A method for identifying power grid faults for wind turbines

Abstract

The embodiment of the present application discloses a method for distinguishing power grid faults for wind turbines, and relates to the technical field of power grid fault detection. The method includes: obtaining the state measurement parameters of the power grid where the target wind turbine is located in the target time series; determining the original regression coefficient and the original basic value; determining the state measurement parameter error based on the original regression coefficient and the original basic value; obtaining the first basic value and the first regression coefficient with the state measurement parameter error minimized as the optimization target; obtaining the first discrimination data according to the first regression coefficient, the first basic value and the corresponding state measurement parameter error; performing feature mapping conversion on the target sampling time in the subsequence to obtain the corresponding local feature data; fusing the local feature data corresponding to multiple subsequences to obtain fused data; performing regression analysis on the fused data to obtain the overall feature data; performing fault discrimination processing based on the overall feature data to obtain the corresponding predicted fault type and probability.

Description

Technical Field

[0001] The present application relates to the technical field of power grid fault detection, and in particular to a power grid fault identification method for a wind turbine generator system. Background Art

[0002] As an important part of renewable energy, wind power generation has developed rapidly in recent years. With the continuous expansion of wind power generation and the increase in the capacity of wind turbines, the stability of wind turbine grid-connected operation has become particularly important. In order to ensure the stable operation of the power grid, wind turbines must have the ability to cope with grid failures.

[0003] Traditional grid fault detection methods often have problems such as lag and low accuracy, which are difficult to meet the requirements of modern wind farms for fault response speed and accuracy. Therefore, studying a fast and accurate grid fault identification method has become an important topic in the field of wind power technology. Summary of the invention

[0004] To solve the above problems, the present invention provides a method for determining a power grid fault for a wind turbine generator system. The method includes:

[0005] Acquire state measurement parameters of the power grid where the target wind turbine is located in the target time series, wherein the state measurement parameters include power grid load data, wind turbine power generation data, power grid topology data, power grid voltage and frequency data, and the target time series includes multiple subsequences, each of which includes multiple sampling moments;

[0006] Determine the original regression coefficient and the original base value corresponding to each state metric parameter in the subsequence;

[0007] Determine a state metric parameter error based on the state metric parameter corresponding to each of the sampling moments and the original regression coefficient and the original base value corresponding to the state metric parameter;

[0008] The original base value and the original regression coefficient are updated with the goal of minimizing the state metric parameter error to obtain a first base value and a first regression coefficient;

[0009] Obtaining first discriminant data according to the first regression coefficient, the first basic value, and a first state metric parameter error of the state metric parameter;

[0010] Performing feature mapping conversion on the target sampling moments in the subsequences according to the first discriminant data to obtain local feature data corresponding to each of the subsequences;

[0011] fusing the local feature data corresponding to the plurality of subsequences to obtain fused data, and performing regression analysis on the fused data to obtain the overall feature data of the target wind turbine group;

[0012] Fault identification processing is performed based on the overall characteristic data of the target wind turbine generator set to obtain the predicted fault type and probability corresponding to the power grid where the target wind turbine generator set is located.

[0013] In some embodiments, the method further comprises:

[0014] An active support control strategy is determined according to the predicted fault type and probability corresponding to the power grid where the target wind turbine is located. The active support control strategy includes at least one control instruction for adjusting the operating state of the target wind turbine to adapt to the predicted fault type.

[0015] In some embodiments, the determining of the active support control strategy according to the predicted fault type and probability corresponding to the power grid where the target wind turbine is located includes:

[0016] When the probability is greater than a first preset threshold, obtaining current power generation data of the target wind turbine group, wherein the current power generation data includes power output;

[0017] Analyzing the degree of influence correlation between the predicted fault type and the current power generation data to determine the parameters to be adjusted;

[0018] Corresponding control instructions are generated according to the parameters to be adjusted to achieve active support control of the target wind turbine group.

[0019] In some embodiments, the step of obtaining a state measurement parameter of a power grid in which a target wind turbine generator set is located in a target time series includes:

[0020] For each of the subsequences in the target time series, obtaining original state measurement parameters of the target wind turbine group in the subsequence;

[0021] Preprocessing the original state metric parameters to obtain preprocessed state metric parameters corresponding to each of the subsequences, wherein the preprocessing includes data cleaning and compression integration;

[0022] Performing preset rule conversion on the preprocessed state measurement parameters to obtain the state measurement parameters corresponding to each of the subsequences;

[0023] Based on the state measurement parameters corresponding to the multiple subsequences, the state measurement parameters of the power grid where the target wind turbine group is located in the target time sequence are obtained.

[0024] In some embodiments, the obtaining of original state measurement parameters of the target wind turbine group in the subsequence includes:

[0025] For each sampling moment in the subsequence, obtaining grid load data, wind turbine generation data, grid topology data, grid voltage and frequency data of the grid where the target wind turbine is located at the sampling moment;

[0026] The grid load data, wind turbine generation data, grid topology data, grid voltage and frequency data are combined into the original state measurement parameters of the grid where the target wind turbine is located at the sampling time;

[0027] Based on the original state metric parameters corresponding to the target wind turbine group at the plurality of sampling moments, the original state metric parameters of the power grid where the target wind turbine group is located in the subsequence are obtained.

[0028] In some embodiments, the preprocessing includes compressing and integrating the data obtained after data cleaning to obtain compression integration parameters, wherein the compression integration parameters include compression integration parameters corresponding to each sampling moment in the subsequence at multiple feature dimensions;

[0029] The performing of preset rule conversion on the preprocessed state metric parameters to obtain the state metric parameters corresponding to each of the subsequences includes:

[0030] For each compression integration parameter under each feature dimension, obtaining a maximum parameter and a minimum parameter among the compression integration parameters under the feature dimension at multiple sampling moments;

[0031] Calculating a first parameter difference between the maximum parameter and the minimum parameter, and a second parameter difference between the compression integration parameter of the feature dimension corresponding to each sampling moment and the minimum parameter;

[0032] Taking the ratio between the second parameter difference corresponding to each sampling moment and the first parameter difference as the normalized state measurement parameter corresponding to the feature dimension at the sampling moment;

[0033] Obtaining the state measurement parameter corresponding to the sampling moment according to the normalized state measurement parameters in the plurality of feature dimensions at the sampling moment;

[0034] The state metric parameter corresponding to each of the subsequences is obtained based on the state metric parameters corresponding to the multiple sampling moments.

[0035] In some embodiments, determining the state metric parameter error based on the state metric parameter corresponding to each sampling moment and the original regression coefficient and the original base value corresponding to the state metric parameter includes:

[0036] For each of the sampling moments, mapping conversion is performed on the sampling moments based on the original base value of the state metric parameter and the original regression coefficient of the state metric parameter to obtain a first mapping conversion result;

[0037] When the absolute value of the difference between the first mapping conversion result and the state metric parameter at the sampling moment is less than or equal to a second preset threshold, determining a sampling moment error corresponding to the sampling moment based on the absolute value of the difference;

[0038] When the absolute value of the difference between the first mapping conversion result and the state metric parameter at the sampling moment is greater than the second preset threshold, determining a sampling moment error corresponding to the sampling moment based on the absolute value of the difference and the second preset threshold;

[0039] The sampling moment errors corresponding to the multiple sampling moments are summed to obtain the state metric parameter error.

