Wind power data anomaly detection method and device, electronic equipment and storage medium

Abstract

The invention discloses a wind power data anomaly detection method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring target operation data of a wind power station; determining data of which the power is smaller than a preset threshold value in the target operation data as first abnormal data; screening out second abnormal data from the first abnormal data according to the probability of the first abnormal data; clustering the second abnormal data to obtain a clustering result, and screening out target abnormal data from the second abnormal data based on the clustering result; and training a predetermined to-be-trained wind power data anomaly detection model according to the target operation data and the target anomaly data to obtain a trained and updated wind power data anomaly detection model, and performing anomaly detection on the operation data of the wind power plant station according to the wind power data anomaly detection model. According to the technical scheme, high-precision detection of the abnormal data can be realized, and the risk of missing report and false report of the abnormal data is effectively reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power data detection, and particularly to a method, device, electronic device and storage medium for detecting abnormal wind power data. Background Art

[0002] At present, the widespread deployment of measuring points in power plants makes the collection of production equipment status data a key link in the digital monitoring of power systems. These measuring point data provide important data support for monitoring the operation status of power plants and making operation decisions for power systems. However, in actual production, affected by external environmental factors, power grid state fluctuations, aging of measuring point equipment, etc., abnormal data generally exists in the collected data set, seriously affecting the normal operation of upper-layer applications.

[0003] In the past few decades, researchers at home and abroad have carried out extensive and in-depth explorations on abnormal data detection methods in power systems, from traditional statistical methods to modern machine learning and artificial intelligence technologies. Whether it is traditional methods based on rules and models or modern technologies that use complex algorithms to automatically identify and correct data deviations, various methods have played an important role in improving detection accuracy, reducing false alarm rates, and enhancing data recovery efficiency.

[0004] However, due to the influence of climate, the measuring point data of wind power plants are highly uncontrollable, the boundary between normal data and abnormal data is relatively blurred, and it is difficult for existing solutions to accurately capture, resulting in a relatively high risk of missed reports and false alarms of abnormal data. Summary of the Invention

[0005] The present invention provides a method, device, electronic device and storage medium for detecting abnormal wind power data, which can achieve high-precision detection of abnormal data and effectively reduce the risk of missed reports and false alarms of abnormal data.

[0006] According to one aspect of the present invention, there is provided a method for detecting abnormal wind power data, the method comprising:

[0007] Obtaining target operation data of a wind power plant; wherein, the target operation data consists of data pairs composed of wind speed and power;

[0008] Determining data with power less than a preset threshold in the target operation data as first abnormal data;

[0009] Screening out second abnormal data from the first abnormal data according to the probability of the first abnormal data;

[0010] Clustering the second abnormal data to obtain a clustering result, and screening out target abnormal data from the second abnormal data based on the clustering result;

[0011] Based on the target operation data and the target abnormal data, train a pre-determined wind power data anomaly detection model to obtain a trained and updated wind power data anomaly detection model, so as to detect anomalies in the operation data of a wind farm according to the wind power data anomaly detection model.

[0012] According to another aspect of the present invention, there is provided a wind power data anomaly detection device, which includes:

[0013] A target operation data acquisition module, configured to acquire the target operation data of a wind farm; wherein, the target operation data consists of data pairs composed of wind speed and power;

[0014] A first abnormal data determination module, configured to determine the data with power less than a preset threshold in the target operation data as the first abnormal data;

[0015] A second abnormal data determination module, configured to screen out the second abnormal data from the first abnormal data according to the probability of the first abnormal data;

[0016] A target abnormal data screening module, configured to cluster the second abnormal data to obtain a clustering result, and screen out the target abnormal data from the second abnormal data based on the clustering result;

[0017] A wind power data anomaly detection model obtaining module, configured to train a pre-determined wind power data anomaly detection model based on the target operation data and the target abnormal data to obtain a trained and updated wind power data anomaly detection model, so as to detect anomalies in the operation data of a wind farm according to the wind power data anomaly detection model.

[0018] According to another aspect of the present invention, there is provided an electronic device, which includes:

[0019] At least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute a wind power data anomaly detection method according to any embodiment of the present invention.

[0020] According to another aspect of the present invention, there is provided a computer-readable storage medium, which stores computer instructions for causing a processor to implement a wind power data anomaly detection method according to any embodiment of the present invention when executed.

