Method and apparatus for identifying outlier of wind power curve, device and storage medium

Abstract

A method and apparatus for identifying an outlier of a wind power curve, a device, and a storage medium, which relate to the technical field of wind power data anomaly identification, and are used to solve the technical problem of relatively poor results identifying anomalous data in high-density clustered wind power data. The present invention comprises: acquiring wind power raw data; normalizing the wind power raw data, to obtain normalized data; generating a scatter plot by using the normalized data; identifying the scatter plot, to obtain dispersed noise points; removing the dispersed noise points from the scatter plot, to obtain preliminary denoised data; generating a fitting standard power curve of the preliminary denoised data; calculating a mean standard deviation of wind speed on the basis of the preliminary denoised data; and determining an outlier on the basis of the mean standard deviation of wind speed and the fitting standard power curve.

Description

Wind power curve abnormal value identification method, device, equipment and storage medium

[0001] This application claims priority to the Chinese patent application filed with the China Patent Office on January 24, 2024, with application number 202410098866.4 and invention name “Method, device, equipment and storage medium for identifying abnormal values ​​of wind power curve”, the entire contents of which are incorporated by reference into this application. Technical Field

[0002] The present invention relates to the technical field of wind power data anomaly identification, and in particular to a method, device, equipment and storage medium for identifying abnormal values ​​of a wind power curve. Background Art

[0003] Wind turbine operating data is a crucial component of power big data and forms the foundation for subsequent research on wind power fluctuation characteristics and wind power forecasting. However, due to factors such as wind curtailment, meter errors, and environmental climate, the data collected by wind turbine supervisory control and data acquisition (SCADA) systems contains numerous outliers. Directly using this data for analysis would significantly compromise the accuracy and reliability of subsequent forecasts. Therefore, identifying outliers in wind turbine measured data and obtaining reliable wind farm operating data is essential for evaluating wind farm operational status and forecasting wind speed and power.

[0004] Currently, outlier detection in unlabeled wind power datasets often relies on unsupervised learning methods, such as the Density-Based Spatial Clustering of Applications with Noise (DBSCAN). This algorithm divides regions of sufficient density into clusters and discovers clusters of arbitrary shapes in noisy spatial databases, defining a cluster as the largest set of densely connected points. Therefore, DBSCAN is effective for identifying dispersed outliers, but less effective for densely clustered outliers.

[0005] Summary of the Invention

[0006] The present invention provides a method, device, equipment and storage medium for identifying abnormal values ​​of wind power curves, which are used to solve the technical problem of poor abnormal data identification effect in high-density concentrated wind power data.

[0007] The present invention provides a method for identifying abnormal values ​​of a wind power curve, comprising:

[0008] Obtaining original wind power data;

[0009] Normalizing the original wind power data to obtain normalized data;

[0010] generating a scatter plot using the normalized data;

[0011] Identify the scatter plot to obtain dispersed noise points;

[0012] Remove the dispersed noise points from the scatter plot to obtain preliminary denoised data;

[0013] generating a fitted standard power curve for the preliminary denoised data;

[0014] Calculating the average standard deviation of wind speed based on the preliminary denoised data;

[0015] An outlier is determined according to the wind speed average standard deviation and the fitted standard power curve.

[0016] Optionally, the step of normalizing the original wind power data to obtain normalized data includes:

[0017] Obtaining maximum power data, minimum power data, maximum wind speed data, and minimum wind speed data from the original wind power data;

[0018] The maximum power data, the minimum power data, the maximum wind speed data, and the minimum wind speed data are used to perform normalization processing on the original wind power data to obtain normalized data.

