Wind energy data quality control method and device based on clustering algorithm

Abstract

The invention provides a wind energy data quality control method and device based on a clustering algorithm, and belongs to the technical field of wind energy data processing. A fitting function is determined based on S-shaped curve characteristics between the wind speed and the wind energy generating capacity, and parameters of the fitting function are initialized; calculating the shortest distance from each sample point to the fitting function; clustering the distance sets of all the sample points based on a Gaussian mixture model to obtain a normal data cluster and an abnormal data cluster; and iteratively updating the parameters of the fitting function by using the sample points of the normal data cluster until a stable threshold value is reached, and using the finally obtained sample points of the normal data cluster as wind energy data subjected to quality control processing for fan performance monitoring. Through S-shaped curve fitting, the distance is used as an anomaly judgment feature, the distance is clustered by using GMM, and fitting parameters are iteratively updated, so that accurate identification and elimination of abnormal data are realized, and the quality of wind energy data is optimized.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of wind energy data processing, and in particular, relates to a wind energy data quality control method and device based on a clustering algorithm. BACKGROUND

[0002] Global economic growth and the improvement of human living standards have accelerated the demand for energy, while the limitations of traditional fossil energy and the harm to the environment have become increasingly prominent, which has accelerated the development of renewable energy. As a renewable, pollution-free and widely distributed energy, wind energy has been increasingly valued. Wind power plants are configured with a series of wind turbines for power generation according to site selection considerations such as wind conditions, surrounding terrain, power transmission lines, etc. The power generated by each turbine is transmitted to a substation and then enters the power grid. However, due to fluctuations in wind and energy demand, wind turbines are subject to complex conditions under alternating loads. Poor working conditions are more likely to cause wind turbine performance to decline, reduce wind output, and further increase operating and maintenance costs. Accurate performance monitoring is the key to ensuring the reliable operation of wind turbine generators.

[0003] In the prior art, the power curve (i.e., the relationship curve between wind speed at the hub height of the wind turbine and the corresponding alternating current output power) is the core indicator for wind turbine performance monitoring and has been widely used in wind farm development, wind turbine state monitoring, data preprocessing and quality control, etc. However, the current wind energy data quality control method has the following shortcomings. (1) Poor non-linear adaptability: wind speed and power have a non-linear S-shaped relationship, and traditional quality control methods (such as fixed threshold method) are difficult to adapt to this non-linear characteristic, which is prone to misjudgment of normal data or omission of abnormal data. (2) Low accuracy of anomaly detection: it relies on a single statistical rule and cannot effectively distinguish between normal fluctuation data and real abnormal data, especially under complex working conditions (such as seasonal wind speed changes), the quality control effect is unstable; (3) Static parameters: the parameters of the fitted power curve are fixed, and the dynamic changes of the data are not considered, which is prone to cause the quality control accuracy to decrease due to parameter deviation after long-term use, and cannot meet the demand for high-quality data for historical data optimization and deep learning data set construction. SUMMARY

[0004] Therefore, the purpose of the present application is to provide a wind energy data quality control method and device based on a clustering algorithm, which can accurately identify and eliminate abnormal data and optimize the quality of wind energy data.

[0005] The wind energy data quality control method based on a clustering algorithm provided by the embodiments of the present application comprises the following steps: Collect wind energy data of an electric field wind tower or a meteorological wind tower, and preprocess the collected wind energy data to obtain sample points; the wind energy data includes wind speed and corresponding wind power generation; determine a fitting function based on the S-shaped curve characteristics between the wind speed and the wind power generation amount, and initialize the parameters of the fitting function; calculate the shortest distance of each sample point to the fitting function to obtain a distance set; and cluster the distance set of all sample points based on a Gaussian mixture model to obtain a normal data cluster and an abnormal data cluster; update the parameters of the fitting function by using the sample points of the normal data cluster iteratively until a stable threshold is reached, and finally obtain the sample points of the normal data cluster as wind energy data after quality control for fan performance monitoring.

[0006] In some embodiments, the sample points are represented by a binary tuple , i is the serial number of the sample point, , and v i and P i are the wind speed and the corresponding wind power generation amount of the sample point, respectively.

