Component maintenance method, device and equipment based on correlation analysis of wind turbine generator fault ride-through load characteristics and power grid voltage drop depth, and medium

Abstract

The invention provides a part maintenance method, device and equipment based on correlation analysis of wind turbine generator fault ride-through load characteristics and power grid voltage drop depth, and a medium, and relates to the field of wind turbine generator monitoring. The method comprises the following steps: obtaining a feature variable time sequence data set, standardizing the feature variable time sequence data set to obtain standardized data, dividing the standardized data into small, medium and large levels by a preset partition boundary through a preset membership function, and constructing a fuzzy transaction set; and processing the fuzzy transaction set by using a preset Apriori correlation analysis algorithm, and carrying out correlation analysis on the load characteristics of the fault ride-through wind turbine generator and the voltage drop depth of the power grid to obtain a correlation rule of each correlation object. According to the method, the internal incidence relation between the power grid disturbance parameters and the unit load characteristics is determined through data driving, and the practical problems of mechanical fault prevention, operation optimization and maintenance strategy optimization are solved.

Description

Technical Field

[0001] The present invention relates to the field of wind turbine generator set monitoring, and in particular to a component maintenance method, device, equipment and medium based on correlation analysis of wind turbine fault ride-through load characteristics and grid voltage drop depth. Background Art

[0002] Grid-connected wind turbines are subject to both grid disturbances and wind excitation. The dynamic loads on the transmission chain are complex, and the dynamic characteristics of the blades and tower mechanical structures are significantly affected. The mechanical components are closely interconnected. If the impact of electromechanical coupling on the load characteristics of the wind turbine is not considered, the probability of mechanical failure will increase, which will have a significant impact on the power quality.

[0003] Currently, wind turbine load response analysis primarily uses quantitative simulation or qualitative theoretical methods to analyze the impact of grid voltage faults on turbine load characteristics. Grid voltage faults specifically include varying wind speeds, voltage dip depths, symmetrical or asymmetrical voltage fault types, and active power recovery rates. Turbine load characteristics specifically include yaw shafting loads, hub loads, tower loads, mechanical fatigue loads on blades, towers, and main shafts, and the frequency and location of drive train torsional vibrations. For wind turbines with unknown structures or parameters that may exist during actual operation, it is currently impossible to use data-driven methods to clearly define the inherent correlation between grid disturbance parameters and turbine load characteristics, thereby hindering the maintenance of related components.

[0004] However, there is currently no technical solution that can solve the above technical problems, and there is no component maintenance method, device, equipment and medium based on the deep correlation analysis between the fault ride-through load characteristics of wind turbines and the grid voltage drop. Summary of the Invention

[0005] The present invention provides a component maintenance method, device, equipment and medium based on the deep correlation analysis of the fault ride-through load characteristics of wind turbines and the grid voltage drop. Through data-driven means, the intrinsic correlation between the grid disturbance parameters and the unit load characteristics is clarified to solve the practical problems of mechanical failure prevention, operation optimization and maintenance strategy optimization.

[0006] In a first aspect, the present invention provides a component maintenance method based on a deep correlation analysis of wind turbine fault ride-through load characteristics and grid voltage drop, comprising:

[0007] Performing classification feature extraction on a full-characteristics dataset of a grid-connected wind turbine to obtain a characteristic variable time series dataset, wherein the characteristic variable time series dataset includes a grid voltage time series set, a wind turbine output time series set, and a wind turbine load time series set;

[0008] Standardizing the characteristic variable time series data set to obtain standardized data, dividing the standardized data into three levels of small, medium, and large based on preset partition boundaries using a preset membership function, and constructing a fuzzy transaction set;

[0009] Processing the fuzzy transaction set, performing correlation analysis on the load characteristics of the fault ride-through wind turbine and the grid voltage drop depth, and obtaining an association rule for each associated object, wherein the association rule at least includes a support degree and a confidence degree of the associated object;

[0010] For any associated object, a target maintenance strategy for the associated object is determined from a preset maintenance strategy library according to the association rule of the associated object.