[0019] Optionally, the step of identifying the scatter plot to obtain dispersed noise points includes:

[0020] Randomly select any unvisited data point in the scatter plot as a visited point;

[0021] Determining a first neighborhood of the visited point according to a preset radius, and obtaining a first point set of the first neighborhood;

[0022] Counting the number of data points in the first neighborhood;

[0023] When the number of data points is not less than a preset minimum data point threshold, generating a point cluster of the visited points;

[0024] Marking each data point in the first point set as a new visited point;

[0025] Determining a second neighborhood of the newly visited point according to a preset radius, and obtaining a second point set of the second neighborhood;

[0026] When the number of data points in the second neighborhood is not less than the preset minimum data point threshold, using the second point set to update the first point set;

[0027] Determine whether the new visited point has a member, if not, add the new visited point to the point cluster, if so, add the new visited point to a preset outlier point set;

[0028] When all the data in the first point set have been accessed, return to the step of randomly selecting any unvisited data point in the scatter plot as a visited point until all the data points in the scatter plot have been accessed, and output the data points in the outlier points as scattered noise points.

[0029] Optionally, the step of generating a fitting standard power curve for the preliminary denoised data includes:

[0030] Dividing the preliminary denoised data into a plurality of power intervals;

[0031] Calculating the wind speed mean and power mean of each power interval as the density midpoint;

[0032] Taking wind speed as the independent variable and power as the target variable, a cubic polynomial model of the power interval is established;

[0033] With the goal of minimizing the sum of absolute values ​​of power residuals, an optimization model is generated using the density midpoint and the parameters to be fitted of the cubic polynomial model;

[0034] Solving the optimization model by a gradient descent method to obtain parameter values ​​of the parameters to be fitted;

[0035] Substituting the parameter values ​​into the cubic polynomial model, a fitting standard power curve is generated.

[0036] The present invention also provides a device for identifying abnormal values ​​of a wind power curve, comprising:

[0037] Wind power raw data acquisition module, used to acquire wind power raw data;

[0038] A normalization module, configured to perform normalization processing on the original wind power data to obtain normalized data;

[0039] A scatter plot generating module, configured to generate a scatter plot using the normalized data;

[0040] A scattered noise point recognition module is used to recognize the scatter plot and obtain scattered noise points;

[0041] A denoising module is used to remove the dispersed noise points from the scatter plot to obtain preliminary denoised data;

[0042] A fitting standard power curve generating module, used for generating a fitting standard power curve of the preliminary denoised data;

[0043] A wind speed average standard deviation calculation module, configured to calculate the wind speed average standard deviation based on the preliminary denoised data;

[0044] An outlier determination module is used to determine an outlier based on the wind speed average standard deviation and the fitted standard power curve.

[0045] Optionally, the normalization module includes:

[0046] A reference data acquisition submodule is used to obtain the maximum power data, minimum power data, maximum wind speed data and minimum wind speed data from the original wind power data;

[0047] The normalization submodule is configured to perform normalization processing on the original wind power data using the maximum power data, the minimum power data, the maximum wind speed data, and the minimum wind speed data to obtain normalized data.

[0048] Optionally, the dispersed noise point identification module includes:

[0049] a visited point selection submodule, configured to randomly select any unvisited data point in the scatter plot as a visited point;

[0050] a first neighborhood and first point set determination submodule, configured to determine a first neighborhood of the visited point according to a preset radius, and obtain a first point set of the first neighborhood;

[0051] a data point counting submodule, configured to count the number of data points in the first neighborhood;

[0052] a point cluster generation submodule, configured to generate a point cluster of the visited points when the number of data points is not less than a preset minimum data point number threshold;

[0053] A new visited point marking submodule, configured to mark each data point in the first point set as a new visited point;

[0054] a second neighborhood and second point set determination submodule, configured to determine a second neighborhood of the newly visited point according to a preset radius, and obtain a second point set of the second neighborhood;

[0055] an updating submodule, configured to update the first point set with the second point set when the number of data points in the second neighborhood is not less than the preset minimum data point number threshold;

[0056] a judgment submodule, configured to judge whether the new visited point has a member, and if not, add the new visited point to the point cluster; and if so, add the new visited point to a preset outlier point set;

[0057] The output submodule is used to return to the step of randomly selecting any unvisited data point in the scatter plot as a visited point when all the data in the first point set have been accessed, until all the data points in the scatter plot have been accessed, and output the data points in the outlier points as scattered noise points.

