed data to ensure that the extracted features are highly correlated with the fault warning. The fault warning data attribute classification is implemented based on the support vector machine algorithm and nonlinear kernel function to improve the accuracy of fault identification. The fault warning model is implemented using the Logistic regression algorithm to predict the probability of fault occurrence. The regression coefficient is solved by the maximum likelihood estimation method to ensure the optimal estimation of the model parameters and improve the accuracy and reliability of the fault warning. Through precise data processing and model training, the model fit is higher and the prediction error is smaller. It can more accurately predict the fault of the wind turbine transmission chain and achieve early warning.

[0047] Details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0049] Figure 1 is a structural block diagram of a wind turbine generator data processing system in one embodiment of the present application;

[0050] Figure 2 is a flow chart of a method for processing wind turbine generator data in one embodiment of the present application;

[0051] Figure 3 It is a structural block diagram of an electronic device in one embodiment of the present application. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. Based on the embodiments provided in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0053] Obviously, the drawings described below are only some examples or embodiments of the present application. For ordinary technicians in this field, the present application can also be applied to other similar scenarios based on these drawings without creative work. In addition, it can also be understood that although the efforts made in this development process may be complicated and lengthy, for ordinary technicians in this field related to the content disclosed in this application, some changes in design, manufacturing or production based on the technical content disclosed in this application are just conventional technical means, and should not be understood as insufficient content disclosed in this application.

[0054] Reference to "embodiments" in this application means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those of ordinary skill in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0055] Unless otherwise defined, the technical terms or scientific terms involved in this application should be understood by people with ordinary skills in the technical field to which this application belongs. The words "one", "a", "a", "the" and the like involved in this application do not indicate a quantitative limitation, and may represent the singular or plural. The terms "include", "comprise", "have" and any of their variations involved in this application are intended to cover non-exclusive inclusions; for example, a process, method, system, product or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units that are not listed, or may also include other steps or units inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The "multiple" involved in this application refers to two or more. "And / or" describes the association relationship of associated objects, indicating that there may be three relationships, for example, "A and / or B" can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects before and after are in an "or" relationship. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.

[0056] refer to Figure 1The embodiment of the present application provides a wind turbine data processing system. The system of the present embodiment includes a data acquisition module, a data processing module, a fault warning module and a data iteration module. The data processing system of the present application can run Figure 2 The following contents are combined with the wind turbine data processing method in Figure 1 and Figure 2 The processing system and processing method of the present application are described.

[0057] In this embodiment, the data acquisition module is used to collect and upload multidimensional operation data of the wind turbine in real time, and send the multidimensional operation data to the data processing module; the data processing module is used to preprocess and normalize the multidimensional operation data, and extract key features from the preprocessed data through a convolutional neural network, obtain key feature data and send it to the fault warning module; the data iteration module is used to obtain a historical data set and a Logistic regression model, train the Logistic regression model according to the historical data set, obtain the fault warning model and send it to the fault warning module, wherein the historical data set includes multidimensional operation data collected historically; the fault warning module is used to receive the key feature data and run the fault warning model, calculate the key feature data through the fault warning model, and obtain the probability of fault occurrence, and when the probability of fault occurrence is greater than a preset threshold, issue a warning signal.

[0058] In this embodiment, the data acquisition module includes a cache unit, which is communicatively connected to multiple collection points of the wind turbine generator set; wherein, when data collection is performed, the cache unit is monitored in real time, and when new collection data is detected to be recorded, the new multidimensional operation data is compressed and encrypted to obtain multidimensional operation data; the data acquisition module uses a low-latency transmission protocol and a multiplexing method to transmit the multidimensional operation data from the collection point to the data processing module; an abnormality detection mechanism is used to monitor the data transmission process, and when a data transmission failure is detected, the multidimensional operation data is resent.

[0059] Specifically, the data acquisition module is not only responsible for collecting the operating data of the wind turbine, including but not limited to multi-dimensional data such as wind speed, temperature, and vibration signals, but also has the ability to transmit the collected data to the edge computing device (ie, the data processing module) in real time.

[0060] Specifically, the data acquisition module achieves this function in the following ways:

[0061] The first is real-time data stream transmission. The data acquisition module is designed to transmit the collected data to the edge computing device in real time. Whenever a new data point is recorded, the data will be sent out immediately, rather than waiting until the data accumulates to a certain amount before batch transmission. This method ensures that the latest data can be received and processed by the edge computing device in a timely manner, thereby improving the timeliness of fault warning;

[0062] The second is the low-latency transmission protocol. In order to ensure low latency of data transmission, the data acquisition module adopts a transmission protocol that can quickly transmit data from the collection point to the edge computing node, reducing data lag caused by network delays. This is especially important for fault warning systems that require real-time monitoring and rapid response.