[0011] According to the component maintenance method based on the deep correlation analysis between the fault ride-through load characteristics of the wind turbine and the grid voltage drop provided by the present invention, the classification feature extraction of the full characteristic data set of the grid-connected wind turbine is performed to obtain the characteristic variable time series data set, which is determined according to a preset characteristic variable set construction procedure; the preset characteristic variable set construction procedure includes an information recognition step, a feature matching step, a useless elimination step, and a sequence construction step;

[0012] The program constructed by using the preset characteristic variable set performs classification feature extraction on the full characteristic data set of the grid-connected wind turbine generator set to obtain a characteristic variable time series data set, including:

[0013] In the information identification step, different categories of data in the full characteristic data set of the grid-connected wind turbine are identified in sequence according to the data header information, and the data are extracted and stored in categories to obtain grid voltage data, wind turbine output data, and wind turbine load data;

[0014] In the feature matching step, whether the grid voltage data, the wind turbine output data, and the wind turbine load data are required data is determined based on the preset fault type and the load feature variable name; if so, the retained grid voltage data, the retained wind turbine output data, and the retained wind turbine load data are retained and obtained;

[0015] In the useless elimination step, data that fails to pass feature matching is eliminated;

[0016] In the sequence construction step, time series are constructed for the retained grid voltage data, retained wind turbine output data, and retained wind turbine load data according to the timestamps in the data files, and saved as the characteristic variable time series data set.

[0017] According to the component maintenance method based on the correlation analysis between the fault ride-through load characteristics of wind turbines and the grid voltage drop depth provided by the present invention, the standardized data is divided into three levels of small, medium and large according to preset partition boundaries using a preset membership function to construct a fuzzy transaction set, including:

[0018]

[0019] Among them, u min 、u med and u max These are the membership functions corresponding to small, medium, and large levels, r1, r2, r3, and r4 are the boundary values ​​of the preset partition boundaries, and x is the standardized data.

[0020] According to the component maintenance method based on the correlation analysis between the fault ride-through load characteristics of the wind turbine and the grid voltage drop depth provided by the present invention, before constructing the fuzzy transaction set, the boundary values ​​r1, r2, r3 and r4 of the preset partition boundaries are determined based on the sample probability distribution function:

[0021] Calculate the standardized sample data corresponding to the cumulative probability value of 0.2 as r1;

[0022] Calculate the standardized sample data corresponding to the cumulative probability value of 0.4 as r2;

[0023] Calculate the standardized sample data corresponding to the cumulative probability value of 0.6 as r3;

[0024] Calculate the standardized sample data corresponding to the cumulative probability value of 0.8 as r4.

[0025] According to the component maintenance method based on the deep correlation analysis of the load characteristics of the wind turbine fault ride-through and the grid voltage drop provided by the present invention, the full characteristic data set of the grid-connected wind turbine includes electrical data and load data;

[0026] The electrical data includes generator power, generator speed and grid voltage amplitude;

[0027] The load data includes load data in the global coordinate system, tower coordinate system, nacelle coordinate system, blade coordinate system, hub coordinate system, and yaw coordinate system in the GL specification.

[0028] According to the component maintenance method based on the correlation analysis between the fault ride-through load characteristics of the wind turbine and the grid voltage drop depth provided by the present invention, the fuzzy transaction set is processed to perform correlation analysis on the fault ride-through load characteristics of the wind turbine and the grid voltage drop depth, and an association rule for each associated object is obtained, including:

[0029] Repeat the following steps:

[0030] Scan all items in the fuzzy transaction set, calculate the support of the objects, generate a frequent 1-item set L1 and set k=1;

[0031] To connect, set k to increase by 1, and L k-1 Generate candidate set C k ;

[0032] Prune the candidate set C according to the minimum support k Generate frequent k-itemset L k ;

[0033] Until L is satisfied k If it is empty, the iteration is terminated, the confidence of the object is calculated, and the association rules of each associated object are generated based on the minimum confidence.