[0007] In some embodiments, the determination of the fitting function based on the S-shaped curve characteristics between the wind speed and the wind power generation amount, and the initialization of the parameters of the fitting function, include the following steps: determine the fitting function based on the S-shaped curve characteristics between the wind speed and the wind power generation amount as ; wherein is an input variable reflecting the wind speed, is a fitting function value reflecting the wind power generation amount under the corresponding wind speed, is a steepness parameter, is a wind power generation amount upper limit parameter, is a turning point parameter; initialize the parameters of the fitting function; wherein , , .

[0008] In some embodiments, the calculation of the shortest distance of each sample point to the fitting function includes the following steps: perform derivative calculation on the fitting function to obtain the derivative of the fitting function ; arbitrarily select a point in the fitting function as a reference point , determine the tangent slope of the reference point based on the derivative of the fitting function, and obtain the tangent equation of the fitting function at the reference point based on the tangent slope; determine the sample points on the normal line at the reference point, calculate the perpendicular distance of each sample point to the tangent equation based on the distance formula of a point to a straight line in a plane, and record the perpendicular distance as the shortest distance of each sample point to the fitting function .

[0009] In some embodiments, the clustering of the distance set of all sample points based on the Gaussian mixture model to obtain the normal data cluster and the abnormal data cluster comprises the following steps: Setting the clustering categories as the normal data cluster and the abnormal data cluster; Using the expectation maximization algorithm to estimate the parameters of the Gaussian mixture model, including the weight, mean and variance of each clustering category; For the shortest distance of each sample point , calculating the probability of belonging to the normal data cluster, and according to the set probability judgment threshold, assigning a normal sample label or an abnormal sample label to each sample point.

[0010] In some embodiments, the sample points assigned with the normal sample label are used as new fitting data for iterative updating of the parameters of the fitting function.

[0011] In some embodiments, the parameters of the fitting function are iteratively updated by the following steps: Updating the parameters of the fitting function based on the new fitting data; wherein the steepness parameter and the inflection point parameter are optimized by the nonlinear least squares method, and the upper limit of wind power generation parameter is optimized by the maximum wind power generation of the new fitting data; Based on the new fitting data and the updated fitting function, the shortest distance calculation and clustering are performed again, and the iterative updating is performed until the stable threshold is reached; wherein the stable threshold is set based on the absolute value of the difference between the steepness parameter or the inflection point parameter in the last two iterations.

[0012] In some embodiments, a wind energy data quality control device based on a clustering algorithm is also provided, which comprises: A collection module for collecting wind energy data of an electric field wind tower or a meteorological wind tower, and pre-processing the collected wind energy data to obtain sample points; the wind energy data includes wind speed and corresponding wind power generation; An initialization module for determining a fitting function based on the S-shaped curve characteristics between wind speed and wind power generation, and initializing the parameters thereof; A clustering module for calculating the shortest distance of each sample point to the fitting function to obtain a distance set; and clustering the distance set of all sample points based on the Gaussian mixture model to obtain the normal data cluster and the abnormal data cluster; An updating module is configured to iteratively update parameters of the fitting function by using sample points of the normal data cluster until a stable threshold is reached, and finally obtain sample points of the normal data cluster as wind energy data after quality control for wind turbine performance monitoring.

[0013] In some embodiments, an electronic device is also provided, which includes a processor, a memory, and a bus, the memory stores machine readable instructions executable by the processor, when the electronic device is running, the processor communicates with the memory through the bus, and the machine readable instructions are executed by the processor to perform the steps of the wind energy data quality control method based on the clustering algorithm.

[0014] In some embodiments, a computer readable storage medium is also provided, which stores a computer program, when the computer program is run by a processor, the steps of the wind energy data quality control method based on the clustering algorithm are performed.