[0063] The third is data compression and encryption. Before data transmission, the data acquisition module will compress the data to reduce the time and bandwidth required for transmission. At the same time, in order to protect the security of the data, the data in transmission will be encrypted to prevent the data from being stolen or tampered with during the transmission process.

[0064] Fourth, the data acquisition module has the ability to dynamically adjust the transmission rate. According to the network conditions and the processing power of the edge computing device, it can automatically adjust the speed of data transmission to ensure the stability and reliability of data transmission even in poor network conditions.

[0065] The fifth is the anomaly detection and retransmission mechanism. In order to deal with possible packet loss or transmission errors during data transmission, the data acquisition module is also equipped with an anomaly detection and data retransmission mechanism. Once a data transmission failure is detected, the system will automatically resend the relevant data until it is confirmed that the edge computing device has successfully received it.

[0066] The sixth is the multiplexing technology. The data acquisition module supports multiplexing technology and can handle data collection and transmission tasks from multiple data sources at the same time. In this way, even in the face of large-scale or multi-point distributed wind turbines, it can efficiently complete the centralized transmission of data.

[0067] Through the above measures, the data acquisition module can not only ensure the accuracy and completeness of data collection, but also ensure that these data can be transmitted to the edge computing device in a timely and secure manner, providing a solid foundation for subsequent data processing and fault warning.

[0068] In a detailed embodiment, the preprocessing unit is used to receive data from the data acquisition module, clean the collected data, use the quartile method to process abnormal data, and if there are abnormal values ​​in the sample data, determine the upper and lower limits of the abnormal values ​​in the sample data according to the interquartile range, and normalize the data. In order to achieve the above functions, the data processing module includes a preprocessing unit and a feature extraction unit, wherein the preprocessing unit is used to preprocess and normalize the multidimensional operation data; the feature extraction unit is used to extract key features from the preprocessed data through a convolutional neural network.

[0069] Specifically, the preprocessing unit can clean up the dirty data and abnormal data in the operating data, and reasonably interpolate or remove the null values ​​in the data according to the physical characteristics of different measuring points. After that, the reasonable range of the data of each measuring point is calculated in combination with statistical principles and the data beyond the range is cleaned. The extraction of high-quality healthy data sets through the monitoring configuration software (SCADA) data preprocessing work is the basis for building a wind turbine fault warning model. The SCADA abnormal data processing method is used to eliminate the abnormal data generated by the uncertainty of wind speed and wind direction. Then, the SCADA data parameter selection method is used to eliminate the information redundancy in the SCADA data to avoid the model training falling into the local minimum, or even the gradient explosion directly affecting the training process of the model. Finally, the data normalization operation is used to eliminate the influence of different dimensions between the wind turbine parameters, so as to accelerate the gradient descent speed during model training, improve the prediction accuracy of the model, and may further extend the prediction step. In order to eliminate the influence of the corresponding dimensions of various parameters, prevent the parameters with larger orders of magnitude from dominating during training, reduce the order of magnitude difference to accelerate the iterative convergence speed, and normalize the sample data.

[0070] During the entire data processing and analysis process of wind turbine transmission chain fault detection, an artificial intelligence algorithm is designed to adapt to different operating conditions of wind turbines and realize the matching relationship between characteristics and fault modes under different operating conditions to achieve better processing results.

[0071] The feature extraction unit uses convolutional neural networks to extract key features from the processed data and implements fault warning data attribute classification based on support vector machine algorithms and nonlinear kernel functions;

[0072] The original data of wind turbine transmission chain failure has multiple dimensions, and contains redundant and low-correlation feature dimensions. The convolutional neural network can improve the feature extraction ability of data through convolution pooling. The convolution operation of the convolution kernel can obtain multiple convolution feature maps, which can effectively extract the key features of the original input data. The convolution operation is expressed as formula (1):

[0073]

[0074] In the formula, represents the n′th weight in the mth convolution kernel of the lth layer; x l(n+n′) It represents the n′th weight perception position in the nth convolved local area of ​​the lth layer; c represents the size of the convolution kernel.

[0075] Through the feature compression of the pooling layer, the feature matrix can be reduced in dimension, thereby significantly reducing the model training parameters to obtain the main features. Among them, the maximum pooling operation takes the maximum value in the perception area as the output of the pooling layer, which is expressed as:

[0076] p l(m，t) =y l(m，n) max (n-1)c+1≤1≤jw {a l(m，t)} (2)

[0077] In the formula, a l(m，t) It is represented as the tth activation value of the mth feature map of the lth layer, and w is the pooling width.

[0078] Based on the above content, the foundation is laid for realizing early warning of transmission chain failure of wind turbine generator sets.