[0021] In the technical solution of the embodiment of the present invention, by obtaining the target operation data of a wind farm, the data with a power less than a preset threshold is identified as the first abnormal data. According to the probability distribution characteristics of the first abnormal data, the second abnormal data is screened out from it. Cluster analysis is performed on the second abnormal data, and the target abnormal data is further screened out through the clustering result. Using the target operation data and the target abnormal data, the pre-constructed wind power data anomaly detection model to be trained is iteratively trained to obtain an updated detection model for real-time anomaly detection of the operation data of the wind farm. This technical solution can achieve high-precision detection of abnormal data and effectively reduce the risks of missed reports and false reports of abnormal data.

[0022] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present invention, nor is it used to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. Brief Description of the Drawings

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0024] Figure 1 is a flowchart of a wind power data anomaly detection method provided in Embodiment 1 of the present invention;

[0025] Figure 2 is a schematic diagram of a wind power data anomaly detection process provided in Embodiment 2 of the present invention;

[0026] Figure 3 is a schematic diagram of a wind power data anomaly detection method provided in Embodiment 3 of the present invention;

[0027] Figure 4 is a schematic diagram of the structure of a wind power data anomaly detection device provided in Embodiment 4 of the present invention;

[0028] Figure 5 is a schematic diagram of the structure of an electronic device for implementing the wind power data anomaly detection method of the embodiment of the present invention. Detailed Description of the Embodiments

[0029] To enable those skilled in the art to better understand the solution of the present invention, the following will clearly and completely describe the technical solution in the embodiments of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0030] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances, so that the embodiments of the present invention described here can be implemented in an order other than those illustrated or described here. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0031] Embodiment 1

[0032] Figure 1 is a flowchart of a wind power data anomaly detection method provided according to Embodiment 1 of the present invention. This embodiment is applicable to the situation of detecting anomalies in wind power data. This method can be executed by a wind power data anomaly detection device, which can be implemented in the form of hardware and / or software, and the wind power data anomaly detection device can be configured in a device. For example, the device can be a device with communication and computing capabilities such as a background server. As Figure 1 shown, the method includes:

[0033] S110. Obtain the target operation data of the wind farm; wherein, the target operation data consists of data pairs composed of wind speed and power.

[0034] In this solution, a wind farm is a centralized facility that generates electricity using wind energy. By deploying a large number of wind turbines, wind energy is converted into electrical energy and connected to the power grid, which is an important part of the renewable energy field.

[0035] In this embodiment, the target operation data consists of data pairs composed of wind speed and power. Wind speed refers to the speed of air flow per unit time; power refers to the electrical energy output by the wind turbine per unit time.

[0036] Specifically, deploy RTU (remote terminal control system) at the wind farm station unit to be tested to collect the target operation data of the wind turbine unit. The target operation data collected by RTU is aggregated to the SCADA (Supervisory Control And Data Acquisition) system. Perform preprocessing operations on the data sets collected in the system and construct wind speed-power (vp) data pairs. Among them, the preprocessing operations include removing missing values, unifying data types, and normalizing data.

[0037] S120: Determine the data in the target operation data whose power is less than a preset threshold as first abnormal data.

[0038] The first abnormal data is composed of a data pair consisting of wind speed and power.

[0039] In this embodiment, the threshold value can be set according to the wind power data anomaly detection requirement. Preferably, the threshold value can be set to 0.

[0040] In this solution, data with power less than 0 in the target operation data is determined as the first abnormal data.

[0041] S130: Filter out second abnormal data from the first abnormal data according to the probability of the first abnormal data.

[0042] In this solution, the probability of the first abnormal data can be calculated based on a probability calculation formula; the probability of the first abnormal data can also be estimated based on the data frequency; and the probability of the first abnormal data can also be calculated using a multidimensional Gaussian distribution.

[0043] Furthermore, the first abnormal data with a probability less than or equal to a preset probability threshold may be determined as the second abnormal data, wherein the probability threshold may be set according to the abnormality detection requirement of wind power data.

[0044] S140: Cluster the second abnormal data to obtain a clustering result, and filter out target abnormal data from the second abnormal data based on the clustering result.

[0045] Among them, the target abnormal data consists of a data pair consisting of wind speed and power.

[0046] In this embodiment, the second abnormal data may be clustered using a DBSCAN algorithm to obtain a clustering result; the second abnormal data may also be clustered using a K-means algorithm to obtain a clustering result.

[0047] Further, determine the cluster centers of the largest cluster result and the remaining cluster results in the clustering results; where the remaining cluster results are the other cluster results in the clustering results except for the largest cluster result. Take the cluster center of the largest cluster result as a benchmark, and calculate the distance between the cluster centers of the remaining cluster results and this benchmark. If the distance between the two exceeds a pre-set distance threshold, then the data corresponding to the remaining cluster results is determined as target abnormal data.