[0034] According to the component maintenance method based on deep correlation analysis of wind turbine fault ride-through load characteristics and grid voltage drop provided by the present invention, the association rules of each associated object include:

[0035] When the grid voltage drops deeper, the changing trends of generator speed, generator power, generator torque, blade pitch angle, left and right acceleration of the tower top, and front and back acceleration of the tower top;

[0036] As well as the changing trends of the maximum loads of the hub x-axis bending moment, hub y-axis bending moment, hub z-axis bending moment, blade root swing bending moment, blade root flapping bending moment, blade root torque, tower top left and right bending moment, tower top fore-and-aft bending moment, tower top torque, tower bottom left and right bending moment, tower bottom fore-and-aft bending moment, and tower bottom torque.

[0037] According to the component maintenance method based on deep correlation analysis of wind turbine fault ride-through load characteristics and grid voltage sag provided by the present invention, determining the target maintenance strategy of the associated object from a preset maintenance strategy library according to the association rules of the associated object includes:

[0038] If the support degree of the associated object is greater than 20% and the confidence degree of the associated object is greater than or equal to 15%, the target maintenance strategy is determined to be a high-priority maintenance plan. The high-priority maintenance plan includes setting the unit to automatically trigger a protection action and generate an emergency maintenance work order when a power grid fault occurs, and checking related associated objects within a preset time period;

[0039] If the support degree of the associated object is greater than 12% and the confidence degree of the associated object is greater than or equal to 10%, determining the target maintenance strategy as a medium-priority maintenance plan, the medium-priority maintenance plan includes incorporating the associated object into the monthly maintenance plan, monitoring the load parameters of the associated object, and determining the adjustment of the control strategy after manual review;

[0040] When the support degree of the associated object is greater than 8% and the confidence degree of the associated object is greater than ≥3%, the target maintenance strategy is determined to be a low-priority maintenance plan, which includes monitoring and recording the operating data of the associated object without triggering active maintenance, and recording quarterly summary analysis.

[0041] In a second aspect, a component maintenance device based on deep correlation analysis of wind turbine fault ride-through load characteristics and grid voltage drop is provided, comprising:

[0042] An extraction unit, the extraction unit being used to perform classification feature extraction on a full characteristic data set of a grid-connected wind turbine to obtain a characteristic variable time series data set, the characteristic variable time series data set including a grid voltage time series set, a wind turbine output time series set, and a wind turbine load time series set;

[0043] A construction unit, the construction unit being configured to standardize the characteristic variable time series data set to obtain standardized data, and to divide the standardized data into three levels of small, medium, and large based on preset partition boundaries using a preset membership function, thereby constructing a fuzzy transaction set;

[0044] a processing unit configured to process the fuzzy transaction set, perform correlation analysis on the load characteristics of the fault ride-through wind turbine and the grid voltage drop depth, and obtain an association rule for each associated object, wherein the association rule includes at least a support and a confidence of the associated object;

[0045] A determination unit is configured to determine, for any associated object, a target maintenance strategy for the associated object from a preset maintenance strategy library according to an association rule of the associated object.

[0046] In a third aspect, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the component maintenance method based on the deep correlation analysis between the fault ride-through load characteristics of the wind turbine and the grid voltage drop is implemented.

[0047] In a fourth aspect, a computer-readable storage medium is provided, which stores a calculation program. When the calculation program is executed by a processor, the processor implements the component maintenance method based on the correlation analysis between the fault ride-through load characteristics of the wind turbine and the grid voltage drop depth.

[0048] Compared with analyzing the impact of grid voltage faults on unit load characteristics by quantitative simulation or qualitative theory, the present invention clarifies the intrinsic correlation between grid disturbance parameters and unit load characteristics through data-driven means, and can realize the correlation analysis of load characteristics of wind turbines with unknown structures or parameters that may exist during actual operation, thereby improving engineering practicality; using a preset feature variable set to build a program to extract data to construct a feature variable time series data set, it can realize the analysis of large amounts of data, making the correlation analysis results universal and representative; based on the membership function, continuous variables are expanded into three levels of variables: large, medium, and small to intuitively describe their states; and the definition of transaction number calculation in the Apriori algorithm is expanded, which can realize the support and confidence calculation of continuous variables, so that the Apriori algorithm can complete the correlation analysis of continuous variables in load data;

