CN114997256A - Method, device and storage medium for detecting abnormal power of wind farm - Google Patents

Abstract

The invention relates to a method, equipment and a storage medium for detecting abnormal power of a wind power plant, wherein the method comprises the following steps: s1, acquiring a power and wind speed scattered point data set D of the wind power plant; s2, identifying discrete abnormal data in the power and wind speed scattered data set by adopting a local abnormal factor LOF algorithm to obtain an intermediate data set D'; and step S3, identifying the stacking abnormal data in the intermediate data set D' by adopting an isolated forest algorithm IF. Compared with the prior art, the method has the advantages of high abnormal data detection accuracy and high detection efficiency.

Description

Method, device and storage medium for detecting abnormal power of wind farm

technical field

The invention relates to the field of wind farm control, in particular to a method, equipment and storage medium for detecting abnormal power of a wind farm.

Background technique

In the actual operation of the wind farm, abnormal data will appear due to the influence of the bad operating environment, the abnormal load shedding of the units, and the failure of the measurement and communication equipment. The basis for predicting the output of wind farms, the equivalent output curve of wind farms based on historical data is also often used to evaluate the operating status of wind farms. Therefore, it is very important to effectively identify these abnormal wind power data.

At present, there are mainly the following methods for the detection of abnormal data: 1) Anomaly detection based on the probability and statistical model, this method judges whether the data is abnormal by calculating the deviation between the standard data and the actual data, but the selection of the standard data set has certain blindness 2) Anomaly data detection based on clustering, this method divides the samples with unknown label information into several subsets according to certain rules and requirements, and divides the samples that belong to a subset with a degree lower than the threshold. Points are regarded as abnormal data; 3) abnormal data detection based on density and distance, this method calculates the number of sample points in a certain unit space in multi-dimensional space, this method does not require a large amount of training data.

The above density-based anomaly detection algorithm mainly displays outlier samples in different density datasets intuitively and concisely by calculating the anomaly level of each sample point in the dataset. However, for a given data set, it is necessary to obtain the Euclidean distance between each data object and all other points, as well as the local reachable distance of the data point, which requires a large amount of calculation. When the scale of the data set is large, the algorithm operating efficiency is low.

In view of the above situation, it is urgent to design a method for detecting abnormal data of wind farms with high accuracy and high efficiency.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a detection method, device and storage medium for abnormal power of a wind farm with high accuracy and high detection efficiency in order to overcome the above-mentioned defects of the prior art.

The object of the present invention can be realized through the following technical solutions:

The invention provides a method for detecting abnormal power of a wind farm, the method comprising the following steps:

Step S1, obtaining the power and wind speed scatter data set D of the wind farm;

Step S2, using the local abnormal factor LOF algorithm to identify discrete abnormal data in the power and wind speed scatter data set, and obtain an intermediate data set D';

In step S3, the isolated forest algorithm IF is used to identify the stacked abnormal data in the intermediate data set D'.

Preferably, the data of the power and wind speed scatter data set of the wind farm in the step S1 includes scattered data and stacked data.

Preferably, the step S2 is specifically: calculating the local abnormality factor LOF corresponding to each data point in the data set, treating the data points whose local abnormality factor LOF value is higher than the LOF threshold as abnormal data, and removing the abnormal data.

Preferably, the local outlier factor LOF is the average value of the ratio of the local reachable density of the data points in the k-distance neighborhood to the local reachable density of the data points.

Preferably, the step S3 is specifically:

1) Randomly sample the intermediate data set D' to construct an isolated binary tree and construct an isolated forest model;

2) For any sample point in each isolated binary tree, traverse the isolated binary tree from the root node until reaching the leaf node, calculate the average path length c(p) of the sample point, and perform anomaly score for the sample based on the anomaly index function;

3) The measured points whose abnormal score value is higher than the score threshold are regarded as abnormal points and removed.

Preferably, the expression of the abnormal index function is:

In the formula, h _x is the path length of the sample point x from the leaf node to the root node, E(h _x ) is the mean path length of the sample point x in all isolated binary trees iTree; c(p) is the isolated sample data constructed by p. The average path length of a binary tree.