[0048] S150. Train the pre-determined wind power data anomaly detection model to be trained based on the target operation data and the target abnormal data, and obtain the trained and updated wind power data anomaly detection model, so as to detect anomalies in the operation data of the wind farm according to the wind power data anomaly detection model.

[0049] Among them, the wind power data anomaly detection model is the Qwen large model. The core architecture of the Qwen model is based on Transformer. The self-attention mechanism in Transformer can effectively process long-distance context dependencies and also supports parallel computing, greatly improving the training and inference efficiency of the model.

[0050] In this solution, a training set, a validation set, and a test set can be constructed based on the target operation data and the target abnormal data, and these data sets can be used to fine-tune and iteratively train the Qwen large model to make it adapt to the task of wind power anomaly data detection.

[0051] Further, after obtaining the trained and updated wind power data anomaly detection model, detect anomalies in the operation data of the wind farm according to the wind power data anomaly detection model, and exclude the abnormal data in the operation data.

[0052] Optionally, training the pre-determined wind power data anomaly detection model to be trained based on the target operation data and the target abnormal data to obtain the trained and updated wind power data anomaly detection model includes:

[0053] Determine the initial parameters of the wind power data anomaly detection model to be trained;

[0054] Input the target operation data and the target abnormal data into the wind power data anomaly detection model to be trained, and output predicted abnormal data;

[0055] Based on the predicted abnormal data and the target abnormal data, determine the loss function value;

[0056] Adjust the initial parameters of the wind power data anomaly detection model to be trained according to the loss function value to obtain the trained and updated wind power data anomaly detection model.

[0057] In this embodiment, the wind power data anomaly detection model to be trained is the Qwen large model.

[0058] Specifically, the initial parameters of the wind power data anomaly detection model to be trained include network structure parameters and training hyperparameters. Among them, the network structure parameters include the number of model layers, the number of attention heads, and the embedding dimension, etc. These parameters determine the feature extraction ability and information interaction efficiency of the model; the training hyperparameters include key indicators such as the learning rate and the number of training epochs, which are used to regulate the training process and convergence speed of the model. By scientifically configuring these parameters, the performance of the model in the anomaly data detection task can be effectively optimized.

[0059] In this solution, the target operation data and the target anomaly data are input into the wind power data anomaly detection model to be trained. The wind power data anomaly detection model to be trained performs in-depth analysis and processing on the target operation data and the target anomaly data, executes the prediction operation, and finally outputs the predicted anomaly data.

[0060] Among them, the loss function is a function used to measure the difference between the model prediction result and the true result. The loss function can include mean squared error, mean absolute error, mean absolute percentage error, logarithmic loss function, etc.

[0061] Specifically, during the model training process, the predicted anomaly data and the target anomaly data are used to calculate the loss function value. By calculating this loss function value, the difference between the model predicted anomaly data and the target anomaly data can be scientifically quantified, and then the prediction performance and generalization ability of the model in the current training state can be comprehensively and accurately evaluated.

[0062] Furthermore, during the process of updating the parameters, multiple iterative trainings need to be performed on the entire target operation data and the target anomaly data. Each iteration repeats the above process. Through continuous iterative training, the parameters of the model are gradually adjusted, so that the predicted anomaly data of the model is getting closer and closer to the target anomaly data, and the loss function value gradually decreases. When the loss function value reaches a pre-set small value, or when the change of the loss function value is no longer obvious after multiple iterations, it is considered that the model has converged. At this time, the obtained model is the wind power data anomaly detection model after training and updating.

[0063] Specifically, the target operation data and the target anomaly data are divided according to the unit and time period to construct a sample set; the sample set is divided into a training set, a test set, and a validation set according to the ratio of 70%, 15%, and 15%; the Hugging Face open-source training tool is used to repeatedly train and evaluate the Qwen large model, change its model parameters, and construct a wind power data anomaly detection model.

[0064] By constructing a wind power data anomaly detection model, the wind power data anomaly detection model can effectively capture complex feature parameters hidden behind the data due to its unique characteristics. Compared with existing technical solutions, this model can more accurately discover abnormal data and significantly improve the detection accuracy.

[0065] Optionally, after obtaining the trained and updated wind power data anomaly detection model, the method further includes:

[0066] Obtain the operation data of the wind farm; wherein, the operation data consists of data pairs composed of wind speed and power;

[0067] Input the operation data into the wind power data anomaly detection model to output abnormal data.

[0068] In this solution, the trained wind power data anomaly detection model is deployed in a server cluster.

[0069] Specifically, collect the operation data of wind turbines, summarize the data collected by RTU into the SCADA system, and save it in the time series database IoTDB; input the operation data in the time series database into the wind power data anomaly detection model for quality inspection, and screen out abnormal data.