[0049] Because wind turbines face complex grid disturbances and wind excitation during actual operation, traditional theoretical models have difficulty accurately predicting their dynamic responses. This is especially true for turbines with unknown structures or parameters. Data-driven methods can directly extract patterns from operating data without relying on precise physical models, thereby more accurately capturing the complex relationship between grid disturbances and load characteristics. For the output association rules, the higher the support and confidence, the stronger the rule, and the greater the correlation between the two. By finding the association rules between grid disturbances and load characteristics, the present invention can optimize the unit design and control strategy to reduce mechanical component fatigue and damage caused by load fluctuations. The weight is set according to the correlation in the objective function of the optimized control. In terms of preventing mechanical failures, the output association rules can help identify potential mechanical failure risks in advance and reduce the probability of mechanical failure. For example, objects with high correlation are predicted and their failure probability is focused on. Maintenance strategies are optimized based on association rules, reducing unnecessary maintenance costs and extending the service life of the unit. For example, objects with high correlation are focused on maintenance, while objects with low correlation can have the frequency and degree of maintenance reduced. The present invention can solve practical problems such as mechanical failure prevention, operation optimization, and maintenance strategy optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0051] Figure 1 This is a flow chart of a component maintenance method based on the correlation analysis of wind turbine fault ride-through load characteristics and grid voltage drop depth provided by the present invention;

[0052] Figure 2 It is a data processing logic diagram of the characteristic variable time series construction program provided by the present invention;

[0053] Figure 3 This is a flow chart of the wind turbine load characteristic correlation analysis based on the Apriori algorithm provided by the present invention;

[0054] Figure 4 This is one of the schematic diagrams of the characteristic variable time series set provided by the present invention;

[0055] Figure 5 This is the second schematic diagram of the characteristic variable time series set provided by the present invention;

[0056] Figure 6 This is the third schematic diagram of the characteristic variable time series set provided by the present invention;

[0057] Figure 7 This is a schematic structural diagram of a component maintenance device based on correlation analysis of wind turbine fault ride-through load characteristics and grid voltage drop depth provided by the present invention;

[0058] Figure 8 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0059] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0060] Aiming at the correlation analysis of wind turbine load characteristics, the present invention compiles a feature variable set construction program including four steps: information identification, feature matching, useless elimination, and sequence construction. Data is extracted to construct a feature variable time series data set. In addition, a correlation analysis method is designed to dynamically partition continuous variables into three levels: large, medium, and small based on a membership function. The membership function of this correlation analysis method expands continuous variables into three levels: large, medium, and small to intuitively describe their states. The definition of transaction number calculation in the Apriori algorithm is extended to realize support and confidence calculation of continuous variables, thereby completing the correlation analysis of each item in the wind turbine load characteristic transaction set.

[0061] The present invention will introduce the principle, implementation steps, and related parameters of the correlation analysis method in the following description, and disclose the correlation analysis between the load characteristics of wind turbines and the depth of grid voltage drop based on the correlation analysis method. Figure 1 This is a flow chart of a component maintenance method based on the analysis of the correlation between the load characteristics of wind turbine fault ride-through and the depth of grid voltage drop, provided by the present invention. The component maintenance method based on the analysis of the correlation between the load characteristics of wind turbine fault ride-through and the depth of grid voltage drop includes:

[0062] Step 101: performing classification feature extraction on a full characteristic data set of a grid-connected wind turbine to obtain a characteristic variable time series data set, wherein the characteristic variable time series data set includes a grid voltage time series set, a wind turbine output time series set, and a wind turbine load time series set;

[0063] Step 102: Standardize the characteristic variable time series data set to obtain standardized data, and divide the standardized data into three levels of small, medium, and large based on preset partition boundaries using a preset membership function to construct a fuzzy transaction set;

[0064] Step 103: Process the fuzzy transaction set, perform correlation analysis on the load characteristics of the fault ride-through wind turbine and the grid voltage drop depth, and obtain an association rule for each associated object, wherein the association rule at least includes support and confidence of the associated object;

[0065] For any associated object, a target maintenance strategy for the associated object is determined from a preset maintenance strategy library according to the association rule of the associated object.