Preferably, the expression of the path length h _x of the sample point from the leaf node to the root node is:

Preferably, the expression of the average path length c(p) is:

In the formula, H _p-1 =ln(p-1)+ξ, ξ is Euler's constant, and p is the number of samples.

According to a second aspect of the present invention, an electronic device is provided, comprising a memory and a processor, the memory stores a computer program, and the processor implements any one of the methods when executing the program.

According to a third aspect of the present invention, there is provided a computer-readable storage medium on which a computer program is stored, and when the program is executed by a processor, any one of the methods is implemented.

Compared with the prior art, the present invention has the following advantages:

1) The present invention overcomes the defect that the existing isolated forest algorithm IF algorithm is difficult to identify discrete abnormal data points and the local abnormal factor LOF algorithm has a poor identification effect for large data volume and stacked abnormal point clusters, and adopts the LOF algorithm. The detection algorithm combined with the IF algorithm classifies and detects abnormal data with different characteristics of wind farm power and wind speed scattered points, which improves the detection accuracy and detection efficiency of abnormal data;

2) The isolated forest algorithm adopted in the present invention realizes multiple segmentation of the data set through the isolated binary tree, thereby accurately identifying abnormal data stacked together with high distribution density.

Description of drawings

Fig. 1 is the method flow chart of the present invention;

Figure 2 is a schematic diagram of the reachable distance in the LOF algorithm;

Figure 3 shows the distribution of abnormal data of wind farms;

Fig. 4 is the abnormal data detection result graph based on LOF algorithm;

FIG. 5 is a graph of the abnormal data detection result based on the IF algorithm.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of the present invention.

Example

As shown in FIG. 1 , this embodiment provides a method for detecting abnormal power of a wind farm, and the method includes the following steps:

Step S1, obtaining the power and wind speed scattered data set D of the wind farm, including scattered data and stacked data;

Step S2, using the local abnormal factor LOF algorithm to identify discrete abnormal data in the power and wind speed scatter data set, and obtain an intermediate data set D', specifically:

Calculate the local abnormal factor LOF corresponding to each data point in the data set, regard the data points with the local abnormal factor LOF value higher than the LOF threshold as abnormal data, and remove the abnormal data; among them, the local abnormal factor LOF is the data in the k-distance neighborhood The average of the ratio of the local reachability density of a point to the local reachability density of the data point.

Step S3, using the isolated forest algorithm IF to identify the stacked abnormal data in the intermediate data set D', including the following sub-steps:

1) Randomly sample the intermediate data set D' to construct an isolated binary tree and construct an isolated forest model;

2) For any sample point in each isolated binary tree, traverse the isolated binary tree from the root node until reaching the leaf node, calculate the average path length c(p) of the sample point, and perform anomaly score for the sample based on the anomaly index function;

The expression of the abnormal indicator function is:

In the formula, h _x is the path length of the sample point x from the leaf node to the root node, E(h _x ) is the mean path length of the sample point x in all isolated binary trees iTree; c(p) is the isolated sample data constructed by p. The average path length of a binary tree.

The expression of the path length h _x of the sample point from the leaf node to the root node is:

The expression of the average path length c(p) is:

In the formula, H _p-1 =ln(p-1)+ξ, ξ is Euler's constant, and p is the number of samples.

3) The measured points whose abnormal score value is higher than the score threshold are regarded as abnormal points and removed.

Next, the local abnormal factor algorithm LOF and the isolated forest algorithm IF adopted in the present invention are described in detail.

1. The anomaly detection method based on the local outlier factor (LOF) is a typical density-based outlier detection method; wherein, the local outlier factor is used to characterize the degree of abnormality of the data.