[0040] In some embodiments, the target sampling moment includes a first target moment that is the frontmost and a second target moment that is the backmost in the subsequence, and performing feature mapping conversion on the target sampling moments in the subsequence according to the first discrimination data to obtain local feature data corresponding to each of the subsequences includes:

[0041] Adding a first product of a first regression coefficient corresponding to the first discriminant data and a first target moment in the subsequence to the first base value to obtain local feature data corresponding to the first target moment;

[0042] Adding a second product of a first regression coefficient corresponding to the first discriminant data and a second target moment in the subsequence to the first base value to obtain local feature data corresponding to the second target moment;

[0043] Based on the local feature data corresponding to the first target moment and the local feature data corresponding to the second target moment, the local feature data corresponding to each of the subsequences is obtained.

[0044] In some embodiments, performing regression analysis on the fused data to obtain overall characteristic data of the target wind turbine generator set includes:

[0045] Performing a univariate linear regression analysis on the fused data to obtain a second regression coefficient and a second base value corresponding to the state measurement parameter;

[0046] Obtain second discriminant data according to the second regression coefficient, the second basic value, and a second state metric parameter error of the state metric parameter;

[0047] A feature mapping conversion is performed according to the second discriminant data to obtain overall feature data of the target time series.

[0048] In some embodiments, the fault identification processing is performed based on the overall characteristic data of the target wind turbine generator set to obtain the predicted fault type and probability corresponding to the power grid where the target wind turbine generator set is located, including:

[0049] The overall characteristic data of the target wind turbine is input into a trained machine learning model for processing to obtain the predicted fault type and probability corresponding to the power grid where the target wind turbine is located; wherein the machine learning model is trained in the following manner:

[0050] Acquire historical fault data and its corresponding fault types as training samples, wherein the historical fault data includes overall characteristic data of the power grid in which the wind turbine is located under a fault state, and the fault type includes at least one of voltage drop, voltage rise, voltage interruption, frequency deviation, harmonic distortion, three-phase imbalance, overvoltage and overcurrent;

[0051] The fault type corresponding to the historical fault data is used as a sample label, and the initial machine learning model is iteratively trained based on the historical fault data and the sample label, and the training is terminated until the preset training conditions are met to obtain the trained machine learning model, wherein the machine learning model includes one or more of a random forest, a support vector machine, and a neural network model.

[0052] The beneficial effects of the power grid fault identification method for wind turbines provided in the embodiments of this specification may include at least:

[0053] (1) By performing a secondary regression analysis on the state measurement parameters corresponding to the power grid where the target wind turbine is located, the data of each characteristic dimension can be fitted with high accuracy, thereby improving the accuracy and reliability of power grid fault identification and prediction results;

[0054] (2) By implementing the preset rule conversion to normalize all compression integration parameters to roughly the same level, the local characteristic data and overall characteristic data corresponding to the target wind turbine can be fitted more quickly and accurately in the subsequent process;

[0055] (3) By using the state measurement parameters detected or statistically obtained for the power grid where the target wind turbine is located, fault identification and prediction of the power grid can be achieved, and active support control can be implemented based on this. The operating state of the wind turbine can be adjusted in time when the power grid fails or is about to fail, thereby reducing the impact of the fault on the power grid and improving the stability and reliability of the entire system.

[0056] Additional features will be described in part in the following description. For those skilled in the art, it will become apparent by reviewing the following and accompanying drawings, or it may be understood by the production or operation of the examples. The features of this specification may be implemented and obtained by practicing or using various aspects of the methods, tools, and combinations described in the following detailed examples. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] This specification will be further described in the form of exemplary embodiments, which will be described in detail by the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same number represents the same structure, wherein:

[0058] Figure 1 is a schematic diagram of an exemplary application scenario of a power grid fault identification system for a wind turbine generator set according to some embodiments of this specification;

[0059] Figure 2 is an exemplary flow chart of a method for determining a power grid fault of a wind turbine generator system according to some embodiments of this specification;

[0060] Figure 3 is a flowchart of an exemplary process for obtaining state metric parameters according to some embodiments of the present specification. DETAILED DESCRIPTION

[0061] In order to more clearly illustrate the technical solutions of the embodiments of this specification, the following is a brief introduction to the drawings required for the description of the embodiments. Obviously, the drawings described below are only some examples or embodiments of this specification. For ordinary technicians in this field, this specification can also be applied to other similar scenarios based on these drawings without creative work. Unless it is obvious from the language environment or otherwise explained, the same reference numerals in the figures represent the same structure or operation.

[0062] It should be understood that the "system", "device", "unit" and / or "module" used in this specification is a method for distinguishing different components, elements, parts, parts or assemblies at different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.

[0063] As shown in this specification and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "comprise" and "include" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0064] Flowcharts are used in this specification to illustrate the operations performed by the system according to the embodiments of this specification. It should be understood that the preceding or following operations are not necessarily performed precisely in order. Instead, the steps may be processed in reverse order or simultaneously. At the same time, other operations may also be added to these processes, or one or more operations may be removed from these processes.

[0065] The identification of power grid faults is usually based on preset thresholds. This method is to detect the operating status of the power grid by setting a series of thresholds. Specifically, certain key parameters of the power grid (such as voltage, current, frequency, etc.) are compared with the preset thresholds. If these parameters exceed the normal range, it is considered that the power grid has a fault.

[0066] The advantage of threshold judgment is that it is simple, intuitive and easy to implement. However, this method also has certain limitations. For example, for some complex fault conditions, a single threshold judgment may not accurately identify the fault. In addition, the selection of threshold is also a key issue. If the threshold is set too high or too low, it may lead to misjudgment or missed judgment, which will not only affect the effectiveness of the active support control strategy, but also affect the timeliness of active support control.

[0067] In view of the above problems, the embodiment of the present application provides a method for distinguishing grid faults for wind turbines, which distinguishes and predicts grid faults by using state measurement parameters detected or statistically obtained for the grid where the target wind turbine is located, and implements active support control based on the state measurement parameters, thereby improving the stability of grid operation. The method for distinguishing grid faults for wind turbines provided by the embodiment of the present application is described in detail below in conjunction with the accompanying drawings.

[0068] Figure 1 1 is a schematic diagram of an exemplary application scenario of a power grid fault identification system for a wind turbine generator set according to some embodiments of this specification. Figure 1 , the system 100 may include a terminal device 110, a state measurement parameter acquisition device 120, a processing device 130, and a network 140. The terminal device 110 may be connected to the processing device 130 via the network 140, wherein the network 140 may include any suitable network capable of realizing information and / or data exchange, and as an example only, the network 140 may include a cable network, a wired network, an optical fiber network, a telecommunication network, an intranet, the Internet, a local area network (LAN), a wide area network (WAN), a wireless local area network (WLAN), a metropolitan area network (MAN), a public switched telephone network (PSTN), a Bluetooth network, a ZigBee network, a near field communication (NFC) network, etc. or any combination thereof.