[0066] In step 101, the preset feature variable set construction procedure includes an information recognition step, a feature matching step, a useless elimination step, and a sequence construction step. Figure 2 This is the data processing logic diagram of the feature variable time series construction program provided by the present invention, such as Figure 2 As shown, firstly, a feature variable set construction program including four steps of information identification, feature matching, useless elimination and sequence construction is written, and then data is extracted to construct a feature variable time series dataset.

[0067] Optionally, the extracting classification features from the full characteristic dataset of the grid-connected wind turbine generator set to obtain the characteristic variable time series dataset is determined according to a preset characteristic variable set construction program; and extracting classification features from the full characteristic dataset of the grid-connected wind turbine generator set using the preset characteristic variable set construction program to obtain the characteristic variable time series dataset includes:

[0068] In the information identification step, different categories of data in the full characteristic data set of the grid-connected wind turbine are identified in sequence according to the data header information, and the data are extracted and stored in categories to obtain grid voltage data, wind turbine output data, and wind turbine load data;

[0069] In the feature matching step, whether the grid voltage data, the wind turbine output data, and the wind turbine load data are required data is determined based on the preset fault type and the load feature variable name; if so, the retained grid voltage data, the retained wind turbine output data, and the retained wind turbine load data are retained and obtained;

[0070] In the useless elimination step, data that fails to pass feature matching is eliminated;

[0071] In the sequence construction step, time series are constructed for the retained grid voltage data, retained wind turbine output data, and retained wind turbine load data according to the timestamps in the data file, that is, the grid voltage time series set, the wind turbine output time series set, and the wind load time series set are saved as the characteristic variable time series data set.

[0072] Optionally, the full characteristic data set of the grid-connected wind turbine generator system includes electrical data and load data;

[0073] The electrical data includes generator power, generator speed and grid voltage amplitude;

[0074] The load data includes load data in the global coordinate system, tower coordinate system, nacelle coordinate system, blade coordinate system, hub coordinate system, and yaw coordinate system in the GL specification.

[0075] In step 102, the data is standardized using the deviation standardization method shown in the following formula to generate the transaction set D1. Where x is the sample data, x* is the standardized data, and x max is the maximum value of the sample data, x min is the minimum value of the sample data:

[0076]

[0077] Then, the normalized data is divided into three levels of small, medium, and large according to preset partition boundaries by a preset membership function, and a fuzzy transaction set is constructed, including:

[0078]

[0079] Among them, u min 、u med and u max These are the membership functions corresponding to small, medium, and large levels, r1, r2, r3, and r4 are the boundary values ​​of the preset partition boundaries, and x is the standardized data.

[0080] Optionally, the present invention processes the feature variable time series data set, converts the continuous variable into a variable dynamically partitioned into three levels: large, medium, and small to intuitively describe its state, generates the fuzzy transaction set required by the Apriori algorithm, and divides the standardized data into three levels: small, medium, and large according to specific partition boundaries through the membership function, constructs the fuzzy transaction set D that can describe the state of the continuous variable, and the membership function can continuously take values ​​in the interval [0,1], where u min (x),u med (x),u max (x) is the membership function corresponding to small, medium and large levels respectively, and [0, r1, r2, r3, r4, 1] is the partition boundary.

[0081] Optionally, before constructing the fuzzy transaction set, boundary values ​​r1, r2, r3, and r4 of the preset partition boundaries are determined based on the sample probability distribution function:

[0082] Calculate the standardized sample data corresponding to the cumulative probability value of 0.2 as r1;

[0083] Calculate the standardized sample data corresponding to the cumulative probability value of 0.4 as r2;

[0084] Calculate the standardized sample data corresponding to the cumulative probability value of 0.6 as r3;

[0085] Calculate the standardized sample data corresponding to the cumulative probability value of 0.8 as r4.