[0070] By detecting wind power abnormal data, the characteristics of the wind power data anomaly detection model enable it to effectively capture complex feature parameters hidden behind the data. Compared with existing technical solutions, it can more accurately discover abnormal data and improve the detection accuracy.

[0071] The technical solution of the embodiment of the present invention obtains the target operation data of the wind farm, identifies the data with power less than the preset threshold as the first abnormal data, screens out the second abnormal data according to the probability distribution characteristics of the first abnormal data, performs clustering analysis on the second abnormal data, further screens out the target abnormal data through the clustering results, and uses the target operation data and the target abnormal data to iteratively train the pre-constructed wind power data anomaly detection model to be trained to obtain an updated detection model for real-time anomaly detection of the operation data of the wind farm. By implementing this technical solution, the traditional statistical analysis method is combined with the unsupervised learning algorithm to efficiently classify wind power data and quickly construct a large-scale wind power data set. Subsequently, the large model technology is used to utilize the powerful feature extraction and pattern recognition capabilities of the large model to adaptively learn complex multi-dimensional data, achieve high-precision elimination of abnormal data, and effectively reduce the risks of missed reports and false reports of abnormal data.

[0072] Embodiment Two

[0073] Figure 2A schematic diagram of the wind power data anomaly detection process provided in the second embodiment of the present invention. The relationship between this embodiment and the above embodiment is a detailed description of the probability calculation process of the first abnormal data. As Figure 2 shown, the method includes:

[0074] S210. Obtain the target operation data of the wind farm; wherein, the target operation data consists of data pairs composed of wind speed and power.

[0075] S220. Determine the data with power less than the preset threshold in the target operation data as the first abnormal data.

[0076] S230. Sort the first abnormal data in ascending order according to wind speed and power to obtain the sorted first abnormal data.

[0077] In this solution, the wind speed can be used as the primary sorting basis and arranged in ascending order; when the wind speed values are the same, the power is used as the secondary sorting criterion and sorted again in ascending order, so as to realize the re - sorting of the first abnormal data.

[0078] S240. Divide the sorted first abnormal data according to the preset wind speed interval and preset power interval to obtain each abnormal data interval.

[0079] Among them, the wind speed interval and power interval can be set according to the requirements of wind power data anomaly detection. For example, it can be set with a power interval of 25kW and a wind speed interval of 25m / s.

[0080] In this embodiment, according to the wind speed interval and power interval, the sorted first abnormal data is divided into several equal abnormal data intervals.

[0081] S250. Determine the probability of each abnormal data interval.

[0082] In this solution, by dividing the wind speed and power data in each abnormal data interval using a multi - dimensional Gaussian distribution, the probability of each abnormal data interval appearing is determined.

[0083] Optionally, determining the probability of each abnormal data interval includes:

[0084] Calculate the mean vector and covariance matrix of the wind speed and power in the abnormal data interval;

[0085] Based on the mean vector and the covariance matrix, calculate the Mahalanobis distance of the wind speed and power in the abnormal data interval;

[0086] According to the Mahalanobis distance and the covariance matrix, calculate the probability of the abnormal data interval.

[0087] Specifically, the mean vector of wind speed and power is calculated using the following formula:

[0088] ;

[0089] Wherein, represents the mean vector of wind speed and power, represents the data pair composed of wind speed and power, represents the number of data pairs within the abnormal data interval.

[0090] In this embodiment, the covariance matrix of wind speed and power is calculated using the following formula:

[0091] ;

[0092] Wherein, represents the covariance matrix of wind speed and power.

[0093] In this solution, the Mahalanobis distance of wind speed and power is calculated using the following formula:

[0094] ;

[0095] Wherein, represents the Mahalanobis distance of wind speed and power, represents the matrix of data pairs composed of wind speed and power.

[0096] Furthermore, the probability of the abnormal data interval is calculated using the following formula:

[0097] ;

[0098] Wherein, represents the probability of the abnormal data interval, represents the dimension of the data pair composed of wind speed and power, .

[0099] By determining the probability of each abnormal data interval, the second abnormal data is screened out from the first abnormal data, and then a training data set is constructed. It has strong versatility, can adaptively discover clusters of any shape, can effectively identify various types of abnormal data clusters, and provides rich and diverse samples for the training of the wind power data anomaly detection model; it can effectively overcome the defect that the traditional solution has poor identification effect in the scenario of high proportion of abnormal data.

[0100] S260. Determine the data corresponding to the abnormal data interval whose probability is less than or equal to the preset probability threshold as the second abnormal data.