[0015] The wind energy data quality control method and device based on the clustering algorithm collect wind energy data of an electric field wind measurement tower or a meteorological wind measurement tower, and preprocess the collected wind energy data to obtain sample points; the wind energy data includes wind speed and corresponding wind power generation; a fitting function is determined based on the S-shaped curve characteristics between the wind speed and the wind power generation, and parameters of the fitting function are initialized; the shortest distance of each sample point to the fitting function is calculated to obtain a distance set; and the distance set of all sample points is clustered based on a Gaussian mixture model to obtain a normal data cluster and an abnormal data cluster; the parameters of the fitting function are iteratively updated by using the sample points of the normal data cluster until a stable threshold is reached, and finally the sample points of the normal data cluster are obtained as wind energy data after quality control for wind turbine performance monitoring. Thus, on the one hand, the S-shaped curve is used to fit the nonlinear relationship between the wind speed and the corresponding wind power generation, which accurately matches the actual operating characteristics of the wind turbine and avoids the fitting deviation of the traditional threshold method; on the other hand, the distance is used as an abnormal feature, and the probability clustering of the Gaussian mixture model can effectively distinguish between normal fluctuations and real abnormalities, thereby improving the abnormal detection accuracy; and the fitting parameters are iteratively updated to adapt to dynamic changes in data, and high quality control accuracy can still be maintained in long-term use. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0017] Figure 1 A flow chart of the wind energy data quality control method based on the clustering algorithm is shown; Figure 2 A wind energy quality control result schematic diagram of A1 station in province A is shown; Figure 3 A wind energy quality control result schematic diagram of B1 station in province B is shown; Figure 4 A structure schematic diagram of the wind energy data quality control device based on the clustering algorithm is shown; Figure 5 A structure schematic diagram of the electronic device is shown. DETAILED DESCRIPTION

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. It should be understood that the drawings in the present application only serve the purpose of description and illustration, and are not used to limit the scope of protection of the present application. In addition, it should be understood that the schematic drawings are not drawn according to the actual proportions. The flow chart shows the operations implemented according to some embodiments of the present application. It should be understood that the operations of the flow chart can not be implemented in sequence, and the steps without logical context relationship can be reversed in sequence or implemented simultaneously. In addition, one or more other operations can be added to the flow chart or removed from the flow chart under the guidance of the content of the present application by those skilled in the art.

[0019] In addition, the described embodiments are only some of the embodiments of the present application, not all the embodiments. The components of the embodiments of the present application described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0020] It should be noted that the term "comprising" will be used in the embodiments of the present application to indicate the presence of the features declared thereafter, but does not exclude the addition of other features.

[0021] In view of the technical problems proposed in the background art, the present application provides a wind energy data quality control method and device based on a clustering algorithm, which can realize accurate identification and elimination of abnormal data and optimize the quality of wind energy data.

[0022] Reference is made to the drawings accompanying the specification Figure 1The application provides a wind energy data quality control method based on a clustering algorithm, and the method comprises the following steps: S1, collecting wind energy data of an electric field wind tower or a meteorological wind tower, and pre-processing the collected wind energy data to obtain sample points; the wind energy data comprises wind speed and corresponding wind power generation capacity; S2, determining a fitting function based on the S-shaped curve characteristics between the wind speed and the wind power generation capacity, and initializing the parameters of the fitting function; S3, calculating the shortest distance of each sample point to the fitting function to obtain a distance set; and clustering the distance set of all sample points based on a Gaussian mixture model to obtain a normal data cluster and an abnormal data cluster; S4, iteratively updating the parameters of the fitting function by using the sample points of the normal data cluster until a stable threshold is reached, and taking the finally obtained sample points of the normal data cluster as wind energy data subjected to quality control processing, which is used for wind turbine performance monitoring.

[0023] Specifically, in step S1, the data source is first determined, and it is clear that the wind energy data comes from a meteorological wind tower or a wind field wind tower, and covers at least one observation station (the observation station can include a single or multiple wind turbines), for example, referring to Table 1, the data of 6 observation stations are selected, wherein the C1 station includes wind energy data of 45 wind turbines.

