Wind power invalid data identification method and system based on adaptive DBSCAN clustering, electronic device and storage medium

Abstract

The present application relates to a wind power invalid data identification method and system based on adaptive DBSCAN clustering, an electronic device and a storage medium, comprising: obtaining an original wind power data set of a wind turbine in a normal operating state; dividing the original wind power data set into m original wind power data subsets according to the size of wind speed using a K-means algorithm; clustering the m original wind power data subsets using an adaptive DBSCAN algorithm, outputting the clustered noise point data as wind power invalid data, obtaining all wind power invalid data in the original wind power data set, dividing the original wind power data set into different categories through a k-means clustering algorithm, improving the similarity of data in the same category, and reducing the similarity between different categories, reducing the subjective influence of human setting parameters by adaptively selecting the parameters of the DBSCAN algorithm, and improving the stability of the identification result.

Description

Technical Field

[0001] This invention belongs to the field of wind power technology, and in particular relates to a method, system, electronic device and storage medium for identifying invalid wind power data based on adaptive DBSCAN clustering. Background Technology

[0002] Wind power is considered a green, low-carbon, and sustainable clean energy source, and its development is of great significance. Wind power generation refers to generating electricity using wind power. In wind power generation, the wind turbine is one of the key pieces of equipment. The turbine records data such as wind speed, wind direction, power, and rotor speed at regular intervals. This data is called wind power data. However, due to various reasons such as meteorological conditions, wind power data may sometimes contain errors or invalid data. For example, equipment failure, abnormal data processing, or external interference can lead to abnormal wind power data. The presence of invalid data can adversely affect wind power prediction and turbine performance monitoring in wind power generation systems. Therefore, the identification and processing of invalid data can play an important role in industry and economics. By effectively identifying and removing invalid data, the performance of wind power generation systems can be improved, their stability and reliability enhanced, and the risks and costs of wind power generation reduced. Simultaneously, for large-scale wind farms, the identification and processing of invalid data can also improve the utilization rate of wind resources, reduce investment costs, and thus better promote the development of renewable energy.

[0003] Currently, the most commonly used invalid data identification methods in the power industry include statistical identification methods, curve modeling, and clustering-based methods. Statistical identification methods essentially use statistical characteristics such as mean, variance, median, and quantiles to identify invalid data with significant deviations. However, when the proportion of outliers in the original data is large, statistical identification methods are not thorough enough in identifying invalid data. Curve modeling uses a large amount of normal data to build a curve model and identifies data outside the boundary as outliers. While this method has high algorithm accuracy, it takes longer and requires a large amount of normal data as the basis for model building, making it unsuitable for processing real-time data. Clustering-based identification methods are typical unsupervised learning algorithms. Depending on the clustering principle, clustering algorithms can be divided into distance-based clustering, density-based clustering, hierarchical clustering, and grid-based clustering. Most clustering algorithms are now quite sensitive to parameter selection; different parameter settings lead to different identification results. Human factors have a significant impact on the identification results, resulting in inconsistent invalid data identification performance and low stability. Summary of the Invention

[0004] To address the problems existing in the background technology, this invention provides a method, system, electronic device, and storage medium for identifying invalid wind power data based on adaptive DBSCAN clustering. This addresses the issues that existing clustering algorithms are highly sensitive to parameter selection, leading to different identification results due to different parameter settings, and that human factors have a significant impact on the identification results, resulting in inconsistent identification performance and low stability of invalid data.

[0005] To achieve the above-mentioned technical objectives, one aspect of the present invention provides a method for identifying invalid wind power data based on adaptive DBSCAN clustering, comprising:

[0006] S1: Obtain the raw wind power dataset of the wind turbine under normal operating conditions and preprocess the raw wind power dataset;

[0007] S2: Based on the wind speed, the original wind power dataset is divided into m subsets of original wind power data using the K-means algorithm;

[0008] Preferably, the step of dividing the original wind power dataset into m subsets using the K-means algorithm includes:

[0009] S21: Use the elbow method to determine the number m of the original wind power data subsets, treat each original wind power data as a data point, and treat each original wind power data subset as a cluster;

[0010] S22: Randomly select m data points as center points, and define the loss function J based on the distance between each data point and the center point of its cluster:

[0011]

[0012] Among them, C i μ represents the set of all data points in the i-th cluster. i Let |x-μ| represent the center point of the i-th cluster. i | represents the distance from data point x to the center point μ. i The Euclidean distance, where J represents the loss function;

[0013] S23: Let t = 0, 1, 2, ... be the number of iterations, and repeat the following process until the loss function J converges;

[0014] S24: For each data point, assign it to the cluster to which the nearest center point belongs;

[0015] S25: For each cluster, recalculate the center point of that cluster; once the loss function converges, complete the partitioning of the original wind power data subset.