[0058] Optionally, the fitting standard power curve generating module includes:

[0059] A power interval division submodule, configured to divide the preliminary denoised data into a plurality of power intervals;

[0060] A density midpoint calculation submodule, configured to calculate the wind speed mean and power mean of each power interval as the density midpoint;

[0061] A cubic polynomial model establishment submodule is used to establish a cubic polynomial model of the power range with wind speed as an independent variable and power as a target variable;

[0062] An optimization model generation submodule, configured to generate an optimization model by using the density midpoint and the parameters to be fitted of the cubic polynomial model with the goal of minimizing the sum of the absolute values ​​of the power residuals;

[0063] A solution submodule, configured to solve the optimization model by a gradient descent method to obtain parameter values ​​of the parameters to be fitted;

[0064] The fitting standard power curve generation submodule is used to substitute the parameter values ​​into the cubic polynomial model to generate a fitting standard power curve.

[0065] The present invention further provides an electronic device, comprising a processor and a memory:

[0066] The memory is used to store program code and transmit the program code to the processor;

[0067] The processor is configured to execute any one of the above methods for identifying abnormal values ​​of a wind power curve according to instructions in the program code.

[0068] The present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium is used to store program code, and the program code is used to execute the method for identifying abnormal values ​​of a wind power curve as described in any one of the above items.

[0069] From the above technical solution, it can be seen that the present invention has the following advantages: the present invention discloses a method for identifying outliers in a wind power curve, and specifically discloses: obtaining original wind power data; normalizing the original wind power data to obtain normalized data; generating a scatter plot using the normalized data; identifying the scatter plot to obtain dispersed noise points; removing the dispersed noise points from the scatter plot to obtain preliminary denoised data; generating a fitted standard power curve for the preliminary denoised data; calculating the average standard deviation of wind speed based on the preliminary denoised data; and determining outliers based on the average standard deviation of wind speed and the fitted standard power curve. The present invention normalizes the original wind power data to eliminate the influence between the different dimensions of wind speed and power, then filters out dispersed outliers in the power curve, and then models the standard power curve for the data after the dispersed outliers are filtered out. The method combines the average standard deviation of wind speed to identify outliers in dense wind power data, thereby realizing the identification of abnormal data in high-density concentrated wind power data and improving the accuracy of wind power data anomaly identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0071] FIG1 is a flowchart of a method for identifying abnormal values ​​of a wind power curve provided by an embodiment of the present invention;

[0072] FIG2 is a flowchart of a method for identifying abnormal values ​​of a wind power curve according to another embodiment of the present invention;

[0073] FIG3 is a schematic diagram of a scatter plot provided by an embodiment of the present invention;

[0074] Figure 4 shows the image of preliminary denoised data after preliminary processing by the DBSCAN algorithm;

[0075] FIG5 is a schematic diagram of the standard power curve after fitting;

[0076] FIG6 is a schematic diagram of outlier classification after processing using the least absolute residual method;

[0077] FIG7 is a structural block diagram of a device for identifying abnormal values ​​of a wind power curve provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0078] The embodiments of the present invention provide a method, apparatus, device and storage medium for identifying abnormal values ​​of a wind power curve, which are used to solve the technical problem of poor abnormal data identification effect in high-density concentrated wind power data.

[0079] In order to make the purpose, features, and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0080] Please refer to FIG1 , which is a flowchart of a method for identifying abnormal values ​​of a wind power curve provided by an embodiment of the present invention.

[0081] The present invention provides a method for identifying abnormal values ​​of a wind power curve, which may specifically include the following steps:

[0082] Step 101: Obtaining original wind power data;

[0083] The original data of wind power includes wind speed, power and other data.

[0084] In the embodiment of the present invention, the original wind power data can be collected by a wind turbine supervisory control and data acquisition system (SCADA).