[0069] In some embodiments, the processing device 130 may refer to a server corresponding to an application. Specifically, the application refers to the power grid fault detection software installed in the terminal device 110, and the processing device 130 may be understood as a power grid fault detection server, which is used to process the state measurement parameters detected or statistically obtained for the power grid where the target wind turbine is located to achieve fault identification and prediction of the power grid, and feed back the fault identification result (or prediction result) to the terminal device 110 for display. In an embodiment of the present application, the terminal device 110 may include a mobile device (such as a mobile phone, a personal digital assistant (PDA), a dedicated mobile terminal, etc.), a tablet computer, a laptop computer, etc. or any combination thereof. In some embodiments, the processing device 130 may be a part of the terminal device 110.

[0070] In some embodiments, the terminal device 110 can respond to user operations or generate a fault identification request for the power grid where any target wind turbine is located according to a preset program, and send the fault identification request to the processing device 130. The processing device 130 can obtain the state measurement parameters of the power grid and the target wind turbine included in the power grid in the target time series (such as power grid load data, wind turbine power generation data, power grid topology data, power grid voltage and frequency data, etc.) through the state measurement parameter acquisition device 120, and then process the state measurement parameters to obtain local feature data corresponding to each subsequence; further, the local feature data of multiple subsequences contained in the target time series can be fused to obtain fused data, and the fused data can be regression analyzed to obtain the overall feature data of the target wind turbine; finally, the processing device 130 can perform fault identification processing based on the overall feature data of the target wind turbine to obtain the predicted fault type and probability corresponding to the power grid where the target wind turbine is located.

[0071] It should be noted that, in the embodiment of the present application, the aforementioned state measurement parameter acquisition device 120 may include various monitoring terminals (such as sensors) and / or software devices for statistically analyzing data detected by the monitoring terminals, etc. For example, in some embodiments, various sensors may be used to detect grid load data, wind turbine generation data, grid topology data, grid voltage and frequency data. In some embodiments, statistical analysis may also be performed on the data detected by the sensors to obtain data such as grid load change trends, power generation data change trends, voltage change trends, and frequency change trends.

[0072] It should also be noted that in the embodiment of the present application, each subsequence can be understood as a time series corresponding to a time period (i.e., a subinterval) contained in the target time series, and the state measurement parameters in the multiple subsequences contained in the aforementioned target time series can be understood as multiple discrete points, the horizontal coordinates of the multiple discrete points in the coordinate axis are the sampling moments contained in the subsequence, and the vertical coordinates are the state measurement parameters corresponding to each sampling moment.

[0073] Figure 2 is an exemplary flow chart of a method for determining a power grid fault for a wind turbine generator set according to some embodiments of this specification. In some embodiments, the method 200 for determining a power grid fault for a wind turbine generator set can be executed by processing logic, which can include hardware (e.g., circuits, dedicated logic, programmable logic, microcode, etc.), software (instructions running on a processing device to perform hardware simulation), etc., or any combination thereof. In some embodiments, Figure 2 One or more operations in the flowchart of the grid fault determination method 200 for wind turbines can be performed by Figure 1 The processing device 130 shown is implemented. For example, the power grid fault determination method 200 for wind turbines can be stored in a storage device in the form of instructions or computer programs, and called and / or executed by the processing device 130. The following describes the execution process of the power grid fault determination method 200 for wind turbines provided in the embodiment of the present application with the processing device 130 as the execution subject.

[0074] Reference Figure 2 In the embodiment of the present application, the grid fault determination method 200 for a wind turbine generator system may include:

[0075] Step S210, obtaining state measurement parameters of the power grid where the target wind turbine is located in the target time series, wherein the state measurement parameters include power grid load data, wind turbine power generation data, power grid topology data, power grid voltage and frequency data, and the target time series includes multiple subsequences, each of which includes multiple sampling moments.

[0076] In the embodiment of the present application, the processing device 130 can obtain the state measurement parameters of the power grid where the target wind turbine is located in the target time series after receiving the fault identification request sent by the terminal device 110. The state measurement parameters include power grid load data, wind turbine power generation data, power grid topology data, power grid voltage and frequency data, etc. The state measurement parameters can be collected by the state measurement parameter acquisition device 120 and stored in the storage device, and the processing device 130 can obtain the state measurement parameters from the storage device.

[0077] Specifically, in the embodiments of the present application, the grid load data may include historical load, peak load, load change trend, etc., which may be obtained through smart meters, SCADA systems, and power management systems. Wind turbine power generation data may include power output, predicted power generation, etc., which may be obtained through the real-time monitoring system of the wind turbine. Grid topology data may include the configuration and status of grid lines, transformers, switches and other equipment, which may be obtained through monitoring terminals such as sensors. Grid voltage and frequency data may be collected in real time in the grid through sensors and / or monitoring equipment.

[0078] In an embodiment of the present application, the aforementioned target time series may include multiple subsequences, wherein each of the subsequences may include multiple sampling moments. For example, in some embodiments, the target time series may be represented as X = {x1, x2, …, x n}. Among them, x1, x2, …, x n is the n subsequences contained in the target time series X. Exemplarily, in some embodiments, the subsequence can be expressed as: x1={(t1, c1, c2,……,c n ), (t2, c1,c2,……,c n ),……, (t m , c1, c2,……,c n )};x2={(t m+1 , c1, c2,……,c n ), (t m+2 , c1,c2,……,c n ),……, (t 2m , c1, c2,……,c n )}; ... Among them, t1, t2, t m ,t m+1 ,t m+2 ,t 2m Respectively represent the 1st, 2nd, mth, m+1st, m+2nd, 2mth sampling moments. c1, c2,……,c n Respectively represent the 1st to nth state measurement parameters collected at each sampling moment. As an example only, in some embodiments, c1 may correspond to grid load data, c2 may correspond to wind turbine generation data, c3 may correspond to grid topology data, c4 may correspond to grid voltage data, c5 may correspond to grid frequency data... In some embodiments, c1, c2,...,c nEach state measurement parameter in may include one or more sub-parameters. For example, c1 may include historical load, peak load, load change trend, etc. The multiple sub-parameters may be arranged according to preset rules to obtain a state measurement parameter in the subsequence. Other state measurement parameters are similar and will not be described in detail in this specification.

[0079] Reference Figure 3 In some embodiments, the process of obtaining the state measurement parameters of the power grid where the target wind turbine is located in the target time series may include the following steps:

[0080] Step S2101: for each subsequence in the target time series, obtain the original state measurement parameters of the target wind turbine in the subsequence.

[0081] Step S2102, preprocessing the original state measurement parameters to obtain preprocessed state measurement parameters corresponding to each of the subsequences, wherein the preprocessing includes data cleaning and compression integration.

[0082] Step S2103, performing a preset rule conversion on the preprocessed state metric parameters to obtain a state metric parameter corresponding to each of the subsequences.

[0083] Step S2104: obtaining the state measurement parameters of the power grid where the target wind turbine is located in the target time sequence based on the state measurement parameters corresponding to the multiple subsequences.

[0084] As an example only, in step S2101, the state measurement parameter collection can be performed once every 2 minutes, the period corresponding to the target time series can be 1 day (24 hours), and the time length corresponding to each subsequence can be 2 hours, then the target time series contains 12 subsequences, and each subsequence includes 60 sampling moments. It should be noted that the length of the above target time series and subsequences is only for example. In the embodiment of the present application, the collection frequency of the state measurement parameter and the length of the target time series and subsequence can be other values.