[0016] S3: Use the adaptive DBSCAN algorithm to cluster the m original wind power data subsets respectively, and output the noise point data after clustering as invalid wind power data to obtain all invalid wind power data in the original wind power dataset.

[0017] Preferably, the step of clustering the m original wind power data subsets using the adaptive DBSCAN algorithm includes:

[0018] S31: For each subset of original wind power data, calculate the distance between each pair of data points in the subset of original wind power data to construct the distance distribution matrix of the subset of original wind power data;

[0019] S32: Arrange the elements in each row of the distance distribution matrix in ascending order to obtain the first intermediate distance distribution matrix;

[0020] S33: The second intermediate distance distribution matrix is ​​obtained by arranging the elements in each column of the first intermediate distance distribution matrix in descending order from largest to smallest;

[0021] S34: Plot each column element in the second intermediate distance distribution matrix as a broken line, and select the broken line that best represents the distribution of data points in the original wind power data subset as the target broken line;

[0022] S35: Fit the target polynomial line using a polynomial function, and solve the polynomial function to obtain the neighborhood radius parameter Eps of the DBSCAN algorithm;

[0023] Preferably, fitting the target polynomial line using a polynomial function includes:

[0024] Initialize the polynomial function f(x) = ax 3 +bc 2 +cx+d, substitute the data points on the target polynomial line into the polynomial function f(x) to solve for the parameters a, b, c, and d. Given the parameters of the polynomial function f(x), solve for the solution x0 when the first derivative of the polynomial function f(x) is equal to -1, and use it as the neighborhood radius parameter Eps of the DBSCAN algorithm. When there are two values ​​for the solution x0 when the first derivative of the polynomial function f(x) is equal to -1, the smaller value x0 is selected as the neighborhood radius parameter Eps of the DBSCAN algorithm.

[0025] S36: Determine the minimum number of neighborhood points MinPts for the DBSCAN algorithm based on the neighborhood radius parameter Eps of the DBSCAN algorithm;

[0026] Preferably, the minimum number of neighborhood points MinPts includes:

[0027]

[0028] Among them, P i The value of represents the number of data points contained within the neighborhood radius Eps of the i-th data point in the original wind power data subset. MinPts represents the minimum number of neighborhood points in the DBSCAN algorithm, and n represents the number of data points in the original wind power data subset.

[0029] S37: Cluster the data points in the original wind power data subset according to the neighborhood radius parameter Eps and the minimum number of neighborhood points MinPts of the DBSCAN algorithm, and divide the data points in the original wind power data subset into core points, boundary points and noise points.

[0030] Preferably, the clustering of data points in the original wind power data subset based on the neighborhood radius parameter Eps and the minimum number of neighborhood points MinPts of the DBSCAN algorithm includes:

[0031] Each data point in the original wind power data subset is treated as an object point. If the number of data points contained within the neighborhood radius Eps of the object point exceeds MinPts, then the object point is a core point. If the number of data points contained within the neighborhood radius Eps of the object point is less than MinPts, but the object point falls within the neighborhood radius of other core points, then the object point is a boundary point. Otherwise, the object point is a noise point.

[0032] Another aspect of the present invention provides a wind power invalid data identification system based on adaptive DBSCAN clustering. The identification system is based on the wind power invalid data identification method based on adaptive DBSCAN clustering, and is characterized by comprising: a data acquisition module, a data preprocessing module, a K-means clustering module, and a DBSCAN clustering module.

[0033] The data acquisition module is used to acquire the original wind power dataset of the wind turbine under normal operating conditions.

[0034] The data preprocessing module is used to preprocess the original wind power dataset;

[0035] The K-means clustering module is used to divide the original wind power dataset into m subsets of original wind power data based on wind speed using the K-means algorithm;

[0036] The DBSCAN clustering module is used to cluster the m original wind power data subsets respectively, and output the clustered noise point data as invalid wind power data to obtain all invalid wind power data in the original wind power dataset.

[0037] Another aspect of the present invention provides an electronic device, including a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements a wind power invalid data identification method based on adaptive DBSCAN clustering.

