Real-time data management method and device for wind power system

Abstract

The invention discloses a real-time data management method and device for a wind power system. The method comprises the following steps: acquiring operation data of a plurality of measuring points; calculating a correlation matrix based on the operation data of the plurality of measuring points; comparing the correlation matrix with a preset threshold interval matrix to obtain abnormal matrix elements exceeding the preset threshold interval matrix in the correlation matrix; wherein the preset threshold value interval matrix is obtained by calculating previous normal operation data; obtaining a plurality of suspected abnormal measuring points according to the positions of the abnormal matrix elements in the correlation matrix; the frequency data of the suspected abnormal measuring points are subjected to diclustering division, and abnormal reminding is carried out on the suspected abnormal measuring points in a high-frequency cluster formed through diclustering division. According to the method, the quality of the operation data can be effectively monitored, the data access abnormal measurement points are automatically identified, and compared with manual troubleshooting of the operation data by management personnel, the method improves the identification efficiency and the identification accuracy of the data access abnormal measurement points.

Description

Technical Field

[0001] This application relates to the technical field of data management, and in particular, to a method and device for real-time data governance of a wind power system. Background Art

[0002] As a key measure to address climate change, reduce greenhouse gas emissions, and ensure energy security, the development and utilization of wind power not only restructure the traditional energy structure but also drive a systematic transformation of the energy industry.

[0003] In the operation and management of wind power equipment, the full-life cycle management of data from acquisition, transmission to storage is crucial. This process includes multi-level transmission and storage links, and problems such as abnormal acquisition, transmission interruption, and storage failure may occur in each link, all of which will lead to a decline in data quality. At the same time, the characteristics of massive data access and high-frequency acquisition further increase the complexity of data governance. When a data interface configuration error occurs, it is often difficult for managers to quickly locate the problem and conduct effective troubleshooting. Summary of the Invention

[0004] Embodiments of this application provide a method and device for real-time data governance of a wind power system, which solve the technical problem that when a data interface configuration error occurs in a wind power equipment in the prior art, it is often difficult for managers to quickly locate the problem and conduct effective troubleshooting.

[0005] In a first aspect, embodiments of this application provide a method for real-time data governance of a wind power system. The method is applied to each type of equipment provided with multiple measurement points, and the method includes: obtaining operation data of the multiple measurement points; calculating a correlation matrix based on the operation data of the multiple measurement points; comparing the correlation matrix with a preset threshold interval matrix to obtain abnormal matrix elements in the correlation matrix that exceed the preset threshold interval matrix, where the preset threshold interval matrix is calculated from previous normal operation data; obtaining multiple suspected abnormal measurement points according to the positions of the abnormal matrix elements in the correlation matrix; performing binary clustering on the frequency data of the suspected abnormal measurement points, and giving an abnormal reminder for the suspected abnormal measurement points in the high-frequency cluster formed by the binary clustering.

[0006] In combination with the first aspect, in a possible implementation manner, calculating the correlation matrix based on the operation data of the multiple measurement points includes: removing outliers from the operation data; determining whether the operation data of each measurement point conforms to a normal distribution; for the operation data of each pair of measurement points that conform to the normal distribution, calculating the correlation coefficient using the Pearson correlation coefficient method; for the operation data of at least two measurement points that do not conform to the normal distribution, calculating the correlation coefficient using the Spearman rank correlation coefficient method; and constructing the correlation matrix with the correlation coefficients of the operation data of each pair of measurement points as matrix elements.

[0007] In combination with the first aspect, in a possible implementation manner, the preset threshold interval matrix calculated from the previous normal operation data includes: dividing the previous normal operation data into a training set and a test set; calculating the correlation coefficients of each pair of measurement points in the training set and forming a correlation reference matrix; splitting the training set into multiple consecutive subsequences according to a preset time window and respectively determining the time-domain correlation matrices for the multiple consecutive subsequences; calculating the residual matrix of each time-domain correlation matrix and the correlation reference matrix, and constructing the preset threshold interval matrix based on the obtained residual matrix; and verifying the preset threshold interval matrix using the test set.