[0085] In an embodiment of the present application, obtaining the original state measurement parameters of the target wind turbine in a subsequence can be understood as: obtaining the original state measurement parameters (i.e., unprocessed state measurement data) of the power grid in which the target wind turbine is located, collected by sensors or a data monitoring system at the 60 sampling moments contained in each subsequence.

[0086] Specifically, in an embodiment of the present application, the processing device 130 can obtain, for each sampling moment in the subsequence, the grid load data, wind turbine power generation data, grid topology data, grid voltage and frequency data of the power grid where the target wind turbine is located at each sampling moment; then, the grid load data, wind turbine power generation data, grid topology data, grid voltage and frequency data are combined to form the original state measurement parameters of the power grid where the target wind turbine is located at the sampling moment; finally, based on the original state measurement parameters corresponding to the target wind turbine at multiple sampling moments, the original state measurement parameters of the power grid where the target wind turbine is located in each subsequence are obtained.

[0087] In step S2102, the preprocessing may include data cleaning and compression integration. Specifically, in the actual application process, the original state measurement parameters collected may have null values, data anomalies, and duplication (for example, data collection is performed twice at the same sampling time). Therefore, in order to ensure the accuracy and effectiveness of the data, the original state measurement parameters need to be cleaned first.

[0088] Specifically, in some embodiments, the original state metric parameters can be detected by the LOF (Local Outlier Factor) algorithm, and the original state metric parameters of the target wind turbine group at each sampling moment in the subsequence that are null values ​​and abnormal values ​​can be removed; further, the original state metric parameters at each sampling moment in the retained original state metric parameters can be subjected to redundant filtering to obtain the cleaned state metric parameters corresponding to each subsequence. It can be understood that in the embodiments of the present application, by cleaning the original state metric parameters and removing abnormal values, null values ​​and redundant values ​​therein, the characterization ability of the original state metric parameters can be improved to a certain extent, thereby improving the accuracy and reliability of subsequent power grid fault identification and prediction.

[0089] Furthermore, the cleaned state metric parameters can be compressed and integrated to obtain the compressed integration parameters corresponding to each subsequence. In the embodiment of the present application, the compression and integration process can be understood as integrating the cleaned data into a unified data structure, aiming to reduce the amount of data, improve storage efficiency and data transmission speed, so as to facilitate subsequent data analysis and processing. As an example only, in some embodiments, the row data can be converted into column data through matrix transposition and other processes to achieve data compression and integration, thereby improving computing performance. For example, in some embodiments, the subsequence x1 can be compressed and integrated as follows:

[0090]

[0091] represents the compressed and integrated subsequence x1. After the above compression and integration, the preprocessed data can be obtained. Further, in step S2103, the processing device 130 can perform a preset rule conversion on the state measurement parameters after the above preprocessing to obtain the state measurement parameters corresponding to each of the subsequences.

[0092] In an embodiment of the present application, the aforementioned compression integration parameters may include compression integration parameters corresponding to multiple feature dimensions at each sampling moment in the subsequence. For the compression integration parameters under each feature dimension, the maximum parameter and the minimum parameter of the compression integration parameters under the feature dimension at multiple sampling moments can be obtained; then the first parameter difference between the maximum parameter and the minimum parameter, and the second parameter difference between the compression integration parameter of the feature dimension corresponding to each sampling moment and the minimum parameter are calculated; further, the processing device 130 can use the ratio between the second parameter difference corresponding to each sampling moment and the first parameter difference as the normalized state measurement parameter corresponding to the feature dimension at the sampling moment; finally, the state measurement parameter corresponding to the sampling moment is obtained according to the normalized state measurement parameters at the sampling moment in multiple feature dimensions, and the state measurement parameter corresponding to each subsequence is obtained based on the state measurement parameters corresponding to multiple sampling moments.

[0093] For example, in some embodiments, ( , ,……, ) is regarded as the compression integration parameter under the first feature dimension (similarly, ( , ,……, ) is regarded as the compression integration parameter under the second feature dimension). Further, the first parameter difference between the maximum parameter and the minimum parameter in the compression integration parameter under the first feature dimension can be calculated. The calculation process of the first parameter difference can be expressed as follows:

[0094]

[0095] in, Represents the first parameter difference corresponding to the first feature dimension, represents the maximum parameter under the first feature dimension, Indicates the minimum parameter under the first feature dimension.

[0096] Furthermore, the second parameter difference between the compressed integration parameter of the feature dimension corresponding to each sampling moment and the minimum parameter can be calculated, and the ratio between the second parameter difference corresponding to each sampling moment and the first parameter difference can be used as the normalized state measurement parameter corresponding to the feature dimension at the sampling moment.

[0097] Specifically, in the embodiment of the present application, the normalized state measurement parameter corresponding to the first feature dimension can be expressed as:

[0098]

[0099] It can be understood that, through the above method, the normalized state measurement parameter corresponding to each feature dimension can be obtained, which will not be described in detail in this specification.

[0100] In an embodiment of the present application, the state measurement parameters corresponding to each sampling moment can be obtained based on the normalized state measurement parameters in multiple feature dimensions at each sampling moment, and then the state measurement parameters corresponding to each of the subsequences can be obtained based on the state measurement parameters corresponding to the multiple sampling moments.

[0101] Exemplarily, in some embodiments, the state metric parameter corresponding to the first sampling moment can be expressed as:

[0102]

[0103] The state measurement parameter corresponding to the second sampling time can be expressed as:

[0104]

[0105] Furthermore, by combining the state metric parameters corresponding to the multiple sampling moments contained in each subsequence, the state metric parameters corresponding to each subsequence can be obtained. Furthermore, by combining the state metric parameters corresponding to multiple subsequences, the state metric parameters of the power grid where the target wind turbine is located in the target time series can be obtained.

[0106] It should be pointed out that in the embodiment of the present application, all compression integration parameters are normalized to approximately the same level through the above-mentioned preset rule conversion, which can make the local characteristic data and overall characteristic data corresponding to the target wind turbine group fit more quickly and accurately in the subsequent process.

[0107] Continue to refer to Figure 2 , the grid fault determination method 200 for wind turbines further includes:

[0108] Step S220: determining the original regression coefficient and the original base value corresponding to each state metric parameter in the subsequence.

[0109] In the embodiment of the present application, the regression coefficient can be understood as a coefficient for describing the relationship between the characteristic parameter and the state metric. Specifically, the original regression coefficient can be obtained by statistical analysis methods based on historical data, such as least squares method, ridge regression or LASSO regression. These regression coefficients can reflect the contribution of different characteristic dimensions to the state metric parameters, thereby providing a basis for subsequent prediction and judgment. In some embodiments, each characteristic dimension can correspond to an original regression coefficient.

[0110] In the embodiment of the present application, the original basic value can be understood as the normal level of the aforementioned state measurement parameter when no power grid fault occurs. In some embodiments, the basic value of each feature dimension under normal operating conditions can be determined through statistical analysis of historical data. These basic values ​​can provide a reference standard for subsequent fault discrimination, so that in actual operation, once the state measurement parameter deviates from the normal level, the potential power grid fault can be quickly identified.