[0101] Wherein, the probability threshold can be set according to the requirements of wind power data anomaly detection. For example, the probability threshold can be set to 0.05.

[0102] Specifically, the data corresponding to the abnormal data interval with a probability less than or equal to 0.05 is classified as the second abnormal data.

[0103] S270. Cluster the second abnormal data to obtain a clustering result, and screen out target abnormal data from the second abnormal data based on the clustering result.

[0104] S280. Train a pre-determined wind power data anomaly detection model to be trained based on the target operation data and the target abnormal data, and obtain an updated wind power data anomaly detection model after training, so as to detect anomalies in the operation data of the wind farm according to the wind power data anomaly detection model.

[0105] The technical solution of the embodiment of the present invention obtains the target operation data of the wind farm, identifies the data with power less than the preset threshold as the first abnormal data among them, sorts the first abnormal data in ascending order of wind speed and power to obtain the sorted first abnormal data, and divides the sorted first abnormal data into intervals according to the preset wind speed interval and preset power interval to obtain each abnormal data interval, calculates the probability of each abnormal data interval, determines the data corresponding to the abnormal data interval with a probability less than or equal to the preset probability threshold as the second abnormal data, performs clustering analysis on the second abnormal data, further screens out target abnormal data through the clustering result, and uses the target operation data and the target abnormal data to iteratively train the pre-constructed wind power data anomaly detection model to be trained, and obtains an updated detection model for real-time anomaly detection of the operation data of the wind farm. By implementing this technical solution, high-precision detection of abnormal data can be achieved, and the risks of missed reports and false reports of abnormal data can be effectively reduced.

[0106] Embodiment III

[0107] Figure 3 FIG. is a schematic diagram of a wind power data anomaly detection method provided by Embodiment III of the present invention. The relationship between this embodiment and the above embodiment is a detailed description of the process of screening target abnormal data. As Figure 3 shown, the method includes:

[0108] S310. Obtain the target operation data of the wind farm; wherein, the target operation data consists of data pairs composed of wind speed and power.

[0109] S320. Determine the data with power less than the preset threshold in the target operation data as the first abnormal data.

[0110] S330. Screen out the second abnormal data from the first abnormal data according to the probability of the first abnormal data.

[0111] S340: Cluster the second abnormal data to obtain a first clustering result.

[0112] In this solution, a clustering algorithm may be used to cluster the second abnormal data to obtain a first clustering result. For example, a DBSCAN algorithm may be used to cluster the second abnormal data.

[0113] Optionally, clustering the second abnormal data to obtain a first clustering result includes:

[0114] Based on the wind turbine models and historical operation data in the wind farm, the clustering radius and clustering threshold are determined;

[0115] Taking each data point in the second abnormal data as a target point, and constructing a neighborhood according to the target point and the clustering radius;

[0116] determining the number of data points within the neighborhood;

[0117] If the number of the data points is greater than the clustering threshold, a neighborhood corresponding to the number of the data points is determined as a first clustering result.

[0118] In this scheme, due to the significant differences in design parameters, operating characteristics and environmental adaptability of wind turbine models, the resulting clustering results are also diverse. In order to accurately grasp the operating rules of wind farms and improve operation and maintenance efficiency, it is necessary to combine the design parameters and historical operating data of various types of wind turbines in the field to build a clustering analysis model that fits the actual working conditions. By integrating multi-dimensional operating data and optimizing the clustering algorithm in a targeted manner, it is possible to achieve a refined classification of the operating status of wind turbines, providing strong support for the formulation of scientific equipment maintenance strategies and power generation scheduling plans.

[0119] Specifically, the clustering radius eps and clustering threshold MinPts of the clustering algorithm are set.

[0120] Furthermore, each data point in the second abnormal data is taken as the target point p, and all points in its eps neighborhood are found. The eps neighborhood is defined as follows:

[0121] ;

[0122] in, are the other data points in the second abnormal data except the target point.

[0123] In this scheme, if the number of data points in the eps neighborhood is greater than MinPts, the point is marked as a core point, and all points in its eps neighborhood are marked as the same cluster, that is, marked as the first clustering result.

[0124] Further, continue to cluster each data point in the second abnormal data until the clustering of each data point in the second abnormal data is completed. Among them, if the number of data points in the neighborhood is less than MinPts, the point is marked as a noise point.

[0125] Using a method that combines traditional statistics and the DBSCAN clustering algorithm to construct a model training data set has the advantages of strong versatility and the ability to adaptively discover clusters of arbitrary shapes. It can effectively discover various types of abnormal data clusters, provide a rich variety of samples for model training, and at the same time effectively overcome the defect that the traditional scheme has poor identification effect in the scenario of high proportion of abnormal data.