Definition 1: k distance, among the points closest to the data point, the distance between the kth closest point and the data point;

Definition 2: k-distance neighborhood, all data sets whose distance from the data point is less than or equal to the k distance of the data point, that is, the set of points in the area with the data point as the center and the k-distance as the radius;

Definition 3: Reachable distance. If the distance between data p and o is relatively close, then the reachable distance from data o to data p is the k-distance of o. When the two are far apart, the reachable distance is regarded as the actual distance between them. Distance d(p,o), as shown in Figure 2;

Definition 4: local reachability density, the local reachability density is the reciprocal of the average reachability density value of the data point p relative to the objects in its k-distance neighborhood;

Definition 5: Local Outlier Factor, for a given k and data point p, the local outlier factor is defined as the average of the ratio of the local reachable density of data points in the k-distance neighborhood to the local reachable density of p. The local anomaly factor directly represents the possibility that a data point is an anomaly. The LOF value is equal to 1 or less than 1, indicating that the local reachability density of the data point is close to the local reachability density of other points in the neighborhood, and the point is an isolated point. The probability of is lower, the larger the LOF value, the higher the degree of isolation, and the more likely it is an abnormal point.

2. Isolated Forest (IF) is an anomaly detection algorithm based on ensemble learning tree model (TreeEmbedding). In the IF algorithm, the distribution is sparse, and the points that are far away from the densely distributed point area are defined as outliers. The basic principle of the IF algorithm is: use a random hyperplane to divide the data space, and use recursive random numbers to divide the data subspace again. After repeated cutting, we know that each subspace has only one sample point left, that is, all data points are Isolation" to stop cutting. It can be seen that the data clusters with high distribution density will be "isolated" after being segmented many times, while the sample points with sparse distribution can be segmented only a few times.

The abnormal data detection process of the algorithm is as follows:

Define the measured point x to traverse each isolated binary tree iTree from the root node node until the number of edges traversed when it encounters an external node is the path length h(x), calculate the average path length of the measured point, and perform anomaly score, The value range of the abnormal score s is (0, 1), and the settings for the abnormal score value s are as follows:

1) When s is close to 1, the smaller the branch of the data point in iTree, the higher the possibility that the data is judged to be abnormal;

2) When s is close to 0, the larger the branch of the data point in iTree, the less likely the data is judged to be abnormal;

3) When s is close to 0.5, the abnormal characteristics of the data points are not obvious, and it is impossible to judge whether it is abnormal or not.

In order to verify the effect of the abnormal data detection method proposed by the present invention, this embodiment is aimed at a wind farm with a rated capacity of 16MW. Group data for anomaly detection.

The algorithm parameters are set as follows: in the LOF algorithm, k is set to 10, and the LOF threshold is set to 1.5; in the IF algorithm, the number of trees, that is, the integration scale, is set to 100, and the number of branches, that is, the sampling scale of each tree is set to 256.

The experimental results shown in Figures 3 to 5 prove that the LOF algorithm has a better identification effect on the scattered T3 abnormal data around the curve, because the other two types of abnormal data show a phenomenon of local accumulation and dense distribution of a large amount of data in a certain area. , the LOF algorithm has a poor recognition effect, and the abnormal data is seriously underreported. For the points with the largest wind speed, the power is not abnormal. However, due to the small amount of data and sparseness, the LOF algorithm falsely reported them as abnormal points. . The IF algorithm uses the tree model to divide the data set multiple times to separate the data points, and the degree of isolation of abnormal data in the tree is relatively obvious. From Figure 5, it is found that the IF algorithm has an obvious effect on the identification of densely distributed abnormal data, but there are also In this way, some normal data at the edge of the data cluster are judged as false positives of abnormal data.

The wind power abnormal data detection method combined with the LOF algorithm and the IF algorithm adopted by the present invention can better identify different types of abnormal data of the wind farm.

The electronic device of the present invention includes a central processing unit (CPU) that can execute various computer program instructions stored in a read only memory (ROM) or loaded into a random access memory (RAM) from a storage unit according to computer program instructions Appropriate action and handling. In RAM, various programs and data required for device operation can also be stored. The CPU, ROM, and RAM are connected to each other through a bus. Input/output (I/O) interfaces are also connected to the bus.

Various components in the device are connected to the I/O interface, including: input units, such as keyboards, mice, etc.; output units, such as various types of displays, speakers, etc.; storage units, such as magnetic disks, optical disks, etc.; and communication units, For example, network cards, modems, wireless communication transceivers, etc. The communication unit allows the device to exchange information/data with other devices through a computer network such as the Internet and/or various telecommunication networks.