[0111] Step S230: determining the state metric parameter error based on the state metric parameter corresponding to each of the sampling moments and the original regression coefficient and the original base value corresponding to the state metric parameter.

[0112] In an embodiment of the present application, the state measurement parameter error can be understood as the difference between the state measurement parameter determined based on the regression coefficient (such as the original regression coefficient or the first regression coefficient involved in the subsequent process) and the base value (such as the original base value or the first base value involved in the subsequent process) and the actual state measurement parameter. This difference can reflect the fitting accuracy and prediction accuracy of the regression analysis.

[0113] Specifically, in some embodiments, for each sampling moment, the sampling moment can be mapped and converted based on the original base value of the state metric parameter and the original regression coefficient of the state metric parameter to obtain a first mapping conversion result. In some embodiments, the mapping conversion process can be a unary function operation. In other words, in some embodiments, the original regression coefficient can be multiplied by the state metric parameter corresponding to each sampling moment, and then the original initial value is added to obtain the corresponding first mapping conversion result.

[0114] Further, if the absolute value of the difference between the first mapping conversion result and the state measurement parameter corresponding to the sampling moment is less than or equal to the second preset threshold, the sampling moment error corresponding to the sampling moment is determined based on the absolute value of the difference; if the absolute value of the difference between the first mapping conversion result and the state measurement parameter corresponding to the sampling moment is greater than the second preset threshold, the sampling moment error corresponding to the sampling moment is determined based on the absolute value of the difference and the second preset threshold. The process can be expressed by the calculation formula as follows:

[0115]

[0116] in, represents the sampling time error corresponding to a certain sampling time; f represents the state measurement parameter corresponding to a certain sampling time, Indicates the first mapping conversion result corresponding to a certain sampling moment; Represents a second preset threshold. In the embodiment of the present application, the second preset threshold can be set or adjusted according to actual needs and is not limited in this specification.

[0117] After obtaining the sampling time error corresponding to each sampling time, the state metric parameter error corresponding to the target time series or each subsequence can be obtained by summing the sampling time errors corresponding to multiple sampling times.

[0118] Step S240, updating the original base value and the original regression coefficient with the goal of minimizing the state metric parameter error to obtain a first base value and a first regression coefficient.

[0119] In an embodiment of the present application, the original base value and the original regression coefficient may be updated with the goal of minimizing the error of the aforementioned state metric parameter to obtain a first base value and a first regression coefficient.

[0120] Specifically, in some embodiments, the original regression coefficient and the original base value may be iteratively updated by an optimization algorithm (such as a gradient descent method, a Newton method, etc.). In each iteration, the values ​​of the regression coefficient and the base value are adjusted according to the size of the state metric parameter error to minimize the state metric parameter error, thereby obtaining the first base value and the first regression coefficient.

[0121] Step S250: obtaining first discrimination data according to the first regression coefficient, the first basic value and a first state metric parameter error of the state metric parameter.

[0122] Specifically, in the embodiment of the present application, the regression coefficient corresponding to the state metric parameter error being minimized can be used as the first regression coefficient, and the basic value corresponding to the state metric parameter error being minimized can be used as the first basic value. Further, the first regression coefficient and the first basic value corresponding to each feature dimension and the first state metric parameter error corresponding to the state metric parameter under the first regression coefficient and the first basic value can be combined to obtain the first discriminant data corresponding to each feature dimension.

[0123] Step S260: performing feature mapping conversion on the target sampling moments in the subsequences according to the first discrimination data to obtain local feature data corresponding to each of the subsequences.

[0124] After obtaining the first discriminant data corresponding to each feature dimension through the above steps, feature mapping conversion can be performed on the target sampling moments in the subsequence according to the first discriminant data to obtain local feature data corresponding to each subsequence. The target sampling moments include the first target moment at the front and the second target moment at the back in the subsequence.

[0125] Specifically, in some embodiments, the first product of the first regression coefficient corresponding to the first discriminant data and the first target moment in the subsequence can be added to the first basic value to obtain local feature data corresponding to the first target moment; then, the second product of the first regression coefficient corresponding to the first discriminant data and the second target moment in the subsequence can be added to the first basic value to obtain local feature data corresponding to the second target moment; finally, based on the local feature data corresponding to the first target moment and the local feature data corresponding to the second target moment, the local feature data corresponding to each of the subsequences is obtained.

[0126] In some embodiments, the determination process of the local feature data can be expressed as follows:

[0127]

[0128]

[0129] in, Represents the local feature data corresponding to the nth feature dimension at the first target moment; Represents the local feature data corresponding to the nth feature dimension at the second target time (i.e., the mth sampling time, which is the last sampling time in the first subsequence); represents the first regression coefficient corresponding to the nth feature dimension, Represents the first basic value corresponding to the nth feature dimension. Similarly, based on the above method, the local feature data corresponding to the first target time and the second target time of all feature dimensions can be obtained.

[0130] Furthermore, based on the local feature data corresponding to the first target moment and the local feature data corresponding to the second target moment, the local feature data corresponding to each of the subsequences can be obtained. Taking the first subsequence as an example, the local feature data corresponding to the subsequence can be expressed as {( , ),……,( , )}.

[0131] Step S270: fusing the local feature data corresponding to the plurality of subsequences to obtain fused data, and performing regression analysis on the fused data to obtain the overall feature data of the target wind turbine group.

[0132] After obtaining the local feature data corresponding to each subsequence through the above steps, the local feature data corresponding to the multiple subsequences can be fused to obtain fused data. The fused data can be expressed as {( , ),……,( , );( , );……}.

[0133] Furthermore, the processing device 130 may perform regression analysis on the fused data to obtain overall characteristic data of the target wind turbine group.

[0134] In some embodiments, a univariate linear regression analysis may be performed on the fused data to obtain a second regression coefficient and a second base value corresponding to the state measurement parameter; then, second discriminant data may be obtained based on the second regression coefficient, the second base value, and the second state measurement parameter error of the state measurement parameter; finally, a feature mapping conversion may be performed based on the second discriminant data to obtain the overall feature data of the target time series.

[0135] As an example only, in some embodiments, in order to capture the time series characteristics of the state measurement parameters of the target wind turbine within 24 hours, a secondary regression analysis can be performed on the local feature data of multiple subsequences for 24 consecutive hours to obtain the overall feature data corresponding to the target wind turbine.

[0136] In the embodiment of the present application, the local feature data of multiple subsequences can be fused first to obtain fused data, specifically, a secondary regression analysis can be performed on the fused data corresponding to the next 24 hours. It should be noted that the fused data obtained by the above-mentioned processing fusion can compress and integrate the original state measurement parameters, thereby greatly reducing the number of features while retaining the feature information, thereby reducing the calculation pressure of the processing device 130.

[0137] In some embodiments, the processing device 130 performs a secondary regression analysis on the fused data to obtain the overall characteristic data corresponding to the target wind turbine group. The specific process can refer to the following steps:

[0138] First, a univariate linear regression analysis can be performed on the fused data to obtain the second regression coefficient and the second base value corresponding to the fused state measurement parameter. Specifically, in some embodiments, the original regression coefficient and the original base value corresponding to the fused data can be determined first (the original regression coefficient and the original base value here are obtained by fitting analysis based on the fused data); then, based on the fused data and the original base value and the original regression coefficient corresponding to the fused data, the error of the second state measurement parameter is determined; and the original base value and the original regression coefficient corresponding to the fused data are updated with the goal of minimizing the error of the second state measurement parameter to obtain the second base value and the second regression coefficient.