[0126] S350. Cluster the first clustering result based on a preset clustering algorithm to obtain a second clustering result.

[0127] In this solution, a clustering algorithm can be used to cluster the first clustering result to obtain a second clustering result. For example, the DBSCAN algorithm can be used to cluster the first clustering result.

[0128] S360. Determine the clustering center of the largest clustering result and the clustering centers of the remaining clustering results in the second clustering result; where, the remaining clustering results are other clustering results in the second clustering result except for the largest clustering result.

[0129] In this solution, after determining the second clustering result, determine the clustering center of the largest clustering result and the clustering centers of the remaining clustering results from the second clustering result.

[0130] S370. If the distance between the clustering center of the remaining clustering results and the clustering center of the largest clustering result is greater than a preset distance threshold, determine the data corresponding to the remaining clustering results as the target abnormal data.

[0131] Among them, the distance threshold can be set according to the abnormal detection requirements of wind power data. For example, the distance threshold can be set to 0.2MinPts.

[0132] In this solution, the distance between the clustering center of the remaining clustering results and the clustering center of the largest clustering result can be calculated according to the distance calculation formula.

[0133] Specifically, when the distance between the clustering center of the remaining clustering results and the clustering center of the largest clustering result is greater than the preset distance threshold, determine the data corresponding to the remaining clustering results as the target abnormal data.

[0134] S380. Train the pre - determined wind power data anomaly detection model to be trained based on the target operation data and the target anomaly data, and obtain the updated wind power data anomaly detection model after training, so as to detect anomalies in the operation data of the wind farm according to the wind power data anomaly detection model.

[0135] In the technical solution of the embodiment of the present invention, by obtaining the target operation data of the wind farm, the data with power less than the preset threshold is identified as the first anomaly data. According to the probability distribution characteristics of the first anomaly data, the second anomaly data is screened out from it. The second anomaly data is clustered to obtain the first clustering result. Based on the preset clustering algorithm, the first clustering result is clustered to obtain the second clustering result. The clustering center of the largest clustering result and the clustering centers of the remaining clustering results in the second clustering result are determined. If the distance between the clustering centers of the remaining clustering results and the clustering center of the largest clustering result is greater than the preset distance threshold, the data corresponding to the remaining clustering results is determined as the target anomaly data. Using the target operation data and the target anomaly data, the pre - constructed wind power data anomaly detection model to be trained is iteratively trained to obtain the updated detection model for real - time anomaly detection of the operation data of the wind farm. By implementing this technical solution, the traditional statistical analysis method and the unsupervised learning algorithm are combined to efficiently classify the wind power data and quickly construct a large - scale wind power data set. Subsequently, using the large - model technology, with the powerful feature extraction and pattern recognition capabilities of the large model, it adaptively learns complex multi - dimensional data to achieve high - precision elimination of anomaly data, effectively reducing the risks of missed reports and false reports of anomaly data.

[0136] Embodiment 4

[0137] Figure 4 It is a schematic structural diagram of a wind power data anomaly detection device provided in Embodiment 4 of the present invention. As Figure 4 shown, the device includes:

[0138] A target operation data acquisition module 410, configured to acquire the target operation data of the wind farm; wherein, the target operation data consists of data pairs composed of wind speed and power;

[0139] A first anomaly data determination module 420, configured to determine the data with power less than the preset threshold in the target operation data as the first anomaly data;

[0140] A second anomaly data determination module 430, configured to screen out the second anomaly data from the first anomaly data according to the probability of the first anomaly data;

[0141] A target anomaly data screening module 440, configured to cluster the second anomaly data to obtain a clustering result, and screen out the target anomaly data from the second anomaly data based on the clustering result;

[0142] The wind power data anomaly detection model obtaining module 450 is configured to train a pre-determined wind power data anomaly detection model to be trained based on the target operation data and the target anomaly data, so as to obtain a trained and updated wind power data anomaly detection model, and perform anomaly detection on the operation data of the wind farm based on the wind power data anomaly detection model.

[0143] Optionally, the second anomaly data determination module 430 includes:

[0144] A sorting unit, configured to sort the first anomaly data in ascending order of wind speed and power to obtain the sorted first anomaly data;

[0145] An interval partitioning unit, configured to partition the sorted first anomaly data according to a preset wind speed interval and a preset power interval to obtain each anomaly data interval;

[0146] A probability determination unit, configured to determine the probability of each anomaly data interval;

[0147] A second anomaly data determination unit, configured to determine the data corresponding to the anomaly data interval with a probability less than or equal to a preset probability threshold as the second anomaly data.