[0086] Optionally, the present invention calculates standardized sample data corresponding to cumulative probability values ​​of 0.2, 0.4, 0.6, and 0.8 as r1, r2, r3, and r4, respectively, to ensure that the data volume of each partition is basically consistent.

[0087] In step 103, the fuzzy transaction set is processed to perform correlation analysis on the load characteristics of the fault ride-through wind turbine and the grid voltage drop depth, and an association rule for each associated object is obtained, including:

[0088] Repeat the following steps:

[0089] Scan all items in the fuzzy transaction set, calculate the support of the objects, generate a frequent 1-item set L1 and set k=1;

[0090] To connect, set k to increase by 1, and L k-1 Generate candidate set C k ;

[0091] Prune the candidate set C according to the minimum support k Generate frequent k-itemset L k ;

[0092] Until L is satisfied k If it is empty, the iteration is terminated, the confidence of the object is calculated, and the association rules of each associated object are generated based on the minimum confidence.

[0093] Alternatively, the Apriori algorithm is a hierarchical algorithm that searches for frequent item sets in ascending order of the number of items contained, and can analyze the associations of items in the transaction set D. The algorithm can be divided into two steps: the first step is to generate frequent item sets based on the minimum support standard, and the second step is to filter strong association rules based on the minimum confidence standard. Suppose there is a transaction set D representing the transaction set composed of wind power data, which is the input data for association analysis, I represents the set of all items in D, and each transaction T in D is the set of individual items, that is, If an association rule "X→Y" exists, it indicates that the occurrence of X is highly likely to lead to the occurrence of Y. The item sets X and Y are both sets of items in I, and X∩Y=Φ. For example, the association rule "PLmax→Pmin" implies that a large voltage dip is likely to result in a reduction in the unit's output active power. Here, "PLmax" and "Pmin" are both wind power data item sets. The item set could also be a disturbance-response combination of "PLmax & PitchAnglemax." Using the Apriori association analysis algorithm, the correlation between the load characteristics of fault-ride-through wind turbines and the depth of the grid voltage dip is analyzed based on the fuzzy transaction set D. The resulting association rules represent the load characteristic correlation analysis results.

[0094] Figure 3 The present invention provides a flow chart of wind turbine load characteristic correlation analysis based on the Apriori algorithm. All items in D are scanned, the support is calculated, and a frequent 1-item set L1 is generated. Through the two steps of connection and pruning, the frequent k-item set L1 is used to generate the frequent k-item set L2. k Explore and generate frequent k+1 item sets L k+1 , and so on, until no new frequent item sets can be found. The connection process refers to the process of L k With L k Connect to generate candidate k+1 item sets C k+1 The pruning process refers to removing C based on the minimum support k+1 The non-frequent candidate items in , thus obtaining the frequent k+1 item set L k+1 .

[0095] Optionally, the Apriori association analysis algorithm calculates rule support as follows: for the association rule "X→Y", its support sup(X→Y) is the percentage of transactions where X and Y occur simultaneously in all transactions, as shown in the following formula:

[0096]

[0097] Based on all the generated frequent item sets, association rules are generated according to the minimum confidence.

[0098] The Apriori association analysis algorithm calculates the confidence of rules as follows: Confidence conf(X→Y) represents the proportion of transactions that occur at the same time as Y among all transactions that occur at X, as shown in formula (6).

[0099]

[0100] Association rules are subject to support and confidence constraints. Based on the setting of minimum support minsup and minimum confidence minconf, if sup(X→Y)>minsup and conf(X→Y)>minconf, the association rule "X→Y" is called a strong association rule.

[0101] Optionally, the present invention adjusts the traditional support and confidence calculation methods for Boolean data accordingly, expands the definition of the number of transactions in the support and confidence calculations, and considers that the number of transactions is equal to the membership value; and for multi-item sets, the number of transactions is defined as the minimum membership value of these items.