[0139] Further, the second discriminant data can be obtained according to the second regression coefficient, the second base value and the second state metric parameter error of the state metric parameter. As an example only, in some embodiments, the second regression coefficient, the second base value corresponding to each feature dimension in the fused data and the second state metric parameter error corresponding to the fused data under the second regression coefficient and the second base value can be combined to obtain the second discriminant data corresponding to each feature dimension.

[0140] After obtaining the second discriminant data corresponding to each feature dimension, feature mapping conversion can be performed on it, so as to obtain the overall feature data corresponding to the aforementioned target time series. Specifically, in some embodiments, the process of performing feature mapping conversion on the second discriminant data can be expressed as follows:

[0141]

[0142]

[0143] in, Represents the feature data corresponding to the nth feature dimension of the fused data at the first sampling time; Indicates the feature data corresponding to the nth feature dimension of the fused data at the last sampling moment (i.e., the sth sampling moment, which is the last sampling moment in all subsequences); Represents the second regression coefficient corresponding to the nth feature dimension of the fused data, represents the second basic value corresponding to the nth feature dimension of the fused data. Based on the above method, the feature data corresponding to all feature dimensions of the fused data at the first sampling time and the last sampling time can be obtained. Further, the overall feature data corresponding to the target time series can be expressed as Q={( , ),( , );……;( , )}.

[0144] Step S280 , performing fault identification processing based on the overall characteristic data of the target wind turbine generator set to obtain a predicted fault type and probability corresponding to the power grid where the target wind turbine generator set is located.

[0145] After the overall characteristic data corresponding to the target wind turbine group is obtained through the above steps, a fault discrimination process can be performed based on the overall characteristic data to obtain the predicted fault type and probability corresponding to the power grid where the target wind turbine group is located. Specifically, in some embodiments of the present application, the overall characteristic data of the target wind turbine group can be input into a trained machine learning model for processing, thereby obtaining the predicted fault type and probability corresponding to the power grid where the target wind turbine group is located.

[0146] Specifically, in an embodiment of the present application, the machine learning module may include one or more of a random forest, a support vector machine, and a neural network model. The machine learning model may be trained in the following manner:

[0147] First, historical fault data and their corresponding fault types are obtained as training samples, wherein the historical fault data include overall characteristic data of the power grid in which the wind turbine is located under a fault state (the overall characteristic data can be obtained based on the method as described above), and the fault type may include at least one of voltage drop, voltage rise, voltage interruption, frequency deviation, harmonic distortion, three-phase imbalance, overvoltage and overcurrent.

[0148] Then, the fault type corresponding to the historical fault data is used as a sample label, and the initial machine learning model is iteratively trained based on the historical fault data and the sample label until the training is terminated when the preset training conditions are met, thereby obtaining the trained machine learning model. Other technical details about model training can be regarded as prior art and will not be discussed in detail in this specification.

[0149] As an example only, in some embodiments of the present application, a fault case database can also be established, which can store a number of historical fault cases and their corresponding fault characteristic data (such as the overall characteristic data obtained by processing in the aforementioned manner). Then, when performing fault identification processing, the overall characteristic data of the target wind turbine group can be compared with the fault characteristic data in the fault case database to obtain the predicted fault type and probability corresponding to the power grid in which the target wind turbine group is located.

[0150] In some embodiments of the present specification, after obtaining the predicted fault type and probability corresponding to the power grid where the target wind turbine is located through the above steps, the processing device 130 can also determine the active support control strategy according to the predicted fault type and probability corresponding to the power grid where the target wind turbine is located. The active support control strategy may include at least one control instruction, which can be used to adjust the operating state of the target wind turbine to adapt to the aforementioned predicted fault type and ensure the stability of the power grid and the safe operation of the wind turbine.

[0151] Specifically, in the embodiment of the present application, the processing device 130 can obtain the current power generation data of the target wind turbine when the probability is greater than a first preset threshold (e.g., 60%), wherein the current power generation data at least includes power output. Then, the degree of influence correlation between the corresponding predicted fault type and the current power generation data is analyzed to determine the parameters to be adjusted; finally, the processing device 130 can generate corresponding control instructions according to the parameters to be adjusted to realize active support control of the target wind turbine.

[0152] For example, if the predicted fault type is voltage drop, and the current power generation data indicates that the power output is high, the processing device 130 may determine that the power output of the wind turbine needs to be reduced to reduce the load on the power grid. In this case, the parameter to be adjusted may include a power set point, and the processing device 130 may generate a control instruction to reduce the power output and send it to the control system of the wind turbine.

[0153] In another possible case, if the predicted fault type is frequency deviation, and the current power generation data indicates that the speed of the wind turbine is high, the processing device 130 may determine that the speed of the wind turbine needs to be adjusted to stabilize the grid frequency. In this case, the parameter to be adjusted may include a speed set point, and the processing device 130 may generate a control instruction to adjust the speed and send it to the control system of the wind turbine.

[0154] For another example, if the predicted fault type is harmonic distortion, and the current power generation data indicates that the harmonic content in the current output by the wind turbine is high, the processing device 130 may determine that the filter of the wind turbine needs to be adjusted to reduce the impact of the harmonics on the power grid. In this case, the parameters to be adjusted may include parameter settings of the filter, and the processing device 130 may generate a control instruction to adjust the filter parameters and send it to the control system of the wind turbine.

[0155] In addition, the processing device 130 can also optimize and adjust other operating parameters of the wind turbine according to the predicted fault type and probability. For example, if the predicted fault type is three-phase imbalance, the processing device 130 may adjust the phase angle of the wind turbine to improve the balance of the output current. In this case, the parameters to be adjusted may include a phase angle setting value, and the processing device 130 can generate a control instruction for adjusting the phase angle and send it to the control system of the wind turbine.

[0156] When implementing the active support control strategy, the processing device 130 may also consider the real-time operating status and environmental conditions (such as meteorological conditions) of the wind turbine to ensure the accuracy and applicability of the control instructions. For example, in the case of high wind speed or large changes in wind direction, the processing device 130 may dynamically adjust the control instructions to adapt to the real-time operating status of the wind turbine.

[0157] In some embodiments, in order to ensure the stability of the power grid and the safe operation of the wind turbines, the processing device 130 can also communicate with the power grid dispatch center to share the operation data and predicted fault information of the wind turbines in real time. By working in collaboration with the power grid dispatch center, the load distribution of the power grid can be further optimized, and the anti-interference ability and reliability of the entire power grid system can be improved.

[0158] In addition, in the embodiment of the present application, the processing device 130 can also adjust other operating parameters of the wind turbine according to the predicted fault type and probability, such as adjusting the blade angle, generator excitation current, grid topology state, etc., to adapt to the fault state of the grid and ensure the safe operation of the wind turbine.