[0024]

[0025] Table 1 Then, the collection parameters are set, for example, the time interval of fixed data collection is 15 minutes, so as to ensure that each time node corresponds to a complete set of observation data; wherein the observation data of a single time node is defined as a sample point, and each sample point is composed of wind speed (WS) at the time and wind power generation capacity (P) at the corresponding time, which is denoted as wherein i is the sample point serial number, and the value range is 1 to the total number of sample points n. In an embodiment, the total number of sample points of each station is shown in Table 2.

[0026]

[0027] Table 2 Further, the collected wind energy data is also subjected to data preprocessing, and the obvious abnormal values and missing values in the data are removed, including but not limited to logically contradictory data with wind speed of 0 but wind power generation capacity greater than 0, missing data with empty wind speed or wind power generation capacity field, so as to ensure that the data integrity after preprocessing meets the subsequent calculation requirements.

[0028] In step S2, based on the nonlinear relationship (S-shaped curve characteristics) between the wind speed and the wind power generation capacity, an S-shaped curve is selected as the fitting function of the two, and the specific formula is as follows:

[0029] in, As an input variable reflecting wind speed, To reflect the fitted function value of wind power generation at the corresponding wind speed, For steepness parameter, This refers to the upper limit parameter for wind power generation. These are the inflection point parameters. The initial fitting parameters are set as follows: referring to the algorithm initialization rules, the kurtosis parameter of the S-shaped fitting curve is... The initial value is set to 1 (the steepness parameter is used to control the slope of the curve near the inflection point); calculate the wind power generation at all sample points. maximum value The upper limit parameter of wind power generation from the S-shaped fitted curve. The initial value is set to this maximum value (the upper limit parameter for wind power generation is used to control the maximum theoretical output value of the curve); calculate the maximum wind speed WS among all sample points. Half of this maximum value is taken as the initial value of the inflection point parameter of the S-shaped fitting curve (the inflection point parameter is used to locate the critical wind speed at which the wind power generation of the curve grows the fastest).

[0030] By constructing a fitting function and initializing its parameters, a benchmark is provided for subsequent distance calculation, initial clustering, and parameter iteration. This ensures that abnormal wind energy data can be gradually eliminated, parameters can be optimized, and finally converged to an accurate fitting curve, thus achieving high-quality wind energy data quality control.

[0031] In step S3, when calculating the shortest distance from each sample point to the fitting function, the derivative of the fitting function constructed in step S1 is first calculated. .

[0032] Then, arbitrarily select a point as a reference point from the fitting function constructed in step S1. and reference point Substitute the derivative of the fitted function We obtain the slope of the tangent at the reference point, and based on the point-slope formula, combined with the coordinates of the reference point and the slope of the tangent, we derive the equation of the tangent to the fitted function at that reference point. .

[0033] Finally, based on the formula for the distance from a point in a plane to a line, the perpendicular distance from each sample point to the tangent equation is calculated. ; among which sample points At the reference point On the normal line, therefore the two satisfy the equation. Then the vertical distance This is the shortest distance from the i-th sample point to the fitted function.

[0034] Further, the distance set of all sample points obtained is clustered. This is because the smaller the shortest distance of the sample point to the fitting function, the higher the matching degree of the wind speed and the wind power generation capacity at that moment, which conforms to the S-shaped relationship of the normal operation of the wind turbine; the larger the shortest distance of the sample point to the fitting function, the lower the matching degree of the wind speed and the wind power generation capacity at that moment, which deviates from the normal operation of the wind turbine, and is most likely to be abnormal data caused by wind turbine / related sensor failure.

[0035] In the present application, a clustering method based on probability distribution is adopted, i.e., Gaussian Mixture Module (GMM), which is used to represent a data set composed of multiple Gaussian distributions (normal distributions), and the mathematical expression is as follows:

[0036] wherein K is the number of Gaussian distributions, is the weight of the kth Gaussian distribution, is the kth Gaussian distribution, and the mean and the variance are respectively and .