[0079] For the fault warning data attribute classification based on the support vector machine algorithm and nonlinear kernel function, the main purpose of the wind turbine transmission chain fault warning control is to realize the autonomous and intelligent control of the wind turbine transmission chain, reliably analyze the operating status of the wind turbine transmission chain, send fault warnings in advance before the transmission chain abnormality occurs and does not cause failures, reduce the operation risk of the transmission chain, and ensure multi-dimensional fault warnings. Therefore, the application layer establishes a fault warning model based on the processing and analysis of the transmission chain operation data of other layers to realize multi-dimensional fault warning and control of the transmission chain, and uses the support vector machine algorithm to complete the establishment of the fault warning mechanism. Since the fault warning control problems have significant inherent coupling correlation, collaborative requirements and multi-dimensional changes, the support vector machine algorithm can deeply mine the data to realize multi-dimensional fault warning control laws, and introduce nonlinear kernel functions to perform optimal calculations to realize fault warning data attribute classification. The algorithm is used to perform attribute division on the multi-dimensional data of the transmission chain fault warning control, which is expressed as:

[0080]

[0081] In the formula, f(x) represents the optimal classification decision function of the data, the symbolic function is represented by sign, b represents the set parameter, x represents the unknown quantity, α m represents the regularization coefficient, y m represents a binary variable, and σ represents the coefficient of volatility.

[0082] After completing the cleaning and preprocessing of the wind turbine transmission chain data, the data feature extraction based on the convolutional neural network, and the division of the transmission chain warning data attributes, the wind turbine transmission chain fault warning based on the edge AI computing architecture can be realized.

[0083] Edge computing uses chips for high-frequency sampling and real-time statistical operations, periodically calculates a set of characteristic parameters, and transmits them to the cloud server through the communication module for diagnosis and predictive analysis. The edge AI computing architecture randomly arranges the specific time of data upload in each preset time period to make the uploaded data volume as uniform as possible. The gateway and controller are regarded as edge computing nodes with computing, storage and data transmission capabilities. For wind turbine transmission chain data collection, the quartile method is used for abnormal data processing, that is, the data is divided into four equal parts, where the interquartile range I QR is the difference between the third quartile and the first quartile, which is expressed as:

[0084] I QR =Q 3 -Q 1 (4)

[0085] In the formula, Q 3 represents the third quartile, Q 1 Represents the first quartile. When there are outliers in the sample data, the upper and lower limits of the outliers in the sample data can be determined based on the interquartile range [F d， F u ], and its calculation is expressed as:

[0086]

[0087] Fault warning module, in order to ensure the reliability of fault warning, the Logistic regression algorithm is used to complete the fault warning. The warning value is represented by E. The warning value E is the threshold for triggering the warning in general. There is no range limit for it in this application. There are many factors that cause transmission chain risks, such as temperature, load rate, etc. The Logistic regression fault warning model is constructed as follows:

[0088]

[0089] In the formula, the regression coefficient of the risk influencing factor is β 1 , β 2 , …, β j Indicates; k 1 , …, k j represents the risk factor; p represents the risk probability. After obtaining the transmission chain fault warning result, the probability of fault occurrence is calculated:

[0090]

[0091] T is mainly used to compare the difference in the mean values ​​of two sets of data. It is an existing technology for solving the regression coefficient by the maximum likelihood estimation method. After the result of formula (6) is obtained, it can be substituted into formula (7) to solve formula (7). The maximum likelihood estimation method is used to solve the regression coefficient e. After the above formula is solved, the probability of fault occurrence can be obtained, and fault identification can be completed. At the same time, an early warning is sent to the display interface to complete the transmission chain fault early warning. At this point, combined with the above-mentioned related content, the design of the wind turbine transmission chain fault early warning algorithm based on the edge AI computing architecture is completed.

[0092] The data iteration module uses the data classified by the data processing module to establish a historical data set and replaces the initial data one by one. According to the feature importance score of the random forest, the top-ranked features are selected. According to the feature importance score, the best feature subset is selected and input into the Logistic regression model for training to obtain the trained fault warning model.

[0093] To sum up, the wind turbine data processing system provided in the embodiment of the present application, the wind turbine data processing system and method, show higher accuracy in the fault warning of the wind turbine transmission chain, can effectively extract the key features of the original input data, and at the same time have comprehensive fault warning management and control functions, can reliably obtain fault warning results, have higher working efficiency, higher fitting degree and smaller error, and can more accurately predict the fault of the wind turbine transmission chain and achieve early warning.

[0094] In a second aspect, the present application provides a method for processing wind turbine generator data. Figure 2 , including the following processes.