[0038] Another aspect of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements a method for identifying invalid wind power data based on adaptive DBSCAN clustering.

[0039] The present invention has at least the following beneficial effects

[0040] This invention utilizes the k-means clustering algorithm to cluster data, dividing the original wind power dataset into different categories. This improves the similarity of data within the same category and reduces the similarity between different categories, enabling accurate identification of invalid data across the entire wind speed range. By employing a segmented adaptive DBSCAN algorithm instead of traditional clustering methods, such as single k-means and single DBSCAN, which have poor identification performance in areas near the cut-off wind speed and in areas with wind speeds exceeding the rated speed, this invention overcomes the dependence of traditional clustering algorithms on the number of clusters or neighborhood parameters. Furthermore, this invention adaptively sets the parameters of the DBSCAN clustering algorithm based on the data distribution of the divided original wind power data subsets, reducing the influence of subjective factors from manually setting parameters, improving the stability of the identification results, and overcoming the problem of poor identification performance caused by different parameters. Attached Figure Description

[0041] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0042] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0043] The accompanying drawings are for illustrative purposes only and are schematic diagrams, not actual pictures. They should not be construed as limiting the invention. To better illustrate the embodiments of the invention, some parts in the drawings may be omitted, enlarged, or reduced, and do not represent the actual product dimensions. It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0044] In the accompanying drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components. In the description of the present invention, it should be understood that if terms such as "upper," "lower," "left," "right," "front," and "rear" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the terms used to describe positional relationships in the drawings are only for illustrative purposes and should not be construed as limiting the present invention. For those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0045] Please see Figure 1 One aspect of the present invention provides a method for identifying invalid wind power data based on adaptive DBSCAN clustering, comprising:

[0046] S1: Obtain the raw wind power dataset of the wind turbine under normal operating conditions and preprocess the raw wind power dataset;

[0047] In this embodiment, the raw wind power data is collected from the operating data of the wind turbine units by modern sensors and monitoring equipment, including multiple indicators such as wind speed, rotational speed, power, current, and vibration. For example, the data of the wind farm is collected using a SCADA system (SCADA system is a computer-based production process control and scheduling automation system that can monitor and control the operating equipment on site). The wind farm has 33 1.6MW doubly-fed wind turbine units. The cut-in wind speed of the units is 2m / s, the rated wind speed is 10m / s, and the cut-out wind speed is 20m / s. The SCADA data sampling frequency is 1S / s. This paper uses 10 minutes of data collected by SCADA as the raw wind power dataset.

[0048] In this embodiment, the preprocessing of the original wind power dataset includes: removing invalid wind power data in the original wind power dataset where the power is less than zero and the wind speed is less than the cut-in wind speed or greater than the cut-out wind speed.

[0049] S2: Based on the wind speed, the original wind power dataset is divided into m subsets of original wind power data using the K-means algorithm;

[0050] Preferably, the step of dividing the original wind power dataset into m subsets using the K-means algorithm includes:

[0051] S21: Use the elbow method to determine the number m of the original wind power data subsets, treat each original wind power data as a data point, and treat each original wind power data subset as a cluster;

[0052] S22: Randomly select m data points as center points, and define the loss function J based on the distance between each data point and the center point of its cluster:

[0053]

[0054] Among them, C i μ represents the set of all data points in the i-th cluster. i Let |x-μ| represent the center point of the i-th cluster. i | represents the distance from data point x to the center point μ. i The Euclidean distance, where J represents the loss function;

[0055] S23: Let t = 0, 1, 2, ... be the number of iterations, and repeat the following process until the loss function J converges;

[0056] S24: For each data point, assign it to the cluster to which the nearest center point belongs;

[0057] S25: For each cluster, recalculate the cluster centroids. Once the loss function converges, the original wind power data subset is partitioned. The loss function J is the objective function of the k-means algorithm; the smaller its value, the more compact the clusters are, the greater the distance between different clusters, and the better the clustering effect. In the k-means algorithm, the centroids are updated iteratively, causing J to continuously decrease until convergence or the number of iterations is reached.

[0058] This invention utilizes the k-means clustering algorithm to cluster data, dividing the original wind power dataset into different categories to improve the similarity of data within the same category and reduce the similarity between different categories, thereby achieving accurate identification of invalid data in wind power monitoring data across the entire wind speed range.

[0059] S3: Use the adaptive DBSCAN algorithm to cluster the m original wind power data subsets respectively, and output the noise point data after clustering as invalid wind power data to obtain all invalid wind power data in the original wind power dataset.