[0037] Specifically, when the distance set of all sample points obtained is clustered based on the Gaussian mixture model, the number of clustering categories is first determined, and the number of components n_components of the Gaussian mixture model GMM is determined to classify the normal data cluster (data conforming to the wind speed-wind power generation capacity rule) and the abnormal data cluster (data deviating from the wind speed-wind power generation capacity rule). And the GMM model parameters are estimated. In an embodiment, the expectation maximization EM algorithm is used to estimate the parameters of the GMM model. The parameters to be estimated include: the weight of each clustering category ( , k = 1, 2,..., n_components, the weight represents the proportion of the corresponding category sample in the total sample, and the sum of the weights of the two categories is 1); the mean of each clustering category ( , k = 1, 2,..., n_components, the mean of the normal data cluster is smaller than that of the abnormal data cluster, and the difference in distance value is large); the variance of each clustering category ( , k = 1, 2,..., n_components, reflecting the dispersion degree of the distance value of the sample in the category); in addition, a probability judgment standard needs to be set, for example, the distance value of each sample point , set its probability of belonging to the normal data cluster ≥ p (0 < p < 1). Further, according to the probability judgment result, a label is assigned to each sample point, for example, if the sample point belongs to the normal data cluster with a probability ≥ p, it is marked as a normal sample (label = 1); if the probability < p, it is marked as an abnormal sample (label = 0). And the sample points marked as abnormal samples (label = 0) are removed, and only the sample points marked as normal samples (label = 1) are retained as new fitting data for subsequent iterative updating of the parameters of the fitting function.

[0038] In step S4, first, the parameters of the fitting function are optimized based on the new fitting data, especially the optimization of the steepness parameter and the inflection point parameter . Among them, the steepness parameter = 1 is a general default value, but the wind speed-wind power generation growth rate of different fans / stations is different, and when refitting, the k value that can make the wind power generation of the normal sample and the theoretical value of the function error is minimized can be found through the nonlinear least squares optimization algorithm. The inflection point parameter is a rough estimate initially set, but the real inflection point (the critical wind speed at which the wind power generation grows the fastest) may shift due to different fan models, and when refitting, the inflection point parameter can also be adjusted to make the wind power generation growth inflection point of the S-shaped curve accurately correspond to the concentrated growth interval of the normal sample according to the principle of minimizing the error. And the wind power generation upper limit parameter is always taken as the maximum wind power generation in the current sample (normal sample obtained by clustering).

[0039] Then the fitting function with updated parameters is substituted into step S3 to recompute the shortest distance of all normal sample points to the new fitting curve, and the GMM clustering and sample screening are reperformed, forming a cycle of parameter updating, distance calculation, and clustering screening, until it is determined that the parameters of the fitting function reach a stable state, and the iteration cycle is stopped. Among them, a parameter stability threshold δ is set, such as the absolute value of the difference between the steepness parameter in the last two iterations ≤ δ, and / or the absolute value of the difference between the inflection point parameter ≤ δ.

[0040] All normal sample points finally retained are wind energy data after quality control, which are used to construct a deep learning / machine learning data set to support fan power prediction, fault diagnosis, and operation and maintenance decisions. In an embodiment, the wind energy quality control results of A1 station in province A and the wind energy quality control results of B1 station in province B are shown in the attached Figure 2 and the attached Figure 3 of the specification, respectively. It can be seen that the normal data is distributed along the S-shaped curve, and the abnormal data deviates from the curve.

[0041] The application provides a wind energy data quality control method based on a clustering algorithm. On one hand, a nonlinear relationship between wind speed and wind power generation is fitted by an S-shaped curve to accurately match the actual operation law of a wind turbine and avoid the fitting deviation of a traditional fixed threshold method. On the other hand, the distance from a sample point to a curve is taken as a feature, and GMM probability clustering is combined to distinguish normal fluctuations from real abnormalities and greatly reduce the misjudgment rate. Furthermore, the fitting parameters are iteratively updated to dynamically adapt to data changes, and high quality control accuracy is still maintained in long-term use.

[0042] Based on the same inventive concept, the application also provides a wind energy data quality control device based on a clustering algorithm. Since the principle of the device in the application for solving the problem is similar to the above-mentioned wind energy data quality control method based on a clustering algorithm, the implementation of the device can be referred to the implementation of the method, and the repeated parts will not be described here.