[0085] Step 102: normalize the original wind power data to obtain normalized data;

[0086] Normalization is a way to simplify calculations, that is, to transform a dimensioned expression into a dimensionless expression, which becomes a scalar.

[0087] After collecting the original wind power data, it can be normalized to eliminate the impact of different dimensions of wind speed and power.

[0088] Step 103, generating a scatter plot using the normalized data;

[0089] After the normalization of the original wind power data is completed, a scatter plot can be generated using the normalized data.

[0090] A scatter plot is an image generated by plotting normalized data on a coordinate graph with wind speed as the horizontal axis and power as the vertical axis.

[0091] Step 104: Identify the scatter plot to obtain dispersed noise points;

[0092] Step 105: removing scattered noise points from the scatter plot to obtain preliminary denoised data;

[0093] In practical applications, due to factors such as wind power curtailment, meter errors, and environmental climate, the data actually collected by the wind turbine supervisory control and data acquisition system (SCADA) contains a large number of outliers, resulting in some scattered noise points in the scatter plot. Therefore, it is necessary to remove the scattered noise points in the scatter plot.

[0094] In one example, DBSCAN (Density-Based Spatial Clustering of Applications with Noise) can be used to identify dispersed noise points.

[0095] DBSCAN is a density-based clustering algorithm that analyzes the density relationship between data points and clusters data points with higher density together to form clusters.

[0096] Step 106, generating a fitted standard power curve of the preliminary denoised data;

[0097] After removing the noise points, a fitted standard power curve of the preliminary denoised data can be generated to determine the outliers in the preliminary denoised data based on the fitted standard power curve.

[0098] Step 107, calculating the average standard deviation of wind speed based on the preliminary denoised data;

[0099] Step 108: determining an outlier based on the wind speed average standard deviation and the fitted standard power curve.

[0100] After generating the fitted standard power curve, the outliers of the fitted standard power curve can be determined by calculating the average standard deviation of the wind speed of the preliminary denoised data.

[0101] The present invention normalizes the original wind power data to eliminate the influence of different dimensions of wind speed and power, then filters out the dispersed outliers in the power curve, and then performs standard power curve modeling on the data after the dispersed outliers are filtered out. In addition, the average standard deviation of wind speed is used to identify outliers in dense wind power data, thereby realizing the identification of abnormal data in high-density concentrated wind power data and improving the accuracy of wind power data anomaly identification.

[0102] Please refer to FIG2 , which is a flowchart of a method for identifying abnormal values ​​of a wind power curve according to another embodiment of the present invention. Specifically, the method may include the following steps:

[0103] Step 201: Obtaining original wind power data;

[0104] Step 202: Acquire maximum power data, minimum power data, maximum wind speed data, and minimum wind speed data from the original wind power data;

[0105] Step 203 , normalizing the original wind power data using the maximum power data, the minimum power data, the maximum wind speed data, and the minimum wind speed data to obtain normalized data;

[0106] Step 204, generating a scatter plot using the normalized data;

[0107] In an embodiment of the present invention, the maximum power data, minimum power data, maximum wind speed data and minimum wind speed data in the original wind power data may be obtained to perform normalization processing on the original wind power data to obtain a scatter plot as shown in FIG3 .

[0108] Step 205: Identify the scatter plot to obtain dispersed noise points;

[0109] In practical applications, due to factors such as wind power curtailment, meter errors, and environmental climate, the data actually collected by the wind turbine supervisory control and data acquisition system (SCADA) contains a large number of outliers, resulting in some scattered noise points in the scatter plot. Therefore, it is necessary to remove the scattered noise points in the scatter plot.