[0148] Optionally, the probability determination unit is specifically configured to:

[0149] Calculate the mean vector and covariance matrix of the wind speed and power within the anomaly data interval;

[0150] Based on the mean vector and the covariance matrix, calculate the Mahalanobis distance of the wind speed and power within the anomaly data interval;

[0151] Calculate the probability of the anomaly data interval according to the Mahalanobis distance and the covariance matrix.

[0152] Optionally, the target anomaly data screening module 440 includes:

[0153] A first clustering result obtaining unit, configured to cluster the second anomaly data to obtain a first clustering result;

[0154] A second clustering result obtaining unit, configured to cluster the first clustering result based on a preset clustering algorithm to obtain a second clustering result;

[0155] A clustering center determination unit, configured to determine the clustering center of the largest clustering result and the clustering centers of the remaining clustering results in the second clustering result; wherein, the remaining clustering results are other clustering results in the second clustering result except the largest clustering result;

[0156] A target abnormal data determination unit, configured to determine the data corresponding to the remaining clustering results as target abnormal data if the distance between the clustering center of the remaining clustering results and the clustering center of the maximum clustering result is greater than a preset distance threshold.

[0157] Optionally, the first clustering result obtaining unit is specifically configured to:

[0158] Determine a clustering radius and a clustering threshold based on the fan models and historical operation data in the wind farm;

[0159] Take each data point in the second abnormal data as a target point, and construct a neighborhood according to the target point and the clustering radius;

[0160] Determine the number of data points in the neighborhood;

[0161] If the number of data points is greater than the clustering threshold, determine the neighborhood corresponding to the number of data points as the first clustering result.

[0162] Optionally, the wind power data anomaly detection model obtaining module 450 is specifically configured to:

[0163] Determine the initial parameters of the wind power data anomaly detection model to be trained;

[0164] Input the target operation data and the target abnormal data into the wind power data anomaly detection model to be trained, and output predicted abnormal data;

[0165] Determine a loss function value based on the predicted abnormal data and the target abnormal data;

[0166] Adjust the initial parameters of the wind power data anomaly detection model to be trained according to the loss function value, and obtain a trained and updated wind power data anomaly detection model.

[0167] Optionally, the device further includes:

[0168] An operation data acquisition module, configured to acquire the operation data of the wind farm; wherein, the operation data consists of data pairs of wind speed and power;

[0169] An abnormal data output module, configured to input the operation data into the wind power data anomaly detection model and output abnormal data.

[0170] The wind power data anomaly detection device provided by the embodiments of the present invention can execute the wind power data anomaly detection method provided by any embodiment of the present invention, and has corresponding functional modules and beneficial effects for executing the method.

[0171] Embodiment Five

[0172] Figure 5 FIG. 1 shows a schematic structural diagram of an electronic device 10 that can be used to implement an embodiment of the present invention. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0173] As Figure 5 shown, the electronic device 10 includes at least one processor 11, and a memory communicatively connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc. The memory stores a computer program executable by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0174] A plurality of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0175] The processor 11 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as a method for detecting abnormal wind power data.

[0176] In some embodiments, a method for detecting abnormal wind power data can be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the method for detecting abnormal wind power data described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to execute a method for detecting abnormal wind power data by any other suitable means (e.g., by means of firmware).

[0177] Various embodiments of the systems and techniques described above in this document can be implemented in digital electronic circuitry, integrated circuit systems, field programmable gate arrays (FPGA), application specific integrated circuits (ASIC), application specific standard products (ASSP), system on a chip (SOC), complex programmable logic devices (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs executable and / or interpretable on a programmable system including at least one programmable processor, which can be a special-purpose or general-purpose programmable processor that receives data and instructions from a storage system, at least one input device, and at least one output device, and transmits the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0178] The computer program for implementing the method of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer programs are executed by the processor, the functions / operations specified in the flowchart and / or block diagram are implemented. The computer programs can be executed entirely on the machine, partially on the machine, as a stand-alone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0179] In the context of the present invention, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, the computer-readable storage medium can be a machine-readable signal medium. More specific examples of the machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0180] In order to provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0181] The systems and techniques described herein can be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer having a graphical user interface or a web browser through which the user can interact with an implementation of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected to each other by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local area network (LAN), wide area network (WAN), blockchain network, and the Internet.

[0182] A computing system may include a client and a server. The client and the server are generally far from each other and usually interact via a communication network. The client-server relationship is created by computer programs running on respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system, and solves the defects of difficult management and weak business scalability existing in traditional physical hosts and VPS services.

[0183] It should be understood that various forms of the processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in the present invention can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved, and no limitation is made herein.