[0043] As shown in the accompanying drawings Figure 4 The application also provides a wind energy data quality control device based on a clustering algorithm. The device comprises: A collection module 401 is configured to collect wind energy data of an electric field wind tower or a meteorological wind tower, and pre-process the collected wind energy data to obtain sample points. The wind energy data includes wind speed and corresponding wind power generation. An initialization module 402 is configured to determine a fitting function based on the S-shaped curve feature between wind speed and wind power generation, and initialize the parameters of the fitting function. A clustering module 403 is configured to calculate the shortest distance from each sample point to the fitting function to obtain a distance set, and cluster the distance set of all sample points based on a Gaussian mixture model to obtain a normal data cluster and an abnormal data cluster. An updating module 404 is configured to iteratively update the parameters of the fitting function using the sample points of the normal data cluster until a stable threshold is reached, and finally obtain the sample points of the normal data cluster as wind energy data after quality control for wind turbine performance monitoring.

[0044] In some embodiments, the initialization module 402 represents the sample point by a binary tuple , where i is the serial number of the sample point, , and wind speed and corresponding wind power generation of the sample point, respectively.

[0045] In some embodiments, the initialization module 402 determines the fitting function based on the S-shaped curve feature between wind speed and wind power generation, and initializes the parameters of the fitting function, including: determining the fitting function as based on the S-shaped curve feature between wind speed and wind power generation; and initializing the parameters of the fitting function. to reflect the input variable of wind speed, to reflect the fitting function value of wind power generation under corresponding wind speed, to reflect the steepness parameter, to reflect the upper limit parameter of wind power generation, to reflect the inflection point parameter; the parameters of the fitting function are initialized and set; wherein, , , .

[0046] In some embodiments, the clustering module 403 calculates the shortest distance of each sample point to the fitting function, including: performing derivative calculation on the fitting function to obtain the fitting function derivative ; selecting an arbitrary point in the fitting function as a reference point , and determining the tangent slope of the reference point based on the fitting function derivative, and obtaining the tangent equation of the fitting function at the reference point based on the tangent slope; determining the sample point on the normal line of the reference point, calculating the perpendicular distance of each sample point to the tangent equation based on the distance formula of a point to a straight line in a plane, and recording the perpendicular distance as the shortest distance of each sample point to the fitting function . And clustering the distance set of all sample points based on the Gaussian mixture model to obtain normal data clusters and abnormal data clusters, including: setting the clustering categories as normal data clusters and abnormal data clusters; using the expectation maximization algorithm to estimate the parameters of the Gaussian mixture model, including the weight, mean and variance of each clustering category; for the shortest distance of each sample point, calculating the probability of belonging to the normal data cluster, and according to the set probability judgment threshold, assigning each sample point a normal sample label or an abnormal sample label. Wherein, the sample points assigned with the normal sample label are used as new fitting data for iterative updating of the parameters of the fitting function.

[0047] In some embodiments, the updating module 404 iteratively updates the parameters of the fitting function, including: updating the parameters of the fitting function based on the new fitting data; wherein, the steepness parameter and the inflection point parameter are optimized by the nonlinear least squares method, and the wind power generation upper limit parameter is optimized by the maximum wind power generation of the new fitting data; based on the new fitting data and the updated fitting function after parameter updating, the shortest distance calculation and clustering are performed again, and the iterative updating is performed until the stable threshold is reached; wherein, the stable threshold is set based on the absolute value of the difference between the steepness parameter or the inflection point parameter in two consecutive iterations.