[0159] It should be pointed out that in the embodiment of the present application, by using the state measurement parameters detected or statistically obtained for the power grid where the target wind turbine is located to realize fault identification and prediction of the power grid, and implementing active support control based on this, the operating state of the wind turbine can be adjusted in time when the power grid fails or is about to fail, thereby reducing the impact of the fault on the power grid and improving the stability and reliability of the entire system.

[0160] In some embodiments of the present application, the processing device 130 may also record the fault processing results and the adjusted operating data in a fault case database to provide more reference data for future fault prediction and active support control.

[0161] In summary, the beneficial effects that may be brought about by the embodiments of the present specification include but are not limited to: (1) In the grid fault identification method for wind turbines provided in some embodiments of the present specification, by performing a secondary regression analysis on the state measurement parameters corresponding to the grid where the target wind turbine is located, the data of each characteristic dimension can be fitted with high accuracy, thereby improving the accuracy and reliability of grid fault identification and prediction results; (2) In the grid fault identification method for wind turbines provided in some embodiments of the present specification, by implementing a preset rule conversion to normalize all compressed integration parameters to approximately the same level, the local characteristic data and overall characteristic data corresponding to the target wind turbine can be fitted more quickly and accurately in the subsequent process; (3) In the grid fault identification method for wind turbines provided in some embodiments of the present specification, by using the state measurement parameters detected or statistically obtained for the grid where the target wind turbine is located to realize grid fault identification and prediction, and based on this, active support control is implemented, so that the wind turbine can adjust its operating state in time when the grid fails or is about to fail, thereby reducing the impact of the fault on the grid and improving the stability and reliability of the entire system.

[0162] It should be noted that different embodiments may produce different beneficial effects. In different embodiments, the beneficial effects that may be produced may be any one or a combination of the above, or may be any other beneficial effects that may be obtained.

[0163] The basic concepts have been described above. Obviously, for those skilled in the art, the above detailed disclosure is only for example and does not constitute a limitation of this specification. Although not explicitly stated here, those skilled in the art may make various modifications, improvements and corrections to this specification. Such modifications, improvements and corrections are suggested in this specification, so such modifications, improvements and corrections still belong to the spirit and scope of the exemplary embodiments of this specification.

[0164] At the same time, this specification uses specific words to describe the embodiments of this specification. For example, "one embodiment", "an embodiment", and / or "some embodiments" refer to a certain feature, structure or characteristic related to at least one embodiment of this specification. Therefore, it should be emphasized and noted that "one embodiment" or "an embodiment" or "an alternative embodiment" mentioned twice or more in different positions in this specification does not necessarily refer to the same embodiment. In addition, certain features, structures or characteristics in one or more embodiments of this specification can be appropriately combined.

[0165] In addition, it will be understood by those skilled in the art that various aspects of this specification may be illustrated and described by a number of patentable categories or situations, including any new and useful process, machine, product or combination of substances, or any new and useful improvements thereto. Accordingly, various aspects of this specification may be performed entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. The above hardware or software may be referred to as "data blocks", "modules", "engines", "units", "components" or "systems". In addition, various aspects of this specification may be represented as a computer product located in one or more computer-readable media, which includes computer-readable program code.

[0166] A computer storage medium may include a propagated data signal containing computer program code, for example, in baseband or as part of a carrier wave. The propagated signal may be in a variety of forms, including electromagnetic, optical, etc., or a suitable combination. A computer storage medium may be any computer-readable medium other than a computer-readable storage medium, which can be connected to an instruction execution system, device or apparatus to communicate, propagate or transmit the program for use. The program code on the computer storage medium may be transmitted via any suitable medium, including radio, cable, fiber optic cable, RF, or similar media, or any combination of the above media.

[0167] The computer program codes required for the operation of each part of this specification can be written in any one or more programming languages, including object-oriented programming languages ​​such as Java, Scala, Smalltalk, Eiffel, JADE, Emerald, C++, C#, VB.NET, Python, etc., conventional procedural programming languages ​​such as C language, Visual Basic, Fortran2003, Perl, COBOL2002, PHP, ABAP, dynamic programming languages ​​such as Python, Ruby and Groovy, or other programming languages, etc. The program code can be run entirely on the user's computer, or run on the user's computer as an independent software package, or run partly on the user's computer and partly on a remote computer, or run entirely on a remote computer or processing device. In the latter case, the remote computer can be connected to the user's computer through any network form, such as a local area network (LAN) or a wide area network (WAN), or connected to an external computer (e.g., via the Internet), or in a cloud computing environment, or used as a service such as software as a service (SaaS).

[0168] In addition, unless explicitly stated in the claims, the order of the processing elements and sequences described in this specification, the use of alphanumeric characters, or the use of other names are not intended to limit the order of the processes and methods of this specification. Although the above disclosure discusses some invention embodiments that are currently considered useful through various examples, it should be understood that such details are only for illustrative purposes, and the attached claims are not limited to the disclosed embodiments. On the contrary, the claims are intended to cover all modifications and equivalent combinations that are consistent with the essence and scope of the embodiments of this specification. For example, although the system components described above can be implemented by hardware devices, they can also be implemented only by software solutions, such as installing the described system on an existing processing device or mobile device.

[0169] Similarly, it should be noted that in order to simplify the description disclosed in this specification and thus help understand one or more embodiments of the invention, in the above description of the embodiments of this specification, multiple features are sometimes combined into one embodiment, figure or description thereof. However, this disclosure method does not mean that the features required by the subject matter of this specification are more than the features mentioned in the claims. In fact, the features of the embodiments are less than all the features of the single embodiment disclosed above.

[0170] In some embodiments, numbers describing the number of components and attributes are used. It should be understood that such numbers used in the description of the embodiments are modified by the modifiers "about", "approximately" or "substantially" in some examples. Unless otherwise specified, "about", "approximately" or "substantially" indicate that the numbers are allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximate values, which may change according to the required features of individual embodiments. In some embodiments, the numerical parameters should take into account the specified significant digits and adopt the general method of retaining the digits. Although the numerical domains and parameters used to confirm the breadth of the range in some embodiments of this specification are approximate values, in specific embodiments, the setting of such numerical values ​​is as accurate as possible within the feasible range.

[0171] Each patent, patent application, patent application publication, and other materials, such as articles, books, specifications, publications, documents, etc., cited in this specification is hereby incorporated by reference in its entirety. Except for application history documents that are inconsistent with or conflicting with the contents of this specification, documents that limit the broadest scope of the claims of this specification (currently or later attached to this specification) are also excluded. It should be noted that if the descriptions, definitions, and / or use of terms in the materials attached to this specification are inconsistent or conflicting with the contents described in this specification, the descriptions, definitions, and / or use of terms in this specification shall prevail.

[0172] Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, as an example and not a limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly introduced and described in this specification.