[0008] In combination with the first aspect, in a possible implementation manner, performing binary clustering on the frequency data of the suspected abnormal measurement points includes: performing binary clustering on the frequency data of the suspected abnormal measurement points using the K-means algorithm to obtain a high-frequency cluster and a low-frequency cluster.

[0009] In a second aspect, an embodiment of the present application provides a real-time data governance device for a wind power system. The real-time data governance device for the wind power system includes: an acquisition module for acquiring the operation data of the multiple measurement points; a calculation module for calculating a correlation matrix based on the operation data of the multiple measurement points; a comparison module for comparing the correlation matrix with a preset threshold interval matrix to obtain abnormal matrix elements in the correlation matrix that exceed the preset threshold interval matrix, where the preset threshold interval matrix is calculated from previous normal operation data; a suspected abnormality module for obtaining multiple suspected abnormal measurement points according to the positions of the abnormal matrix elements in the correlation matrix; a partitioning module for performing binary clustering on the frequency data of the suspected abnormal measurement points; and a reminder module for giving an abnormality reminder for the suspected abnormal measurement points in the high-frequency cluster formed by the binary clustering.

[0010] In combination with the second aspect, in a possible implementation method, the calculation module is specifically used to: eliminate outliers in the operating data; determine whether the operating data of each measuring point conforms to the normal distribution; for each pair of the operating data of the measuring points that conforms to the normal distribution, use the Pearson correlation coefficient method to calculate the correlation coefficient; for at least one of the operating data of two measuring points that do not conform to the normal distribution, use the Spearman rank correlation coefficient method to calculate the correlation coefficient; and construct the correlation matrix using the correlation coefficient of the operating data of each pair of the measuring points as a matrix element.

[0011] In combination with the second aspect, in a possible implementation method, the preset threshold interval matrix is calculated from previous normal operation data, including: dividing the previous normal operation data into a training set and a test set; calculating the correlation coefficient of each pair of measurement points in the training set, and forming a correlation reference matrix; splitting the training set into multiple continuous subsequences according to a preset time window, and determining the time domain correlation matrix for each of the multiple continuous subsequences; calculating the residual matrix between each of the time domain correlation matrices and the correlation reference matrix, and constructing the preset threshold interval matrix based on the obtained residual matrix; and using the test set to verify the preset threshold interval matrix.

[0012] In conjunction with the second aspect, in a possible implementation, the partitioning module is specifically configured to: perform binary clustering on the frequency data of the suspected abnormal measurement points using a K-means algorithm to obtain high-frequency clusters and low-frequency clusters.

[0013] In a third aspect, an embodiment of the present application provides an electronic device, comprising: one or more processors; and a memory storing computer-executable instructions, wherein the computer-executable instructions, when executed by the one or more processors, enable the one or more processors to execute the real-time data management method for a wind power system as described in the first aspect or any possible implementation of the first aspect.

[0014] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, which stores computer-readable instructions. When the computer-readable instructions are executed by a computer, the real-time data management method for the wind power system as described in the first aspect or any possible implementation method of the first aspect is implemented.

[0015] The real-time data management method for wind power systems provided in the embodiment of the present application can effectively monitor the quality of operating data and automatically identify abnormal data access measurement points. Compared with manual investigation of operating data by management personnel, this method improves the efficiency and accuracy of identifying abnormal data access measurement points. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] To more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the accompanying drawings required for the description in the embodiments of the present application. Obviously, the accompanying drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can also be obtained based on these drawings.

[0017] Figure 1 It is a flowchart of the real-time data governance method for the wind power system provided by the embodiments of the present application;

[0018] Figure 2 It is a structural diagram of the real-time data governance device for the wind power system provided by the embodiments of the present application.