[0048] The wind energy data quality control device based on the clustering algorithm provided in the application collects wind energy data of an electric field wind tower or a meteorological wind tower through a collection module, and pre-processes the collected wind energy data to obtain sample points; the wind energy data includes wind speed and corresponding wind power generation; a fitting function is determined based on the S-shaped curve characteristics between the wind speed and the wind power generation through an initialization module, and parameters thereof are initialized and set; the shortest distance of each sample point to the fitting function is calculated through a clustering module to obtain a distance set; and the distance sets of all sample points are clustered based on a Gaussian mixture model to obtain normal data clusters and abnormal data clusters; the parameters of the fitting function are iteratively updated using the sample points of the normal data clusters through an updating module until a stable threshold is reached, and the sample points of the final normal data clusters are taken as wind energy data after quality control processing, which is used for wind turbine performance monitoring. Thus, on the one hand, the nonlinear relationship between the wind speed and the corresponding wind power generation is fitted through the S-shaped curve, the actual operation characteristics of the wind turbine are accurately matched, and the fitting deviation of the traditional threshold method is avoided; on the other hand, the distance is taken as an abnormal feature, the probability clustering of the Gaussian mixture model is combined, the normal fluctuation and the real abnormality can be effectively distinguished, and the abnormal detection accuracy is improved; and the fitting parameters are iteratively updated, and then the data dynamic change is adapted, and the high quality control accuracy can be maintained for a long time.

[0049] Based on the same concept of the application, as shown in the description Figure 5 The electronic device 500 provided by the embodiment of the application includes at least one processor 501, at least one network interface 504 or other user interface 503, a memory 505, and at least one communication bus 502. The communication bus 502 is used to realize the connection and communication between the components. The electronic device 500 can optionally include a user interface 503, including a display (for example, a touch screen, an LCD, a CRT, holographic imaging (Holographic) or a projector (Projector), etc.), a keyboard or a clicking device (for example, a mouse, a trackball, a touchpad or a touch screen, etc.).

[0050] The memory 505 can include a read-only memory and a random access memory, and provides instructions and data for the processor 501. A part of the memory 505 can also include a non-volatile random access memory (NVRAM).

[0051] In some embodiments, the memory 505 stores the following elements, executable modules or data structures, or a subset of them, or an extended set of them: An operating system 5051 includes various system programs, used to realize various basic services and process hardware-based tasks; The application program module 5052 includes various application programs, such as a desktop (launcher), a media player (MediaPlayer), a browser (Browser), and the like, for implementing various application services.

[0052] In the embodiment of the present application, the processor 501 is configured to execute the steps of the wind energy data quality control method based on the clustering algorithm by invoking the program or instruction stored in the memory 505.

[0053] The present application also provides a computer readable storage medium having a computer program stored thereon, the computer program being executed by a processor to perform the steps of the wind energy data quality control method based on the clustering algorithm.

[0054] Specifically, the storage medium can be a general storage medium, such as a mobile disk, a hard disk, or the like, and the computer program stored on the storage medium can be executed to accurately identify and eliminate abnormal data and optimize the quality of wind energy data.

[0055] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other manners. The described device embodiments are only schematic. For example, the division of the units is only a logical function division, and there can be another division manner in actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed coupling or direct coupling or communication connection between the units can be indirect coupling or communication connection through some interface, electrical or mechanical, or in other forms.

[0056] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they can be located in one place, or distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the present embodiment.

[0057] In addition, each functional unit in the embodiments provided in the present application can be integrated into one processing unit, or each unit can be a physically independent unit, or two or more units can be integrated into one unit.

[0058] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the parts of the prior art that make contributions or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0059] Finally, it should be noted that: the above embodiments are only specific embodiments of the present application, which are used to illustrate the technical solutions of the present application, but not to limit them. The protection scope of the present application is not limited thereto. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can modify or easily think of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed by the present application, or replace some technical features with equivalent ones. The modifications, changes or replacements do not make the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application. They should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A wind energy data quality control method based on clustering algorithm, characterized in that, The method includes the following steps: Wind energy data is collected from electric field wind measurement towers or meteorological wind measurement towers, and the collected wind energy data is preprocessed to obtain sample points; the wind energy data includes wind speed and corresponding wind power generation. The fitting function is determined based on the S-shaped curve characteristics between wind speed and wind power generation, and its parameters are initialized. Calculate the shortest distance from each sample point to the fitted function to obtain a distance set; and cluster all sample points' distance sets based on a Gaussian mixture model to obtain normal data clusters and abnormal data clusters. The parameters of the fitting function are iteratively updated using the sample points of the normal data cluster until a stable threshold is reached. The sample points of the final normal data cluster are then used as wind energy data after quality control processing for wind turbine performance monitoring.