[0095] Step S1, collecting and uploading multi-dimensional operation data of the wind turbine generator set in real time.

[0096] Step S2, preprocessing and normalizing the multidimensional operation data, and extracting key features from the processed data through a convolutional neural network to obtain key feature data. Specifically, the multidimensional operation data is decompressed, and the decompressed multidimensional operation data is interpolated for missing values ​​according to the physical characteristics of different measuring points to obtain complete multidimensional operation data; the quartile method is used to calculate the quartiles in the complete multidimensional operation data, and the upper and lower limits of the abnormal values ​​in the data are determined according to the quartiles, and the complete multidimensional operation data that is not within the upper and lower limits is eliminated to obtain normal multidimensional operation data; the normal multidimensional operation data is normalized to obtain data to be characterized.

[0097] Among them, the quartile method is to calculate the first quartile and the third quartile of the multidimensional operation data; calculate the interquartile range based on the first quartile and the third quartile; determine the upper and lower limits of the abnormal value based on the interquartile range; identify and mark the abnormal values ​​from the multidimensional operation data based on the upper and lower limits; and process the marked abnormal values ​​to delete, replace or convert the format.

[0098] Step S3, obtaining a historical data set and a Logistic regression model, and training the Logistic regression model according to the historical data set to obtain the fault warning model, wherein the historical data set includes multi-dimensional operation data collected historically.

[0099] Specifically, a pre-built convolutional neural network architecture is obtained, wherein the convolutional neural network architecture includes an input layer, a convolutional layer, a pooling layer, and a fully connected layer; the data to be characterized is received through the input layer, and specific features in the data to be characterized are extracted through the convolutional layer to obtain a key feature matrix; the key feature matrix is ​​subjected to dimensionality reduction processing through the pooling layer; and the key feature matrix after dimensionality reduction is mapped to a final output category through the fully connected layer to obtain the key feature data.

[0100] In a preferred embodiment, the method also includes: selecting a corresponding SVM kernel function according to user needs, and using grid search and cross-validation methods to adjust the parameters of the SVM to obtain an initial SVM model; obtaining a labeled training data set to train the SVM model, and using a test data set to evaluate the performance of the SVM model; using the trained SVM model to predict the fault category of the key feature data, and using a nonlinear kernel function for optimal calculation to obtain a fault attribute classification of different key feature data.

[0101] Step S4, calculating the key characteristic data through the fault warning model to obtain the probability of fault occurrence, and issuing a warning signal when the probability of fault occurrence is greater than a preset threshold.

[0102] Among them, the construction of the fault warning model includes: obtaining the risk factors of failure and the corresponding risk factor coefficients, and constructing an initial fault warning model according to the risk factors, risk factor coefficients and Logistic regression model; solving the coefficients of the initial fault warning model by the maximum likelihood estimation method to obtain the regression coefficients, and substituting the regression coefficients into the initial fault warning model to obtain the fault warning model.

[0103] In a more specific embodiment, the wind turbine generator data processing method adopts the following technical solution:

[0104] S01, the data acquisition module is responsible for collecting the operating data of the wind turbine generator set using sensors, including but not limited to multi-dimensional data collected by wind speed sensors, temperature sensors and vibration sensors;

[0105] S02. The preprocessing unit performs preliminary cleaning of the data collected by the data acquisition module, removes or marks missing values ​​and obviously erroneous data points, and reasonably interpolates or removes the null values ​​in the data according to the physical characteristics of different measuring points;

[0106] S03. The preprocessing unit uses the quartile method to process abnormal data, marks and processes all data points below or above the upper and lower limits as abnormal values, and uses the SCADA data parameter selection method to eliminate information redundancy, avoid model training from falling into local minima or gradient explosion, and normalize the cleaned data to eliminate the influence of different dimensions between wind turbine parameters, speed up the gradient descent speed during model training, and improve the prediction accuracy, making it suitable for subsequent data processing and analysis;

[0107] S04, the feature extraction unit inputs the data normalized by the preprocessing unit into the convolutional neural network, obtains multiple convolution feature maps through the convolution operation of the convolution kernel, improves the feature extraction capability, reduces the model training parameters through the feature compression of the pooling layer, and obtains the main features;

[0108] S05. The feature extraction unit uses the support vector machine algorithm and nonlinear kernel function to classify the extracted key features to identify the data attributes related to fault warning, deeply explore the multi-dimensional fault warning control rules in the data, and realize the classification of fault warning data attributes;

[0109] S06. The fault warning model constructs a Logistic regression model and uses the maximum likelihood estimation method to solve the regression coefficient;

[0110] S07. Use cross-validation method to evaluate the generalization ability of fault warning model;

[0111] S08. Input the extracted key features into the trained Logistic regression model to calculate the probability of failure;

[0112] S09. Complete fault identification based on the calculated fault occurrence probability, and issue a warning if the fault occurrence probability exceeds a predetermined threshold.