[0019] Figure 3 It is a schematic structural diagram of the electronic device provided by the embodiments of the present application. Specific embodiments

[0020] The terms "including" and "having" and any variations thereof in the description and claims of this article are intended to cover non-exclusive inclusion, so that a process, method, system, product or device including a series of units does not have to be limited to those units, but may include other units not clearly listed or inherent to these processes, methods, systems, products or devices.

[0021] In the embodiments of the present application, words such as "exemplary" or "for example" are used to represent examples, illustrations or explanations. Any embodiment or design solution described as "exemplary" or "for example" in the embodiments of the present application should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Rather, the use of words such as "exemplary" or "for example" is intended to present relevant concepts in a specific manner.

[0022] In the description of the embodiments of the present application, unless otherwise specified, the meaning of "a plurality of" refers to two or more. For example, a plurality of processing units refers to two or more processing units, etc.; a plurality of elements refers to two or more elements, etc.

[0023] The embodiments of the present application propose a real-time data governance method for a wind power system. This method determines a plurality of suspected abnormal measurement points by comparing the correlation matrix of the operation data of multiple measurement points with a preset threshold interval matrix, and performs binary clustering on the frequency data of the suspected abnormal measurement points to find the abnormal measurement points and give a reminder.

[0024] The embodiments of the present application provide a real-time data governance method for a wind power system. This method is applied to each type of device equipped with a plurality of measurement points, and the method includes steps S101 to S106.

[0025] S101. Obtain the operation data of multiple measurement points.

[0026] The wind power system has equipment such as a pitch system and a transmission system. The data obtained from multiple measurement points on each piece of equipment has strong correlation. When performing step S101, according to the source of the data, the data is grouped to obtain the operation data of multiple measurement points of each type of equipment.

[0027] S102. Calculate the correlation matrix based on the operation data of multiple measurement points.

[0028] When performing step S102, calculate the correlation coefficients between the operation data of multiple measurement points pairwise, and then use multiple measurement points as the abscissa and ordinate to form a correlation matrix recording the correlation between the operation data of multiple measurement points pairwise. Any value in the correlation matrix represents the correlation between the two measurement points corresponding to that position.

[0029] For example, there are five measurement points (x, y, z, s, t) set in the pitch system. The correlation matrix obtained after performing step S102 is: In the correlation matrix, xy represents the correlation coefficient between the operation data of measurement point x and the operation data of measurement point y. Similarly, the elements in the correlation matrix also represent the correlation coefficients of the operation data of two measurement points.

[0030] S103. Compare the correlation matrix with the preset threshold interval matrix to obtain the abnormal matrix elements in the correlation matrix that exceed the preset threshold interval matrix. Among them, the preset threshold interval matrix is calculated from the previous normal operation data.

[0031] The preset threshold interval matrix records the normal threshold intervals corresponding to each element in the correlation matrix. For example, the preset threshold interval is The correlation matrix is If yz and sz in the correlation matrix exceed the normal threshold intervals corresponding to yz and sz in the preset threshold interval matrix, then yz and sz are determined as abnormal matrix elements.

[0032] S104. According to the positions of the abnormal matrix elements in the correlation matrix, obtain multiple suspected abnormal measurement points.

[0033] Among them, each abnormal matrix element corresponds to two measurement points, and all the measurement points corresponding to all the abnormal matrix elements are determined as suspected abnormal measurement points.

[0034] For example, the correlation matrix is If yz and sz in the correlation matrix are abnormal matrix elements, then measurement points y, z, and s are suspected abnormal measurement points.

[0035] S105. Perform binary clustering on the frequency data of the suspected abnormal measurement points.

[0036] After executing step S105 , a high-frequency cluster and a low-frequency cluster are obtained, and the suspected abnormal measurement points corresponding to the frequency data in the high-frequency cluster are the measurement points where the abnormality is determined to have occurred.

[0037] S106: Providing abnormal reminders for suspected abnormal measurement points in the high-frequency clusters formed by the binary clustering.

[0038] After executing step S106, the management personnel can obtain an abnormality reminder, thereby enabling the management personnel to quickly locate and investigate the data interface configuration that occurred.