2. The wind energy data quality control method based on clustering algorithm according to claim 1, characterized in that, in, Sample points through binary pairs Indicates that i is the index of the sample point. These represent the wind speed and corresponding wind power generation at the sample points, respectively.

3. The wind energy data quality control method based on clustering algorithm according to claim 2, characterized in that, The process of determining the fitting function based on the S-shaped curve characteristics between wind speed and wind power generation, and initializing its parameters, includes the following steps: The fitting function is determined based on the S-shaped curve characteristics between wind speed and wind power generation. ;in, As an input variable reflecting wind speed, To reflect the fitted function value of wind power generation at the corresponding wind speed, For steepness parameter, This refers to the upper limit parameter for wind power generation. Inflection point parameters; The parameters of the fitting function are initialized; wherein, , , .

4. The wind energy data quality control method based on clustering algorithm according to claim 3, characterized in that, The calculation of the shortest distance from each sample point to the fitted function includes the following steps: The derivative of the fitted function is obtained by taking the derivative of the fitted function. ; Arbitrarily select a point as a reference point in the fitting function. Furthermore, the slope of the tangent at the reference point is determined based on the derivative of the fitting function, and the equation of the tangent at the reference point is obtained based on the slope of the tangent. Determine the sample points on the normal line of the reference point, calculate the perpendicular distance from each sample point to the tangent equation based on the formula for the distance from a point to a line in a plane, and record this perpendicular distance as the shortest distance from each sample point to the fitted function. .

5. The wind energy data quality control method based on clustering algorithm according to claim 4, characterized in that, The method of clustering the distance set of all sample points based on the Gaussian mixture model to obtain normal data clusters and abnormal data clusters includes the following steps: Define the clustering categories as normal data clusters and abnormal data clusters; The parameters of the Gaussian mixture model are estimated using the expectation-maximization algorithm, including the weights, mean, and variance of each cluster. Shortest distance for each sample point Calculate the probability that it belongs to the normal data cluster, and assign a normal sample label or an abnormal sample label to each sample point according to the set probability judgment threshold.

6. The wind energy data quality control method based on clustering algorithm according to claim 5, characterized in that, in, The sample points with normal sample labels are used as new fitting data for iteratively updating the parameters of the fitting function.

7. The wind energy data quality control method based on clustering algorithm according to claim 6, characterized in that, The parameters of the fitted function are iteratively updated in the following manner. Includes the following steps: The parameters of the fitted function are updated based on the new fitted data; among them, the kurtosis parameter is adjusted using a nonlinear least squares method. and inflection point parameters Optimization was performed by adjusting the upper limit parameter of wind power generation based on the maximum wind power generation from the new fitted data. Optimize; Based on the new fitted data and the fitted function with updated parameters, the shortest distance is recalculated and clustered, and this process is iteratively updated until a stable threshold is reached; wherein, the stable threshold is based on the kurtosis parameter in two consecutive iterations. or inflection point parameters The absolute value of the difference is set.

8. A wind energy data quality control device based on a clustering algorithm, characterized in that, The device includes: The data acquisition module is used to collect wind energy data from electric field wind measurement towers or meteorological wind measurement towers, and to preprocess the collected wind energy data to obtain sample points; the wind energy data includes wind speed and corresponding wind power generation. The initialization module is used to determine the fitting function based on the S-curve characteristics between wind speed and wind power generation, and to initialize its parameters. The clustering module is used to calculate the shortest distance from each sample point to the fitted function, thereby obtaining a distance set; and based on the Gaussian mixture model, it clusters the distance sets of all sample points to obtain normal data clusters and abnormal data clusters. The update module is used to iteratively update the parameters of the fitting function using the sample points of the normal data cluster until a stable threshold is reached. The sample points of the final normal data cluster are used as wind energy data after quality control processing for wind turbine performance monitoring.

9. An electronic device, characterized in that, include: The device includes a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, they perform the steps of a wind energy data quality control method based on a clustering algorithm as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of a wind energy data quality control method based on a clustering algorithm as described in any one of claims 1 to 7.

Images