[0113] The data iteration module uses the data classified by the data processing module to establish a historical data set and replace the initial data one by one. According to the feature importance score of the random forest, the top-ranked features are selected. According to the feature importance score, the best feature subset is selected and input into the Logistic regression model for training to obtain the trained fault warning model. The specific steps include the following:

[0114] Create a historical data set:

[0115] Data collection: Data collected for the first time or historically constitutes the historical data set;

[0116] Data preparation: pre-process historical data, including data cleaning, missing value processing, feature engineering, etc., to ensure data quality;

[0117] Feature extraction: Extract useful features from historical data, such as vibration frequency, temperature change, current fluctuation, etc.;

[0118] Model training: Use historical data sets to train the initial fault warning model;

[0119] Real-time data processing and early warning:

[0120] Data collection: collect the operating data of wind turbines in real time through sensors;

[0121] Data preprocessing: preprocess the collected real-time data, including data cleaning and normalization, to ensure that the real-time data is consistent with the format of the historical data set;

[0122] Input model: input the preprocessed real-time data into the trained Logistic regression model;

[0123] Calculate probability: Use the model to predict real-time data and calculate the warning probability of each sample;

[0124] Warning output: Warning output is performed based on the calculated probability. If the probability exceeds the preset threshold, a warning signal is issued;

[0125] Data screening and analysis:

[0126] Data screening: Based on the warning probability output by the model, a threshold is set (for example, 0.8), and data above this threshold is used as candidate warning data. The candidate warning data is initially filtered to remove obviously erroneous or abnormal data points, and the correlation between the warning data and other known fault data is checked to exclude isolated events or atypical situations.

[0127] Case analysis: For each warning case, review historical data to see if there are similar cases, check the situation and treatment measures at the time, analyze warning data from multiple dimensions, such as time series analysis, trend analysis, pattern matching, etc., check the change trend of warning data over time, whether there are periodic or sudden changes, analyze the change trend of warning data over time, determine whether it shows an upward or downward trend, compare warning data with known fault modes, and find similarities;

[0128] Effectiveness judgment: Use deep neural network models to extract features and classify warning data to improve judgment accuracy;

[0129] Multi-source information fusion: Fusion of warning data with multi-source information (such as environmental data, operation records, etc.), comprehensive judgment of the effectiveness of warning, use of association rule mining technology, analysis of the correlation between warning data and other related factors, and consideration of the contextual information of the warning, such as weather conditions, equipment operating status, etc., to improve the comprehensiveness of the judgment;

[0130] Result confirmation: All analysis results are summarized to determine which warnings are real and effective. The warning data that are confirmed to be real and effective are recorded to prepare for subsequent data updates and model iterations. A feedback mechanism is established to record the warning data that are confirmed to be invalid so that the model can be improved in the future.

[0131] Data update and model iteration:

[0132] Data cleaning: Perform necessary cleaning on new data to remove noise and outliers;

[0133] Data balance: Check whether the newly merged data set has any category imbalance problem, count the number of data in each category, and determine whether there is a serious imbalance. If data imbalance is detected, take corresponding data balancing strategies, oversample the minority class data to increase its number, simply copy the minority class data to increase its number, generate synthetic samples through interpolation to increase the minority class data, undersample the majority class data to reduce its number, randomly delete part of the majority class data, use the nearest neighbor algorithm to delete samples similar to the majority class data, combine oversampling and undersampling methods to achieve data balance, oversample the minority class data, undersample the majority class data, first use SMOTE to generate synthetic samples, and then use nearest neighbor undersampling to reduce the majority class data;

[0134] Model retraining:

[0135] Model loading: load the existing model framework and prepare for training;

[0136] Feature selection: Use the feature importance score in random forest to evaluate the importance of existing features, perform correlation analysis, remove highly correlated features, avoid multicollinearity problems, use recursive feature elimination methods to gradually remove unimportant features until the best feature combination is found, introduce new features or transform existing features according to the characteristics of the new data set to capture more effective information;

[0137] Hyperparameter tuning: define the candidate range of hyperparameters, try all possible combinations one by one through grid search to find the best hyperparameter configuration, define a reasonable candidate value range for each hyperparameter, use cross-validation to evaluate the effect of each set of hyperparameters within the candidate range, select the best hyperparameter combination based on the cross-validation results, randomly select hyperparameter combinations from the hyperparameter candidate range for evaluation, randomly select hyperparameter combinations from the candidate range, conduct multiple experiments to reduce the impact of chance, select the best hyperparameter combination based on the experimental results, use Bayesian optimization methods to estimate the optimal values ​​of hyperparameters by building a probability model, build a Gaussian process or other probability model to estimate the effect of hyperparameters, define an objective function that maximizes the performance indicators of the model, optimize hyperparameters in an iterative manner, select the hyperparameter combination that is most likely to improve performance for evaluation each time, set convergence conditions, and stop optimization when the threshold is reached or after a certain number of iterations.