[0110] In one example, step 205 may include the following sub-steps:

[0111] S51, randomly select any unvisited data point in the scatter plot as a visited point;

[0112] S52, determining a first neighborhood of the visited point according to a preset radius, and obtaining a first point set of the first neighborhood;

[0113] S53, counting the number of data points in the first neighborhood;

[0114] S54, when the number of data points is not less than a preset minimum data point threshold, generating a point cluster of visited points;

[0115] S55, marking each data point in the first point set as a new visited point;

[0116] S56, determining a second neighborhood of the newly visited point according to a preset radius, and obtaining a second point set of the second neighborhood;

[0117] S57, when the number of data points in the second neighborhood is not less than a preset minimum data point threshold, using the second point set to update the first point set;

[0118] S58, determining whether the new visited point has a member, if not, adding the new visited point to the point cluster, if so, adding the new visited point to the preset outlier point set;

[0119] S59, when all the data in the first point set have been visited, return to the step of randomly selecting any unvisited data point in the scatter plot as a visited point, until all the data points in the scatter plot have been visited, and output the data points in the outlier points as scattered noise points.

[0120] In the specific implementation, you can first set the neighborhood radius ε and the minimum number of points MinPts. ε is a distance threshold used to define the neighborhood of a data point; MinPts is the preset minimum data point threshold, that is, the minimum number of data points required in the neighborhood.

[0121] Then all points in the scatter plot are set as unvisited points, and any unvisited data point p in the scatter plot is randomly selected as a visited point. Set the first neighborhood within the p given neighborhoods and obtain the first point set N k as follows:

[0122] If there are at least MinPts data points in the neighborhood with radius ε and centered at p, create a point cluster C k : C k ={p};

[0123] For N k For each data point q in , after it is marked as visited, if there are at least MinPts data points in the neighborhood with radius ε centered on q, then the first point set N is updated k :

[0124] If q does not belong to any cluster, add it to C k , otherwise, mark q as an outlier and add it to the outlier set O.

[0125] Step 206, removing scattered noise points from the scatter plot to obtain preliminary denoised data;

[0126] By filtering out the scattered noise points in the outlier set, preliminary denoised data can be obtained.

[0127] In addition, after filtering out the scattered noise points in the outlier set, data points of some clusters can be filtered out according to specific circumstances, and the preliminary denoised data display image shown in Figure 4 is obtained.

[0128] Step 207, generating a fitted standard power curve of the preliminary denoised data;

[0129] It should be noted that the selection of parameters in the DBSCAN algorithm, such as ε and MinPts, has a great influence on the clustering results. Therefore, when there are many clustered outliers in the original data set, it is necessary to improve the clustering effect by appropriately adjusting the parameters. However, this method will bring more negative effects, so further data cleaning is required on this basis.

[0130] In one example, the least absolute residual (LAR) method can be used to perform further data cleaning. Therefore, step 207 may include the following sub-steps:

[0131] S71, dividing the preliminary denoised data into a plurality of power intervals;

[0132] S72, calculating the wind speed mean and power mean of each power interval as the density midpoint;

[0133] S73, using wind speed as the independent variable and power as the target variable, a cubic polynomial model of power interval is established;

[0134] S74, with the goal of minimizing the sum of absolute values ​​of power residuals, generating an optimization model using the parameters to be fitted of the density midpoint and cubic polynomial models;

[0135] S75, solving the optimization model by a gradient descent method to obtain parameter values ​​of the parameters to be fitted;

[0136] S76: Substitute the parameter values ​​into the cubic polynomial model to generate a fitted standard power curve.

[0137] In the specific implementation, the preliminary denoised data obtained after DBSCAN preprocessing can be first divided into several power intervals. Then, the mean wind speed and mean power of each power interval are calculated as the density midpoint of the power interval.

[0138] Then a cubic polynomial model is established, with wind speed v as the independent variable and power P as the target variable:

[0139] in These are parameters to be fitted.

[0140] Define the objective function as the sum of the absolute values ​​of the residuals:

[0141] Where n is the number of density midpoints, (v i ,P i ) is the midpoint of each density.

[0142] The L1 norm penalty term is introduced into the objective function, so that the optimization objective is:

[0143] Among them, λ is the coefficient of the penalty term, which is used to balance the fitting accuracy and model complexity.