The processing unit executes the various methods and processes described above, eg, methods S1 to S3. For example, in some embodiments, the methods S1-S3 may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and/or installed on the device via the ROM and/or the communication unit. When the computer program is loaded into the RAM and executed by the CPU, one or more steps of the methods S1 to S3 described above may be performed. Alternatively, in other embodiments, the CPU may be configured to perform the methods S1-S3 in any other suitable manner (eg, by means of firmware).

The functions described herein above may be performed, at least in part, by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: Field Programmable Gate Arrays (FPGAs), Application Specific Integrated Circuits (ASICs), Application Specific Standard Products (ASSPs), System on Chips (SOCs), Load Programmable Logic Device (CPLD) and so on.

Program code for implementing the methods of the present invention may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general purpose computer, special purpose computer or other programmable data processing apparatus, such that the program code, when executed by the processor or controller, performs the functions/functions specified in the flowcharts and/or block diagrams. Action is implemented. The program code may execute entirely on the machine, partly on the machine, partly on the machine and partly on a remote machine as a stand-alone software package or entirely on the remote machine or server.

In the context of the present invention, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in connection with the instruction execution system, apparatus or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media may include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media would include one or more wire-based electrical connections, portable computer disks, hard disks, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM or flash memory), fiber optics, compact disk read only memory (CD-ROM), optical storage, magnetic storage, or any suitable combination of the foregoing.

The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited to this. Any person skilled in the art can easily think of various equivalents within the technical scope disclosed by the present invention. Modifications or substitutions should be included within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims (10)

1. A method for detecting abnormal power of a wind power plant is characterized by comprising the following steps:

s1, acquiring a power and wind speed scattered point data set D of the wind power plant;

s2, identifying discrete abnormal data in the power and wind speed scattered data set by adopting a local abnormal factor LOF algorithm to obtain an intermediate data set D';

and step S3, identifying the stacking abnormal data in the intermediate data set D' by adopting an isolated forest algorithm IF.

2. The method for detecting abnormal power of a wind farm according to claim 1, wherein the data of the power and wind speed scatter data set of the wind farm in the step S1 comprises scattered data and stacked data.

3. The method for detecting abnormal power of a wind farm according to claim 1, wherein the step S2 specifically comprises: and calculating a local abnormal factor LOF corresponding to each data point in the data set, regarding the data points with the local abnormal factor LOF values higher than the LOF threshold value as abnormal data, and removing the abnormal data.

4. The method for detecting abnormal power of a wind power plant according to claim 3, wherein the local abnormality factor LOF is an average value of a ratio of a local reachable density of the data points in a k-distance neighborhood to a local reachable density of the data points.

5. The method for detecting abnormal power of a wind farm according to claim 1, wherein the step S3 specifically comprises:

1) randomly sampling the intermediate data set D' to construct an isolated binary tree and construct an isolated forest model;

2) for any sample point in each isolated binary tree, traversing the isolated binary tree from a root node until reaching a leaf node, calculating the average path length c (p) of the sample point, and carrying out abnormal scoring on the sample based on an abnormal index function;

3) and (4) regarding the measured points with the abnormal scoring values higher than the scoring threshold values as abnormal points and removing the abnormal points.

6. A method for detecting abnormal power of a wind farm according to claim 5, characterized in that the expression of the abnormal index function is:

in the formula, h _x From leaf node for sample point xPath Length from Point to root node, E (h) _x ) The path length mean value of the sample point x in all isolated binary trees iTree; c (p) the average path length of the isolated binary tree constructed for p sample data.

7. The method for detecting abnormal power of wind power plant according to claim 6, wherein the path length h of the sample point from the leaf node to the root node is _x The expression of (a) is:

8. method for detecting abnormal power of a wind farm according to claim 6, characterized in that the average path length c (p) is expressed by:

in the formula, H _p-1 Where ═ ln (p-1) + ξ, ξ is the euler constant and p is the number of samples.

9. An electronic device comprising a memory and a processor, the memory having stored thereon a computer program, wherein the processor, when executing the program, implements the method of any of claims 1-8.

10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the method according to any one of claims 1 to 8.

Images