[0138] The generalization ability of the model is evaluated by the cross-validation method. The specific operations are as follows:

[0139] Divide the dataset:

[0140] Divide the historical data set into a training set and a test set, with 80% of the data used for training and 20% for testing;

[0141] When performing cross-validation, the training set is further divided into K subsets (K = 5 or 10);

[0142] Perform K-fold cross validation:

[0143] Select K-1 subsets from the K subsets as training data, and the remaining subset as validation data;

[0144] Use K-1 subsets of data to train the Logistic regression model;

[0145] Use the remaining subset (validation set) to evaluate the performance of the model;

[0146] Repeat the above steps K times, selecting a different subset as the validation set each time;

[0147] Collect performance indicators for each verification (such as accuracy, precision, recall, F1 score, etc.);

[0148] Calculate the average performance:

[0149] Calculate the average performance index of K validations to evaluate the overall generalization ability of the model;

[0150] Comprehensive indicators such as average accuracy, average precision, and average recall can be calculated;

[0151] Select the best model:

[0152] Based on the results of cross-validation, select a set of regression coefficients with the best performance;

[0153] This set of regression coefficients is the optimal solution obtained through comprehensive analysis of all K validation results;

[0154] Final model evaluation:

[0155] Use the test set (data that was not previously trained) to evaluate the final selected model;

[0156] This step is to further verify the performance of the model on completely independent data.

[0157] Through the above steps, the present invention can achieve early warning of transmission chain failure of wind turbine generator sets, thereby effectively reducing the risk of transmission chain operation and improving the operating efficiency and safety of wind turbine generator sets.

[0158] In summary, the present invention provides a complete solution for wind turbine transmission chain fault warning by integrating data acquisition, preprocessing, feature extraction, fault warning modeling and warning output links, combining edge computing technology and artificial intelligence algorithms.

[0159] Experimental preparation:

[0160] In order to verify the effectiveness of the wind turbine transmission chain fault warning algorithm based on edge AI computing architecture proposed in the present invention, the following experiment was carried out. The present invention built a wind turbine fault simulation experimental platform, which consists of a frequency converter, a fault simulation test bench, a collector, and collection software. The high-speed end speed of the frequency converter is 1800r / min, and the sampling frequency is 512kHZ. The motor speed is set at 1000r / min, and the sampling frequency is 10.24kHZ. Data under different states of the transmission chain are obtained respectively. 250 groups of transmission chain data in different states are collected, totaling 1000 groups. The object of the experimental fault warning is the wind turbine of mainstream domestic manufacturers. Through continuous operation and field test verification in the past year, 96 actual field fault records have been found so far.

[0161] Result analysis:

[0162] The wind turbine transmission chain fault warning algorithm based on edge AI computing architecture proposed in the present invention is compared with the scraper conveyor chain transmission system fault warning method based on LSTM-Adam and the scraper conveyor chain transmission system fault detection and diagnosis method based on ANSYS. The actual fault warning results are compared by using the method of the present invention, the LSTM-Adam method and the ANSYS method respectively. The application status of the wind turbine transmission chain health warning is shown in Table 1;

[0163] Table 1 Application of wind turbine transmission chain health warning

[0164]

[0165]

[0166] According to Table 1, the wind turbine transmission chain fault warning algorithm based on the edge AI computing architecture proposed by the present invention can warn of faults in the gearbox and main bearing, while the LSTM-Adam method can only warn of abnormal data of the main bearing temperature sensor and abnormal data of the non-driving end temperature sensor of the gearbox intermediate shaft, and the ANSYS method can only warn of abnormal data of the gearbox oil pool temperature, abnormal data of the main bearing temperature sensor and abnormal data of the non-driving end temperature sensor of the gearbox intermediate shaft. Therefore, it can be considered that the wind turbine transmission chain fault warning algorithm based on the edge AI computing architecture designed by the present invention has certain feasibility.

[0167] Table 2 Comparison of active warning accuracy of different methods

[0168]

[0169] The active warning accuracy of the method of the present invention is higher than 80%, and it can effectively extract the key features of the original input data during the actual operation and maintenance application of the wind turbine transmission chain. The active warning accuracy of the two methods is relatively low, which proves that the fault warning curve of the method of the present invention has a higher degree of fit with the actual curve, a smaller error, and a more superior performance. The fault warning effect is better and can achieve early warning of the wind turbine transmission chain. Therefore, the method of the present invention has a comprehensive fault warning control function and can reliably obtain fault warning results. It has extremely high working energy efficiency during the application process and meets the user's application needs for fault warning control of the wind turbine transmission chain.