[0039] Therefore, the real-time data management method for wind power systems provided in the embodiment of the present application can effectively monitor the quality of operating data and automatically identify abnormal data access measurement points. Compared with manual investigation of operating data by management personnel, this method improves the efficiency and accuracy of identifying abnormal data access measurement points.

[0040] In some embodiments of the present application, step S102 specifically includes steps S201 to S205.

[0041] S201: Eliminate abnormal values in the operating data.

[0042] Due to sensor failure at the measuring point, interference in data transmission, and other issues, outliers may exist in the obtained operating data. After the outliers are eliminated in step S201, the influence of the outliers on the construction of the correlation matrix is avoided, so that the correlation matrix can more accurately reflect the correlation between the two measuring points.

[0043] For example, the interquartile range method is used to eliminate outliers in the operating data.

[0044] S202: Determine whether the operating data of each measuring point conforms to a normal distribution.

[0045] For example, the Jarque-Bera test method is used to determine whether the operating data of each measuring point conforms to the normal distribution. The Jarque-Bera test method is a statistical method used to test whether the data obeys the normal distribution. It estimates the degree of deviation between the sample data and the normal distribution by combining the two characteristics of skewness and kurtosis, and is more suitable for large sample data.

[0046] S203. For each pair of operating data of the measuring points that conforms to the normal distribution, the correlation coefficient is calculated using the Pearson correlation coefficient method.

[0047] S204 . For the operating data of at least one of two measuring points that do not conform to the normal distribution, calculate the correlation coefficient using the Spearman rank correlation coefficient method.

[0048] Divide the motion data into normal distribution and non - normal distribution, and use step S203 and step S204 to calculate the correlation coefficients between the operation data of each pair of measurement points that conform to the normal distribution and the correlation coefficients between the operation data of at least two measurement points that do not conform to the normal distribution respectively, so that the correlation coefficients are more accurate and avoid the deviation of the calculated correlation coefficients.

[0049] S205. Construct a correlation matrix with the correlation coefficients of the operation data of each pair of measurement points as matrix elements.

[0050] In some embodiments of the present application, the preset threshold interval matrix is obtained by calculating the past normal operation data, including step S301 and step S304.

[0051] S301. Divide the past normal operation data into a training set and a test set.

[0052] The past normal operation data can be obtained by manually reviewing and removing outliers from the operation data of all measurement points in the past year.

[0053] Exemplarily, the past normal operation data is taken as the unit of days, and 20% of the data is randomly selected as the test set, and the remaining 80% of the data is used as the training set.

[0054] S302. Calculate the correlation coefficients of each pair of measurement points in the training set and form a correlation reference matrix.

[0055] When performing step S302, the above steps S202 to S205 can be referred to form a correlation reference matrix.

[0056] S303. Split the training set into multiple consecutive subsequences according to a preset time window, and determine the time - domain correlation matrix for each of the multiple consecutive subsequences. Exemplarily, the preset time window is days.

[0057] S304. Calculate the residual matrix of each time - domain correlation matrix and the correlation reference matrix, and construct a preset threshold interval matrix based on the obtained residual matrix.

[0058] After executing step S303, multiple time - domain correlation matrices are obtained. In step S303, each element in each time - domain correlation matrix is compared with the corresponding element in the correlation reference matrix, and the residual matrix is obtained through the following formula: Where, is the residual matrix corresponding to the k - th time - domain correlation matrix, C ij is the correlation reference matrix, is the k - th time - domain correlation matrix.

[0059] When constructing a preset threshold interval matrix with the obtained residual matrix, a preset threshold interval matrix can be constructed according to the maximum residual value and the minimum residual value corresponding to each element in the correlation reference matrix; the calculation formula is as follows: q ij =[c ij +r ij |max, c ij +r ij |min]; where q ij is the element in the i-th row and j-th column of the preset threshold interval matrix, c ij is the element in the i-th row and j-th column of the correlation reference matrix, r ij | max is the maximum residual value corresponding to the element in the i-th row and j-th column of the correlation reference matrix, r ij | min is the minimum residual value corresponding to the element in the i-th row and j-th column of the correlation reference matrix.