[0144] Use the gradient descent method to update the parameters. Since the absolute value function derivative is not unique at zero, the sub-gradient concept is introduced:

[0145] Where α is the learning rate, is the objective function with respect to β j For the absolute value function, its subgradient is -1 or 1, so the objective function is about β j The subgradient of can be defined as:

[0146] where sgn(·) is the sign function.

[0147] Repeatedly use the gradient descent method to update the parameters until the parameters converge and obtain the parameters Substituting the cubic polynomial model, the fitted standard power curve is shown in Figure 5.

[0148] Step 208, calculating the average standard deviation of wind speed based on the preliminary denoised data;

[0149] The calculation formula for the average standard deviation of wind speed σ is as follows:

[0150] Where m is the total number of power intervals, σ i is the standard deviation of wind speed in each power interval, and σ is the average standard deviation of wind speed.

[0151] Step 209: determining an outlier based on the wind speed average standard deviation and the fitted standard power curve.

[0152] In a specific implementation, the normal power range can be set as the 3σ distance around the standard power curve, and the data points outside this power range can be set as abnormal values. The result is shown in Figure 6.

[0153] Please refer to FIG. 7 , which is a structural block diagram of a device for identifying abnormal values ​​of a wind power curve provided by an embodiment of the present invention.

[0154] An embodiment of the present invention provides a device for identifying abnormal values ​​of a wind power curve, comprising:

[0155] Wind power raw data acquisition module 701, used to acquire wind power raw data;

[0156] Normalization module 702, used to perform normalization processing on the original wind power data to obtain normalized data;

[0157] A scatter plot generating module 703 is used to generate a scatter plot using normalized data;

[0158] The scattered noise point identification module 704 is used to identify the scatter plot and obtain scattered noise points;

[0159] De-noising module 705, used to remove scattered noise points from the scatter plot to obtain preliminary de-noised data;

[0160] A fitting standard power curve generating module 706 is used to generate a fitting standard power curve of the preliminary denoised data;

[0161] The wind speed average standard deviation calculation module 707 is used to calculate the wind speed average standard deviation based on the preliminary denoised data;

[0162] The outlier determination module 708 is configured to determine an outlier based on the wind speed average standard deviation and the fitted standard power curve.

[0163] In this embodiment of the present invention, the normalization module 702 includes:

[0164] The reference data acquisition submodule is used to obtain the maximum power data, minimum power data, maximum wind speed data and minimum wind speed data from the original wind power data;

[0165] The normalization submodule is used to normalize the original wind power data using the maximum power data, the minimum power data, the maximum wind speed data and the minimum wind speed data to obtain normalized data.

[0166] In this embodiment of the present invention, the scattered noise point identification module 704 includes:

[0167] The visited point selection submodule is used to randomly select any unvisited data point in the scatter plot as a visited point;

[0168] A first neighborhood and first point set determination submodule, configured to determine a first neighborhood of the visited point according to a preset radius and obtain a first point set of the first neighborhood;

[0169] A data point counting submodule is used to count the data points in the first neighborhood;

[0170] The point cluster generation submodule is used to generate point clusters of visited points when the number of data points is not less than a preset minimum data point threshold;

[0171] A new visited point marking submodule, for marking each data point in the first point set as a new visited point;

[0172] A second neighborhood and second point set determination submodule, configured to determine a second neighborhood of the newly visited point according to a preset radius and obtain a second point set of the second neighborhood;

[0173] An updating submodule, configured to update the first point set with the second point set when the number of data points in the second neighborhood is not less than a preset minimum data point threshold;

[0174] A judgment submodule is used to judge whether the new visited point has a home. If not, the new visited point is added to the point cluster. If so, the new visited point is added to the preset outlier point set.

[0175] The output submodule is used to return to the step of randomly selecting any unvisited data point in the scatter plot as a visited point when all the data in the first point set have been visited, until all the data points in the scatter plot have been visited, and output the data points in the outliers as scattered noise points.