[0060] DBSCAN is a density-based clustering algorithm capable of discovering clusters of arbitrary shape and size, and it performs well on noisy datasets. The basic idea of ​​DBSCAN is to consider sample points within a given radius Eps as neighbors, and to determine core points using a minimum number of neighbors, MinPts. Core points are sample points with a sufficient number of neighbors within a neighborhood radius Eps. The main definitions of the DBSCAN algorithm are given below:

[0061] Definition 1: (Eps Neighborhood) Given a point p in space, the Eps neighborhood of p is the set of points within a hypersphere centered at p and with radius Eps, i.e.:

[0062] N Eps ={q∈D|dist(p,q)≤Eps}

[0063] 1) In the formula, D is the sample dataset, and dist(p,q) is the distance between points p and q.

[0064] Definition 2: Definition 2 (Core Points and Boundary Points) Given the parameters "neighborhood radius Eps" and "minimum number of neighborhoods min_samples", for an object p∈D, if:

[0065] |N Eps (p)∣≥MinPts

[0066] Then p is called a core point; objects that are not core points but are within the "Eps" neighborhood of a certain core point are called boundary points.

[0067] Preferably, the step of clustering the m original wind power data subsets using the adaptive DBSCAN algorithm includes:

[0068] S31: For each subset of original wind power data, calculate the distance between each pair of data points in the subset of original wind power data to construct the distance distribution matrix of the subset of original wind power data;

[0069] DIST n×n ={dist(i,j)|1≤i≤n,1≤j≤n}

[0070] Where n is the number of data points contained in the original wind power data subset; DIST n×n It is an n-row, n-column real symmetric matrix, where dist(i,j) represents the distance between the i-th data point and the j-th data point.

[0071] S32: Arrange the elements in each row of the distance distribution matrix in ascending order to obtain the first intermediate distance distribution matrix;

[0072] S43: The second intermediate distance distribution matrix is ​​obtained by arranging the elements in each column of the first intermediate distance distribution matrix in descending order from largest to smallest;

[0073] S34: Plot each column element in the second intermediate distance distribution matrix as a broken line, and select the broken line that best represents the distribution of data points in the original wind power data subset as the target broken line;

[0074] Draw each column of elements as a polyline to obtain dist. k k = 1, 2, ..., n. Based on prior information, the broken line that best represents the distribution of the original wind power data subset is selected as the target broken line. A large amount of existing prior information shows that the curve with k = 4 best represents the distribution of the original wind power data subset. Therefore, this invention selects the curve dist4 as the target curve.

[0075] S35: Fit the target polynomial line using a polynomial function, and solve the polynomial function to obtain the neighborhood radius parameter Eps of the DBSCAN algorithm;

[0076] Preferably, fitting the target polynomial line using a polynomial function includes:

[0077] Initialize the polynomial function f(x) = ax 3 +bx 2 +cx+d, substitute the data points on the target polynomial line into the polynomial function f(x) to solve for the parameters a, b, c, and d. Given the parameters of the polynomial function f(x), solve for the solution x0 when the first derivative of the polynomial function f(x) is equal to -1, and use it as the neighborhood radius parameter Eps of the DBSCAN algorithm. When there are two values ​​for the solution x0 when the first derivative of the polynomial function f(x) is equal to -1, the smaller value x0 is selected as the neighborhood radius parameter Eps of the DBSCAN algorithm.

[0078] S36: Determine the minimum number of neighborhood points MinPts for the DBSCAN algorithm based on the neighborhood radius parameter Eps of the DBSCAN algorithm;

[0079] Preferably, the minimum number of neighborhood points MinPts includes:

[0080]

[0081] Among them, P i The value of represents the number of data points contained within the neighborhood radius Eps of the i-th data point in the original wind power data subset. MinPts represents the minimum number of neighborhood points in the DBSCAN algorithm, and n represents the number of data points in the original wind power data subset.

[0082] S37: Cluster the data points in the original wind power data subset according to the neighborhood radius parameter Eps and the minimum number of neighborhood points MinPts of the DBSCAN algorithm, and divide the data points in the original wind power data subset into core points, boundary points and noise points.