[0170] In a third aspect, an embodiment of the present application provides an electronic device, Figure 3 FIG. 1 is a block diagram of an electronic device according to an exemplary embodiment. Figure 3 As shown, the electronic device may include a processor 11 and a memory 12 storing computer program instructions.

[0171] Specifically, the processor 11 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0172] Among them, the memory 12 may include a large-capacity memory for data or instructions. By way of example and not limitation, the memory 12 may include a hard disk drive (HDD), a floppy disk drive, a solid-state drive (SSD), a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 12 may include a removable or non-removable (or fixed) medium. Where appropriate, the memory 12 may be inside or outside a data processing device. In a specific embodiment, the memory 12 is a non-volatile memory. In a specific embodiment, the memory 12 includes a read-only memory (ROM) and a random access memory (RAM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (Programmable Read-Only Memory, PROM for short), an erasable PROM (Erasable ProgrammableRead-Only Memory, EPROM for short), an electrically erasable PROM (Electrically Erasable ProgrammableRead-Only Memory, EEPROM for short), an electrically alterable ROM (Electrically Alterable Read-Only Memory, EAROM for short) or a flash memory (FLASH) or a combination of two or more of these. Under appropriate circumstances, the RAM can be a static random access memory (SRAM) or a dynamic random access memory (DRAM), wherein the DRAM can be a fast page mode dynamic random access memory (FPMDRAM), an extended data output dynamic random access memory (EDODRAM), a synchronous dynamic random access memory (SDRAM), etc.

[0173] The memory 12 may be used to store or cache various data files that need to be processed and / or used for communication, as well as possible computer program instructions executed by the processor 11 .

[0174] The processor 11 implements any one of the wind turbine generator set data processing methods in the above embodiments by reading and executing computer program instructions stored in the memory 12 .

[0175] In one embodiment, the electronic device may further include a communication interface 13 and a bus 10. Figure 3 As shown, the processor 11, the memory 12, and the communication interface 13 are connected via a bus 10 and communicate with each other.

[0176] The communication interface 13 is used to realize the communication between the modules, devices, units and / or equipment in the embodiment of the present application. The communication port 13 can also realize data communication with other components such as: external devices, image / data acquisition equipment, databases, external storage and image / data processing workstations.

[0177] The bus 10 includes hardware, software or both, and couples the components of the electronic device to each other. The bus 10 includes but is not limited to at least one of the following: a data bus, an address bus, a control bus, an expansion bus, and a local bus. By way of example and not limitation, bus 10 may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus, or other suitable buses or a combination of two or more of these. Where appropriate, bus 10 may include one or more buses. Although embodiments of the present application describe and illustrate a particular bus, the present application contemplates any suitable bus or interconnect.

[0178] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a program stored thereon, and when the program is executed by a processor, the wind turbine generator set data processing method provided in the first aspect is implemented.

[0179] The readable storage medium may include but is not limited to: a portable disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, an optical storage device, a magnetic storage device or any suitable combination of the above.

[0180] In a possible implementation, the present invention can also be implemented in the form of a program product, which includes program code. When the program product is run on a terminal device, the program code is used to enable the terminal device to execute the steps of the wind turbine data processing method provided in the first aspect.

[0181] The program code for executing the present invention may be written in any combination of one or more programming languages, and may be executed entirely on a user device, partially on a user device, as an independent software package, partially on a user device and partially on a remote device, or entirely on a remote device.

[0182] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0183] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A method for processing wind turbine generator data, characterized in that: include: Collect and upload multi-dimensional operating data of wind turbines in real time; Preprocessing and normalizing the multidimensional operation data, and extracting key features from the processed data through a convolutional neural network to obtain key feature data; Acquire a historical data set and a Logistic regression model, and train the Logistic regression model according to the historical data set to obtain a fault warning model, wherein the historical data set includes multi-dimensional operation data collected historically; The key characteristic data is calculated by the fault warning model to obtain the probability of fault occurrence, and when the probability of fault occurrence is greater than a preset threshold, a warning signal is issued.

2. The wind turbine generator data processing method according to claim 1, characterized in that: The preprocessing and normalizing of the multi-dimensional operation data includes: Decompressing the multidimensional operation data, and interpolating missing values ​​of the decompressed multidimensional operation data according to physical characteristics of different measuring points to obtain complete multidimensional operation data; Calculate the quartiles in the complete multidimensional operation data using the quartile method, determine the upper and lower limits of abnormal values ​​in the data according to the quartiles, and remove the complete multidimensional operation data that are not within the upper and lower limits to obtain normal multidimensional operation data; The normal multi-dimensional operation data is normalized to obtain data to be characterized.