[0176] In this embodiment of the present invention, the fitting standard power curve generating module 706 includes:

[0177] The power interval division submodule is used to divide the preliminary denoised data into several power intervals;

[0178] The density midpoint calculation submodule is used to calculate the mean wind speed and mean power of each power interval as the density midpoint;

[0179] The cubic polynomial model building submodule is used to build a cubic polynomial model of power range with wind speed as the independent variable and power as the target variable;

[0180] An optimization model generation submodule is used to generate an optimization model using the density midpoint and the parameters to be fitted of the cubic polynomial model with the goal of minimizing the sum of the absolute values ​​of the power residuals;

[0181] The solution submodule is used to solve the optimization model through the gradient descent method to obtain the parameter values ​​of the parameters to be fitted;

[0182] The fitting standard power curve generation submodule is used to substitute parameter values ​​into the cubic polynomial model to generate a fitting standard power curve.

[0183] An embodiment of the present invention further provides an electronic device, the device including a processor and a memory:

[0184] The memory is used to store program codes and transmit the program codes to the processor;

[0185] The processor is configured to execute the wind power curve abnormal value identification method according to the instructions in the program code.

[0186] An embodiment of the present invention further provides a computer-readable storage medium, which is used to store program code, and the program code is used to execute the wind power curve abnormal value identification method of the embodiment of the present invention.

[0187] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0188] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0189] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, apparatus, or computer program products. Thus, embodiments of the present invention may take the form of a fully hardware embodiment, a fully software embodiment, or an embodiment combining software and hardware. Furthermore, embodiments of the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0190] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate a device for implementing the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0191] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device that implements the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0192] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce computer-implemented processing, so that the instructions executed on the computer or other programmable terminal device provide steps for implementing the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0193] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they become aware of the basic creative concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0194] Finally, it should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or terminal device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or terminal device that includes the element.

[0195] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying outliers in a wind power curve, characterized in that, include: Obtaining original wind power data; Normalizing the original wind power data to obtain normalized data; generating a scatter plot using the normalized data; Identify the scatter plot to obtain dispersed noise points; Remove the dispersed noise points from the scatter plot to obtain preliminary denoised data; generating a fitted standard power curve for the preliminary denoised data; Calculating the average standard deviation of wind speed based on the preliminary denoised data; An outlier is determined according to the wind speed average standard deviation and the fitted standard power curve.

2. The method according to claim 1, characterized in that, The step of normalizing the original wind power data to obtain normalized data includes: Obtaining maximum power data, minimum power data, maximum wind speed data, and minimum wind speed data from the original wind power data; The maximum power data, the minimum power data, the maximum wind speed data, and the minimum wind speed data are used to perform normalization processing on the original wind power data to obtain normalized data.

3. The method according to claim 1, characterized in that, The step of identifying the scatter plot to obtain dispersed noise points includes: Randomly select any unvisited data point in the scatter plot as a visited point; Determining a first neighborhood of the visited point according to a preset radius, and obtaining a first point set of the first neighborhood; Counting the number of data points in the first neighborhood; When the number of data points is not less than a preset minimum data point threshold, generating a point cluster of the visited points; Marking each data point in the first point set as a new visited point; Determining a second neighborhood of the newly visited point according to a preset radius, and obtaining a second point set of the second neighborhood; When the number of data points in the second neighborhood is not less than the preset minimum data point threshold, Updating the first point set using the second point set; Determine whether the new visited point has a member, if not, add the new visited point to the point cluster, if so, add the new visited point to a preset outlier point set; When all the data in the first point set have been accessed, return to the step of randomly selecting any unvisited data point in the scatter plot as a visited point until all the data points in the scatter plot have been accessed, and output the data points in the outlier points as scattered noise points.

4. The method according to claim 1, characterized in that, The step of generating a fitting standard power curve for the preliminary denoised data comprises: Dividing the preliminary denoised data into a plurality of power intervals; Calculating the wind speed mean and power mean of each power interval as the density midpoint; Taking wind speed as the independent variable and power as the target variable, a cubic polynomial model of the power interval is established; With the goal of minimizing the sum of absolute values of power residuals, an optimization model is generated using the density midpoint and the parameters to be fitted of the cubic polynomial model; Solving the optimization model by a gradient descent method to obtain parameter values of the parameters to be fitted; Substituting the parameter values into the cubic polynomial model, a fitting standard power curve is generated.