[0083] Preferably, the clustering of data points in the original wind power data subset based on the neighborhood radius parameter Eps and the minimum number of neighborhood points MinPts of the DBSCAN algorithm includes:

[0084] Each data point in the original wind power data subset is treated as an object point. If the number of data points contained within the neighborhood radius Eps of the object point exceeds MinPts, then the object point is a core point. If the number of data points contained within the neighborhood radius Eps of the object point is less than MinPts, but the object point falls within the neighborhood radius of other core points, then the object point is a boundary point. Otherwise, the object point is a noise point.

[0085] This invention replaces traditional clustering methods, such as single k-means and single DBSCAN, with a segmented adaptive DBSCAN algorithm, which has poor recognition performance in areas near the cut-off wind speed and in areas with wind speeds greater than the rated wind speed. This overcomes the dependence of traditional clustering algorithms on the number of clusters or neighborhood parameters. At the same time, this invention adaptively sets the parameters of the DBSCAN clustering algorithm based on the data distribution of the partitioned original wind power data subset, reducing the influence of subjective factors in setting parameters manually, improving the stability of the recognition results, and overcoming the problem of poor recognition performance caused by different recognition results due to different parameters.

[0086] Another aspect of the present invention provides a wind power invalid data identification system based on adaptive DBSCAN clustering. The identification system is based on the wind power invalid data identification method based on adaptive DBSCAN clustering, and is characterized by comprising: a data acquisition module, a data preprocessing module, a K-means clustering module, and a DBSCAN clustering module.

[0087] The data acquisition module is used to acquire the original wind power dataset of the wind turbine under normal operating conditions.

[0088] The data preprocessing module is used to preprocess the original wind power dataset;

[0089] The K-means clustering module is used to divide the original wind power dataset into m subsets of original wind power data based on wind speed using the K-means algorithm;

[0090] The DBSCAN clustering module is used to cluster the m original wind power data subsets respectively, and output the clustered noise point data as invalid wind power data to obtain all invalid wind power data in the original wind power dataset.

[0091] Another aspect of the present invention provides an electronic device, including a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements a wind power invalid data identification method based on adaptive DBSCAN clustering.

[0092] In one embodiment, the electronic device may be a smart device terminal such as a server, mobile phone, or computer.

[0093] Another aspect of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements a wind power invalid data identification method based on adaptive DBSCAN clustering. Any references to memory, storage, database, or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0094] In this embodiment, a segmented adaptive DBSCAN algorithm is used to replace traditional clustering methods, such as single k-means and single DBSCAN, which have poor recognition performance in areas near the cut-off wind speed and in areas with wind speeds greater than the rated wind speed. This overcomes the dependence of traditional clustering algorithms on the number of clusters or neighborhood parameters. At the same time, this invention adaptively sets the parameters of the DBSCAN clustering algorithm based on the data distribution of the partitioned original wind power data subset, reducing the influence of subjective factors in setting parameters manually, improving the stability of the recognition results, and overcoming the problem of poor recognition performance caused by different recognition results due to different parameters.

[0095] In this embodiment, the effectiveness of the proposed invalid data identification method is verified using SCADA data collected from a wind farm in East China. This wind farm has 33 1.6MW doubly-fed induction generator (DFIG) wind turbines. The cut-in wind speed is 2 m / s, the rated wind speed is 10 m / s, and the cut-out wind speed is 20 m / s. The SCADA data sampling frequency is 1 s / s. This paper uses 10 minutes of SCADA data for analysis. First, k-means is used to divide the output power data in the SCADA system into clustering regions. For each region, segmented adaptive DBSCAN clustering is used to identify invalid data, integrating the segmented filtered bad data. Invalid power data mainly includes outliers, power-limited operation data, and deviation cluster data. The k-means-based segmented adaptive DBSCAN clustering method proposed in this paper, based on the distribution characteristics of the collected original power data, clusters the original power data into different classes in an unsupervised manner, effectively distinguishing between power-limited operation data under normal turbine operation conditions, outliers, deviation cluster data, and other invalid data, as well as normal data.