[0184] The above specific embodiments do not constitute a limitation on the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for detecting abnormal wind power data, characterized in that, Including: Obtain the target operation data of the wind farm; wherein, the target operation data consists of data pairs composed of wind speed and power; Determine the data with power less than the preset threshold in the target operation data as the first abnormal data; Screen out the second abnormal data from the first abnormal data according to the probability of the first abnormal data; Cluster the second abnormal data to obtain a clustering result, and screen out the target abnormal data from the second abnormal data based on the clustering result; Train the pre-determined wind power data anomaly detection model to be trained based on the target operation data and the target abnormal data to obtain an updated wind power data anomaly detection model after training, so as to detect anomalies in the operation data of the wind farm according to the wind power data anomaly detection model.

2. The method according to claim 1, characterized in that, Screening out the second abnormal data from the first abnormal data according to the probability of the first abnormal data includes: Sort the first abnormal data according to the rules of increasing wind speed and power to obtain the sorted first abnormal data; Perform interval division on the sorted first abnormal data according to the preset wind speed interval and preset power interval to obtain each abnormal data interval; Determine the probability of each abnormal data interval; Determine the data corresponding to the abnormal data interval with the probability less than or equal to the preset probability threshold as the second abnormal data.

3. The method according to claim 2, characterized in that, Determining the probability of each abnormal data interval includes: Calculate the mean vector and covariance matrix of the wind speed and power within the abnormal data interval; Calculate the Mahalanobis distance of the wind speed and power within the abnormal data interval based on the mean vector and the covariance matrix; Calculate the probability of the abnormal data interval according to the Mahalanobis distance and the covariance matrix.

4. The method according to claim 1, wherein Clustering the second abnormal data to obtain a clustering result, and screening out the target abnormal data from the second abnormal data based on the clustering result includes: Cluster the second abnormal data to obtain a first clustering result; Cluster the first clustering result based on the preset clustering algorithm to obtain a second clustering result; Determine the clustering center of the largest clustering result and the clustering centers of the remaining clustering results in the second clustering result; wherein, the remaining clustering results are other clustering results in the second clustering result except the largest clustering result; If the distance between the clustering center of the remaining clustering results and the clustering center of the largest clustering result is greater than the preset distance threshold, then determine the data corresponding to the remaining clustering results as the target abnormal data.

5. The method according to claim 4, wherein Clustering the second abnormal data to obtain a first clustering result includes: Determine the clustering radius and clustering threshold based on the fan models and historical operation data in the wind farm; Take each data point in the second abnormal data as the target point, and construct a neighborhood according to the target point and the clustering radius; Determine the number of data points in the neighborhood; If the number of data points is greater than the clustering threshold, then determine the neighborhood corresponding to the number of data points as the first clustering result.

6. The method according to claim 1, characterized in that Training the pre-determined wind power data anomaly detection model to be trained based on the target operation data and the target anomaly data to obtain the trained and updated wind power data anomaly detection model, including: Determining the initial parameters of the wind power data anomaly detection model to be trained; Inputting the target operation data and the target anomaly data into the wind power data anomaly detection model to be trained, and outputting predicted anomaly data; Determining the loss function value based on the predicted anomaly data and the target anomaly data; Adjusting the initial parameters of the wind power data anomaly detection model to be trained according to the loss function value to obtain the trained and updated wind power data anomaly detection model.

7. The method according to claim 1, wherein After obtaining the trained and updated wind power data anomaly detection model, the method further includes: Obtaining the operation data of the wind farm; wherein, the operation data consists of data pairs composed of wind speed and power; Inputting the operation data into the wind power data anomaly detection model and outputting anomaly data.

8. A wind power data anomaly detection device, characterized in that, Including: A target operation data acquisition module, configured to acquire the target operation data of the wind farm; wherein, the target operation data consists of data pairs composed of wind speed and power; A first anomaly data determination module, configured to determine, as the first anomaly data, the data in the target operation data whose power is less than a preset threshold; A second anomaly data determination module, configured to screen out the second anomaly data from the first anomaly data according to the probability of the first anomaly data; A target anomaly data screening module, configured to cluster the second anomaly data to obtain a clustering result, and screen out the target anomaly data from the second anomaly data based on the clustering result; A wind power data anomaly detection model obtaining module, configured to train the pre-determined wind power data anomaly detection model to be trained based on the target operation data and the target anomaly data to obtain the trained and updated wind power data anomaly detection model, so as to perform anomaly detection on the operation data of the wind farm according to the wind power data anomaly detection model.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute a wind power data anomaly detection method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a processor to implement a wind power data anomaly detection method according to any one of claims 1-7 when executed.

Images