5. A device for identifying outliers in a wind power curve, characterized in that, include: Wind power raw data acquisition module, used to acquire wind power raw data; A normalization module, configured to perform normalization processing on the original wind power data to obtain normalized data; A scatter plot generating module, configured to generate a scatter plot using the normalized data; A scattered noise point recognition module is used to recognize the scatter plot and obtain scattered noise points; A denoising module is used to remove the dispersed noise points from the scatter plot to obtain preliminary denoised data; A fitting standard power curve generating module, used for generating a fitting standard power curve of the preliminary denoised data; A wind speed average standard deviation calculation module, configured to calculate the wind speed average standard deviation based on the preliminary denoised data; An outlier determination module is used to determine the outlier value based on the average standard deviation of the wind speed and the fitted standard power. The curve identifies outliers.

6. The device according to claim 5, characterized in that, The normalization module includes: A reference data acquisition submodule is used to obtain the maximum power data, minimum power data, maximum wind speed data and minimum wind speed data from the original wind power data; The normalization submodule is configured to perform normalization processing on the original wind power data using the maximum power data, the minimum power data, the maximum wind speed data, and the minimum wind speed data to obtain normalized data.

7. The device according to claim 5, characterized in that, The dispersed noise point identification module includes: a visited point selection submodule, configured to randomly select any unvisited data point in the scatter plot as a visited point; a first neighborhood and first point set determination submodule, configured to determine a first neighborhood of the visited point according to a preset radius, and obtain a first point set of the first neighborhood; a data point counting submodule, configured to count the number of data points in the first neighborhood; a point cluster generation submodule, configured to generate a point cluster of the visited points when the number of data points is not less than a preset minimum data point number threshold; A new visited point marking submodule, configured to mark each data point in the first point set as a new visited point; a second neighborhood and second point set determination submodule, configured to determine a second neighborhood of the newly visited point according to a preset radius, and obtain a second point set of the second neighborhood; an updating submodule, configured to update the first point set with the second point set when the number of data points in the second neighborhood is not less than the preset minimum data point number threshold; a judgment submodule, configured to judge whether the new visited point has a member, and if not, add the new visited point to the point cluster; and if so, add the new visited point to a preset outlier point set; The output submodule is used to return to the step of randomly selecting any unvisited data point in the scatter plot as a visited point when all the data in the first point set have been accessed, until all the data points in the scatter plot have been accessed, and output the data points in the outlier points as scattered noise points.

8. The device according to claim 5, characterized in that, The fitting standard power curve generation module includes: A power interval division submodule, configured to divide the preliminary denoised data into a plurality of power intervals; A density midpoint calculation submodule, configured to calculate the wind speed mean and power mean of each power interval as the density midpoint; A cubic polynomial model establishment submodule is used to establish a cubic polynomial model of the power range with wind speed as an independent variable and power as a target variable; An optimization model generation submodule, configured to generate an optimization model by using the density midpoint and the parameters to be fitted of the cubic polynomial model with the goal of minimizing the sum of the absolute values of the power residuals; A solution sub-module, configured to solve the optimization model by the gradient descent method to obtain the parameter values of the parameters to be fitted; A fitting standard power curve generation sub-module, configured to substitute the parameter values into the cubic polynomial model to generate a fitting standard power curve.

9. An electronic device, characterized in that, The device includes a processor and a memory: The memory is configured to store program codes and transmit the program codes to the processor; The processor is configured to execute the wind power curve outlier recognition method according to any one of claims 1-4 based on the instructions in the program codes.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium is configured to store program codes, and the program codes are used to execute the wind power curve outlier recognition method according to any one of claims 1-4.

Images