[0096] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for identifying invalid wind power data based on adaptive DBSCAN clustering, characterized in that, include: S1: Obtain the raw wind power dataset of the wind turbine under normal operating conditions and preprocess the raw wind power dataset; S2: Based on the wind speed, the original wind power dataset is divided into m subsets of original wind power data using the K-means algorithm; S3: Using the adaptive DBSCAN algorithm to respectively... Clustering is performed on a subset of the original wind power data, and the noise point data after clustering is output as invalid wind power data, resulting in all invalid wind power data in the original wind power dataset. The adaptive DBSCAN algorithm was used to respectively... Clustering of a subset of the original wind power data includes: S31: For each subset of original wind power data, calculate the distance between each pair of data points in the subset of original wind power data to construct the distance distribution matrix of the subset of original wind power data; S32: Arrange the elements in each row of the distance distribution matrix in ascending order to obtain the first intermediate distance distribution matrix; S33: The second intermediate distance distribution matrix is ​​obtained by arranging the elements in each column of the first intermediate distance distribution matrix in descending order from largest to smallest; S34: Plot each column element in the second intermediate distance distribution matrix as a broken line, and select the broken line that best represents the distribution of data points in the original wind power data subset as the target broken line; S35: Fit the target polynomial line using a polynomial function, and solve the polynomial function to obtain the neighborhood radius parameter Eps of the DBSCAN algorithm; S36: Determine the minimum number of neighborhood points for the DBSCAN algorithm based on the neighborhood radius parameter Eps. ; S37: Based on the neighborhood radius parameter Eps and the minimum number of neighborhood points in the DBSCAN algorithm. Clustering is performed on the data points in the original wind power data subset, dividing the data points in the original wind power data subset into core points, boundary points, and noise points.

2. The wind power invalid data identification method based on adaptive DBSCAN clustering according to claim 1, characterized in that, The process of dividing the original wind power dataset into m subsets using the K-means algorithm includes: S21: Use the elbow method to determine the number m of the original wind power data subsets, treat each original wind power data as a data point, and treat each original wind power data subset as a cluster; S22: Random selection Using 10 data points as the centroids, a loss function is defined based on the distance between each data point and the centroid of its cluster. : in, Indicates the first The set of all data points in a cluster Indicates the first The center point of each cluster, Indicates the distance from data point x to the center point. European distance, Represents the loss function; S23: Order For the number of iterations, repeat the following process until the loss function is reached. convergence; S24: For each data point, assign it to the cluster to which the nearest center point belongs; S25: For each cluster, recalculate the cluster center point; once the loss function converges, complete the partitioning of the original wind power data subset.

3. The wind power invalid data identification method based on adaptive DBSCAN clustering according to claim 1, characterized in that, The fitting of the target polynomial line using a polynomial function includes: Initialize polynomial function Substitute the data points on the target polynomial function. Solve for parameters , , and Given a polynomial function Solving polynomial functions under the given parameters Solution when the first derivative equals -1 As the neighborhood radius parameter Eps of the DBSCAN algorithm, where the polynomial function... Solution when the first derivative equals -1 If there are two values, choose the smaller value. Eps is the neighborhood radius parameter in the DBSCAN algorithm.

4. The wind power invalid data identification method based on adaptive DBSCAN clustering according to claim 1, characterized in that, The minimum number of neighborhood points include: in, This represents the first element in the original wind power data subset. The number of data points contained within the neighborhood radius Eps of a data point, where MinPts represents the minimum number of neighborhood points required by the DBSCAN algorithm. This indicates the number of data points in the original wind power data subset.

5. The wind power invalid data identification method based on adaptive DBSCAN clustering according to claim 1, characterized in that, The neighborhood radius parameter Eps and the minimum number of neighborhood points are based on the DBSCAN algorithm. Clustering of data points in a subset of the original wind power data includes: Each data point in the original wind power data subset is treated as an object point. If the number of data points contained within the neighborhood radius Eps of an object point exceeds a certain threshold, the object point is considered an object point. Then the object point is the core point; if the number of data points contained within the neighborhood radius Eps of the object point is less than However, if the object point falls within the radius of the other core points, then the object point is a boundary point; otherwise, the object point is a noise point.

6. A wind power invalid data identification system based on adaptive DBSCAN clustering, wherein the identification system is based on the wind power invalid data identification method based on adaptive DBSCAN clustering as described in any one of claims 1-5, characterized in that, include: Data acquisition module, data preprocessing module, K-means clustering module, and DBSCAN clustering module; The data acquisition module is used to acquire the original wind power dataset of the wind turbine under normal operating conditions. The data preprocessing module is used to preprocess the original wind power dataset; The K-means clustering module is used to divide the original wind power dataset into m subsets of original wind power data based on wind speed using the K-means algorithm; The DBSCAN clustering module is used to respectively... Clustering is performed on a subset of the original wind power data, and the noise point data after clustering is output as invalid wind power data, resulting in all invalid wind power data in the original wind power dataset.

7. An electronic device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, which, when executed by a processor, implements the method as described in any one of claims 1 to 5.

Images