[0060] S305. Verify the preset threshold interval matrix using the test set.

[0061] In some embodiments of the present application, step S105 specifically includes: performing binary clustering on the frequency data of suspected abnormal measurement points using the K-means clustering algorithm to obtain a high-frequency cluster and a low-frequency cluster.

[0062] The K-means clustering algorithm is an unsupervised clustering method that divides data into K clusters to make the data points within each cluster as similar as possible. In step S105, the frequencies of suspected abnormal measurement points are classified into two categories: high frequency and low frequency, that is, K = 2. The core idea of K-means is to minimize the sum of the squared distances from all data points to the centers of their respective clusters through iterative optimization.

[0063] Based on the same concept as the embodiments of the foregoing method, an embodiment of the present application also provides a real-time data governance device 200 for a wind power system. The real-time data governance device 200 for a wind power system can be deployed in a data governance terminal device, improving the identification efficiency and accuracy of abnormal measurement points for data access.

[0064] As Figure 2 shown, the real-time data governance device 200 for a wind power system includes an acquisition module 201, a calculation module 202, a comparison module 203, a suspected abnormal module 204, a division module 205, and a reminder module 206.

[0065] The acquisition module 201 is used to obtain operating data from multiple measuring points. The calculation module 202 is used to calculate a correlation matrix based on the operating data from the multiple measuring points. The comparison module 203 is used to compare the correlation matrix with a preset threshold interval matrix to obtain abnormal matrix elements in the correlation matrix that exceed the preset threshold interval matrix; wherein the preset threshold interval matrix is calculated based on previous normal operating data. The suspected abnormality module 204 is used to obtain multiple suspected abnormal measuring points based on the position of the abnormal matrix elements in the correlation matrix. The division module 205 is used to divide the frequency data of the suspected abnormal measuring points into two clusters. The reminder module 206 is used to issue abnormal reminders for the suspected abnormal measuring points in the high-frequency clusters formed by the two-cluster division.

[0066] The calculation module 202 is specifically used to: eliminate outliers in the operating data; determine whether the operating data of each measuring point conforms to the normal distribution; calculate the correlation coefficient between the operating data of each pair of measuring points that conform to the normal distribution using the Pearson correlation coefficient method; calculate the correlation coefficient between the operating data of at least one of two measuring points that do not conform to the normal distribution using the Spearman rank correlation coefficient method; and construct a correlation matrix using the correlation coefficient of the operating data of each pair of measuring points as a matrix element.

[0067] The preset threshold interval matrix is obtained by calculating the previous normal operation data, including: dividing the previous normal operation data into a training set and a test set; calculating the correlation coefficient of each pair of measurement points in the training set and forming a correlation reference matrix; splitting the training set into multiple continuous subsequences according to a preset time window, and determining the time domain correlation matrix for each of the multiple continuous subsequences; calculating the residual matrix between each time domain correlation matrix and the correlation reference matrix, and constructing a preset threshold interval matrix based on the obtained residual matrix; and using the test set to verify the preset threshold interval matrix.

[0068] The partitioning module 205 is specifically used to perform binary clustering on the frequency data of suspected abnormal measurement points using the K-means algorithm to obtain high-frequency clusters and low-frequency clusters.

[0069] like Figure 3 As shown, an embodiment of the present application also provides an electronic device 300, which includes one or more processors 301 and a memory 302 for storing computer-executable instructions. When the computer-executable instructions are executed by the one or more processors 301, the one or more processors 301 execute the above-mentioned wind power system real-time data management method.

[0070] In an embodiment of the present application, the processor 301 may be a central processing unit (CPU), and the processor 301 may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0071] The memory 302 may include a read-only memory and a random access memory, and provide instructions and data to the processor 301. The memory 302 may also include a non-volatile random access memory. Optionally, the random access memory may be, for example, a high bandwidth memory (HBM).