3. The wind turbine generator data processing method according to claim 2, characterized in that: The method of calculating the quartiles in the complete multidimensional operation data by using the quartile method, determining the upper and lower limits of abnormal values ​​in the data according to the quartiles, and processing the complete multidimensional operation data that is not within the upper and lower limits to obtain normal multidimensional operation data includes: Calculating the first quartile and the third quartile of the multidimensional running data; Calculate the interquartile range based on the first quartile and the third quartile; Determine the upper and lower limits of the outlier value according to the interquartile range; Identifying and marking abnormal values ​​from the multidimensional operation data according to the upper and lower limit ranges; The marked outliers are processed by deletion, replacement or format conversion.

4. The wind turbine generator data processing method according to claim 1, characterized in that: The key features are extracted from the preprocessed data by using a convolutional neural network to obtain key feature data, including: Obtain a pre-built convolutional neural network architecture, wherein the convolutional neural network architecture includes an input layer, a convolution layer, a pooling layer, and a fully connected layer; The data to be characterized is received through the input layer, and specific features in the data to be characterized are extracted through the convolution layer to obtain a key feature matrix; the key feature matrix is ​​subjected to dimensionality reduction processing through the pooling layer; and the key feature matrix after dimensionality reduction is mapped to the final output category through the fully connected layer to obtain the key feature data.

5. The wind turbine generator data processing method according to claim 1, characterized in that: The construction of the fault warning model includes: Obtain risk factors for failure and corresponding risk factor coefficients, and construct an initial failure warning model based on the risk factors, risk factor coefficients and a Logistic regression model; The coefficients of the initial fault warning model are solved by the maximum likelihood estimation method to obtain regression coefficients, and the regression coefficients are substituted into the initial fault warning model to obtain the fault warning model.

6. The method for processing wind turbine generator data according to claim 1, characterized in that: The method further comprises: Select a corresponding SVM kernel function according to user needs, and use grid search and cross-validation methods to adjust the parameters of the SVM to obtain an initial SVM model; Obtain a labeled training data set to train the SVM model, and use a test data set to evaluate the performance of the SVM model; The trained SVM model is used to predict the fault category of the key feature data, and a nonlinear kernel function is used to perform optimal calculation to obtain fault attribute classification of different key feature data.

7. A wind turbine generator data processing system, characterized in that: The system is used to run the wind turbine generator data processing method according to any one of claims 1 to 6, comprising: A data acquisition module is used to collect multi-dimensional operation data of the wind turbine generator set in real time and upload it, and send the multi-dimensional operation data to the data processing module; A data processing module, used for preprocessing and normalizing the multi-dimensional operation data, and extracting key features from the preprocessed data through a convolutional neural network, obtaining key feature data and sending it to the fault warning module; A data iteration module, used for acquiring a historical data set and a Logistic regression model, training the Logistic regression model according to the historical data set, obtaining the fault warning model and sending it to the fault warning module, wherein the historical data set includes multi-dimensional operation data collected historically; The fault warning module is used to receive the key feature data and run the fault warning model, calculate the key feature data through the fault warning model to obtain the probability of fault occurrence, and issue a warning signal when the probability of fault occurrence is greater than a preset threshold.

8. The wind turbine generator data processing system according to claim 7, characterized in that: The data acquisition module includes a cache unit, which is communicatively connected to multiple acquisition points of the wind turbine generator set; The real-time collection and uploading of multi-dimensional operation data of the wind turbine generator set, and sending the multi-dimensional operation data to the data processing module, include: The cache unit is monitored in real time, and when it is detected that new collected data is recorded, the new multi-dimensional operation data is compressed and encrypted to obtain the multi-dimensional operation data; The data acquisition module transmits the multi-dimensional operation data from the acquisition point to the data processing module using a low-latency transmission protocol and a multiplexing method; An abnormality detection mechanism is used to monitor the data transmission process, and when a data transmission failure is detected, the multi-dimensional operation data is resent.

9. The wind turbine generator data processing system according to claim 7, characterized in that: The data processing module includes a preprocessing unit and a feature extraction unit, wherein: The preprocessing unit is used to perform preprocessing and normalization processing on the multidimensional operation data; The feature extraction unit is used to extract key features from the preprocessed data through a convolutional neural network.

10. An electronic device, characterized in that: The invention comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the wind turbine generator set data processing method according to any one of claims 1 to 7 when executing the computer program.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the wind turbine generator set data processing method according to any one of claims 1 to 7 is implemented.

Images