Claims

1. A method for determining a power grid fault for a wind turbine generator system, characterized in that: The method comprises: Acquire state measurement parameters of the power grid where the target wind turbine is located in the target time series, wherein the state measurement parameters include power grid load data, wind turbine power generation data, power grid topology data, power grid voltage and frequency data, and the target time series includes multiple subsequences, each of which includes multiple sampling moments; Determine the original regression coefficient and the original base value corresponding to each state metric parameter in the subsequence; Determine a state metric parameter error based on the state metric parameter corresponding to each of the sampling moments and the original regression coefficient and the original base value corresponding to the state metric parameter; The original base value and the original regression coefficient are updated with the goal of minimizing the state metric parameter error to obtain a first base value and a first regression coefficient; Obtaining first discriminant data according to the first regression coefficient, the first basic value, and a first state metric parameter error of the state metric parameter; Performing feature mapping conversion on the target sampling moments in the subsequences according to the first discriminant data to obtain local feature data corresponding to each of the subsequences; fusing the local feature data corresponding to the plurality of subsequences to obtain fused data, and performing regression analysis on the fused data to obtain the overall feature data of the target wind turbine group; Fault identification processing is performed based on the overall characteristic data of the target wind turbine generator set to obtain the predicted fault type and probability corresponding to the power grid where the target wind turbine generator set is located.

2. The method according to claim 1, characterized in that The method further comprises: An active support control strategy is determined according to the predicted fault type and probability corresponding to the power grid where the target wind turbine is located. The active support control strategy includes at least one control instruction for adjusting the operating state of the target wind turbine to adapt to the predicted fault type.

3. The method according to claim 2, characterized in that The active support control strategy is determined according to the predicted fault type and probability corresponding to the power grid where the target wind turbine is located, including: When the probability is greater than a first preset threshold, obtaining current power generation data of the target wind turbine group, wherein the current power generation data includes power output; Analyzing the degree of influence correlation between the predicted fault type and the current power generation data to determine the parameters to be adjusted; Corresponding control instructions are generated according to the parameters to be adjusted to achieve active support control of the target wind turbine group.

4. The method according to claim 1, characterized in that The step of obtaining the state measurement parameters of the power grid where the target wind turbine generator set is located in the target time series includes: For each of the subsequences in the target time series, obtaining original state measurement parameters of the target wind turbine group in the subsequence; Preprocessing the original state metric parameters to obtain preprocessed state metric parameters corresponding to each of the subsequences, wherein the preprocessing includes data cleaning and compression integration; Performing preset rule conversion on the preprocessed state measurement parameters to obtain a state measurement parameter corresponding to each of the subsequences; Based on the state measurement parameters corresponding to the multiple subsequences, the state measurement parameters of the power grid where the target wind turbine group is located in the target time sequence are obtained.

5. The method according to claim 4, characterized in that The obtaining of the original state measurement parameters of the target wind turbine group in the subsequence includes: For each sampling moment in the subsequence, obtaining grid load data, wind turbine generation data, grid topology data, grid voltage and frequency data of the grid where the target wind turbine is located at the sampling moment; The grid load data, wind turbine generation data, grid topology data, grid voltage and frequency data are combined into the original state measurement parameters of the grid where the target wind turbine is located at the sampling time; Based on the original state metric parameters corresponding to the target wind turbine group at the plurality of sampling moments, the original state metric parameters of the power grid where the target wind turbine group is located in the subsequence are obtained.

6. The method according to claim 4, characterized in that The preprocessing includes compressing and integrating the data obtained after data cleaning to obtain compression integration parameters, wherein the compression integration parameters include compression integration parameters corresponding to multiple feature dimensions at each sampling moment in the subsequence; The performing of preset rule conversion on the preprocessed state metric parameters to obtain the state metric parameters corresponding to each of the subsequences includes: For each compression integration parameter under each feature dimension, obtaining a maximum parameter and a minimum parameter among the compression integration parameters under the feature dimension at multiple sampling moments; Calculating a first parameter difference between the maximum parameter and the minimum parameter, and a second parameter difference between the compression integration parameter of the feature dimension corresponding to each sampling moment and the minimum parameter; Taking the ratio between the second parameter difference corresponding to each sampling moment and the first parameter difference as the normalized state measurement parameter corresponding to the feature dimension at the sampling moment; Obtaining the state measurement parameter corresponding to the sampling moment according to the normalized state measurement parameters in the plurality of feature dimensions at the sampling moment; The state metric parameter corresponding to each of the subsequences is obtained based on the state metric parameters corresponding to the multiple sampling moments.

7. The method according to claim 1, characterized in that The determining of the state metric parameter error based on the state metric parameter corresponding to each sampling moment and the original regression coefficient and the original base value corresponding to the state metric parameter comprises: For each of the sampling moments, mapping conversion is performed on the sampling moments based on the original base value of the state metric parameter and the original regression coefficient of the state metric parameter to obtain a first mapping conversion result; When the absolute value of the difference between the first mapping conversion result and the state metric parameter at the sampling moment is less than or equal to a second preset threshold, determining a sampling moment error corresponding to the sampling moment based on the absolute value of the difference; When the absolute value of the difference between the first mapping conversion result and the state metric parameter at the sampling moment is greater than the second preset threshold, determining a sampling moment error corresponding to the sampling moment based on the absolute value of the difference and the second preset threshold; The sampling moment errors corresponding to the multiple sampling moments are summed to obtain the state metric parameter error.

8. The method according to claim 7, characterized in that The target sampling moments include the first target moment that is closest to the front and the second target moment that is closest to the back in the subsequence, and the feature mapping conversion is performed on the target sampling moments in the subsequence according to the first discrimination data to obtain local feature data corresponding to each of the subsequences, including: Adding a first product of a first regression coefficient corresponding to the first discriminant data and a first target moment in the subsequence to the first base value to obtain local feature data corresponding to the first target moment; Adding a second product of a first regression coefficient corresponding to the first discriminant data and a second target moment in the subsequence to the first base value to obtain local feature data corresponding to the second target moment; Based on the local feature data corresponding to the first target moment and the local feature data corresponding to the second target moment, the local feature data corresponding to each of the subsequences is obtained.

9. The method according to claim 8, characterized in that The performing regression analysis on the fused data to obtain the overall characteristic data of the target wind turbine generator set includes: Performing a univariate linear regression analysis on the fused data to obtain a second regression coefficient and a second base value corresponding to the state measurement parameter; Obtain second discriminant data according to the second regression coefficient, the second basic value, and a second state metric parameter error of the state metric parameter; A feature mapping conversion is performed according to the second discriminant data to obtain overall feature data of the target time series.

10. The method according to claim 9, characterized in that The performing of fault identification processing based on the overall characteristic data of the target wind turbine generator set to obtain the predicted fault type and probability corresponding to the power grid where the target wind turbine generator set is located includes: The overall characteristic data of the target wind turbine is input into a trained machine learning model for processing to obtain the predicted fault type and probability corresponding to the power grid where the target wind turbine is located; wherein the machine learning model is trained in the following manner: Acquire historical fault data and its corresponding fault types as training samples, wherein the historical fault data includes overall characteristic data of the power grid in which the wind turbine is located under a fault state, and the fault type includes at least one of voltage drop, voltage rise, voltage interruption, frequency deviation, harmonic distortion, three-phase imbalance, overvoltage and overcurrent; The fault type corresponding to the historical fault data is used as a sample label, and the initial machine learning model is iteratively trained based on the historical fault data and the sample label, and the training is terminated until the preset training conditions are met to obtain the trained machine learning model, wherein the machine learning model includes one or more of a random forest, a support vector machine, and a neural network model.

Images