[0072] The memory 302 may be a volatile memory, a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchlink DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0073] An embodiment of the present application provides a computer-readable storage medium, which stores computer-readable instructions. When the computer-readable instructions are executed by a computer, the above-mentioned real-time data governance method for a wind power system is implemented.

[0074] The various embodiments in this specification are described in a progressive manner. For the same or similar parts between the various embodiments, reference can be made to each other. The key point of each embodiment is to illustrate the differences from other embodiments.

[0075] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit the present application. Although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the present application.

Claims

1. A real-time data governance method for a wind power system, characterized in that, The method is applied to each type of device with multiple measurement points, and the method includes: Obtaining the operation data of the multiple measurement points; Calculating a correlation matrix based on the operation data of the multiple measurement points; Comparing the correlation matrix with a preset threshold interval matrix to obtain abnormal matrix elements in the correlation matrix that exceed the preset threshold interval matrix; wherein, the preset threshold interval matrix is calculated from past normal operation data; Obtaining multiple suspected abnormal measurement points according to the positions of the abnormal matrix elements in the correlation matrix; Performing binary clustering on the frequency data of the suspected abnormal measurement points, and giving an abnormal reminder for the suspected abnormal measurement points in the high-frequency cluster formed by the binary clustering.

2. The method according to claim 1, wherein The calculating a correlation matrix based on the operation data of the multiple measurement points includes: Removing outliers from the operation data; Judging whether the operation data of each measurement point conforms to a normal distribution; For the operation data between each pair of measurement points that conform to a normal distribution, calculating a correlation coefficient using the Pearson correlation coefficient method; For the operation data between at least two measurement points that do not conform to a normal distribution, calculating a correlation coefficient using the Spearman rank correlation coefficient method; Constructing the correlation matrix with the correlation coefficients of the operation data of each pair of measurement points as matrix elements.

3. The method according to claim 1, wherein The preset threshold interval matrix being calculated from past normal operation data includes: Dividing the past normal operation data into a training set and a test set; Calculating the correlation coefficients between each pair of measurement points in the training set and forming a correlation reference matrix; Dividing the training set into multiple consecutive subsequences according to a preset time window, and respectively determining time-domain correlation matrices for the multiple consecutive subsequences; Calculating the residual matrix between each time-domain correlation matrix and the correlation reference matrix, and constructing the preset threshold interval matrix based on the obtained residual matrix; Validating the preset threshold interval matrix using the test set.

4. The method according to claim 1, wherein The performing binary clustering on the frequency data of the suspected abnormal measurement points includes: Performing binary clustering on the frequency data of the suspected abnormal measurement points using the K-means algorithm to obtain a high-frequency cluster and a low-frequency cluster.

5. A real-time data governance device for a wind power system, characterized in that, Includes: An acquisition module for obtaining the operation data of the multiple measurement points; A calculation module for calculating a correlation matrix based on the operation data of the multiple measurement points; A comparison module for comparing the correlation matrix with a preset threshold interval matrix to obtain abnormal matrix elements in the correlation matrix that exceed the preset threshold interval matrix; wherein, the preset threshold interval matrix is calculated from past normal operation data; A suspected abnormal module for obtaining multiple suspected abnormal measurement points according to the positions of the abnormal matrix elements in the correlation matrix; A division module for performing binary clustering on the frequency data of the suspected abnormal measurement points; A reminder module for giving an abnormal reminder for the suspected abnormal measurement points in the high-frequency cluster formed by the binary clustering.

6. An electronic device, characterized in that, Includes: One or more processors; And A memory storing computer-executable instructions that, when executed by the one or more processors, cause the one or more processors to perform the real-time data governance method for a wind power system as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-readable instructions that, when executed by a computer, implement the real-time data governance method for a wind power system as described in any one of claims 1 to 4.

Images