[0102] In an optional embodiment, the specific process of wind turbine load characteristic correlation analysis based on the Apriori algorithm is as follows:

[0103] (1) Set the algorithm support and confidence;

[0104] (2) Standardize the input data and generate transaction set D1;

[0105] (3) Convert D1 into fuzzy transaction set D by membership function;

[0106] (4) Scan all items in D, calculate the support, generate the frequent 1-item set L1 and set k = 1;

[0107] (5) Connect, set k to increase by 1, and L k-1 Generate candidate set C k ;

[0108] (6) Pruning is performed based on the minimum support of C k Generate frequent k-itemset L k ;

[0109] (7) If L k If L is empty, the iteration is terminated, otherwise repeat steps (4)-(6) until L is satisfied. k is empty;

[0110] (8) Calculate the confidence and generate association rules based on the minimum confidence.

[0111] Optionally, the association rule for each association object includes:

[0112] When the grid voltage drops deeper, the changing trends of generator speed, generator power, generator torque, blade pitch angle, left and right acceleration of the tower top, and front and back acceleration of the tower top;

[0113] As well as the changing trends of the maximum loads of the hub x-axis bending moment, hub y-axis bending moment, hub z-axis bending moment, blade root swing bending moment, blade root flapping bending moment, blade root torque, tower top left and right bending moment, tower top fore-and-aft bending moment, tower top torque, tower bottom left and right bending moment, tower bottom fore-and-aft bending moment, and tower bottom torque.

[0114] Optionally, determining the target maintenance strategy of the associated object from a preset maintenance strategy library according to the association rule of the associated object includes:

[0115] If the support degree of the associated object is greater than 20% and the confidence degree of the associated object is greater than or equal to 15%, the target maintenance strategy is determined to be a high-priority maintenance plan. The high-priority maintenance plan includes setting the unit to automatically trigger a protection action and generate an emergency maintenance work order when a power grid fault occurs, and checking related associated objects within a preset time period;

[0116] If the support degree of the associated object is greater than 12% and the confidence degree of the associated object is greater than or equal to 10%, determining the target maintenance strategy as a medium-priority maintenance plan, the medium-priority maintenance plan includes incorporating the associated object into the monthly maintenance plan, monitoring the load parameters of the associated object, and determining the adjustment of the control strategy after manual review;

[0117] When the support degree of the associated object is greater than 8% and the confidence degree of the associated object is greater than ≥3%, the target maintenance strategy is determined to be a low-priority maintenance plan, which includes monitoring and recording the operating data of the associated object without triggering active maintenance, and recording quarterly summary analysis.

[0118] Optionally, the association analysis result in step 103 of the present invention is obtained based on the support set to 20% and the confidence set to 15% in the association analysis algorithm. After obtaining the association rules and their corresponding support and confidence, the rules are graded based on the support and confidence and different maintenance strategies are adopted. For the association objects with support ≥ 20% and confidence ≥ 15%, a high-priority maintenance plan based on emergency intervention is adopted, that is, the unit is set to automatically trigger protection actions (such as power reduction, pitch angle compensation) when the power grid fails and generate an emergency maintenance work order, and the relevant components are inspected within 24 hours. For the association objects with support ≥ 12% and confidence ≥ 10%, a medium-priority maintenance plan based on planned maintenance is adopted, that is, the component is included in the monthly maintenance plan, the relevant load parameters are monitored, and the control strategy is decided after manual review of the data. For the association objects with support ≥ 8% and confidence ≥ 3%, a low-priority maintenance plan based on observation and recording is adopted, that is, the operating data of the component is continuously monitored and recorded but no active maintenance is triggered. At the same time, a summary analysis is performed every quarter to verify whether the rule is stable.

[0119] Figure 4 This is one of the schematic diagrams of the characteristic variable time series set provided by the present invention. Figure 5 This is the second schematic diagram of the characteristic variable time series set provided by the present invention. Figure 6 This is the third schematic diagram of the characteristic variable time series set provided by the present invention. The present invention will be described in conjunction with the above method for a specific example, and the steps are as follows:

[0120] Step 1: Based on the acquired wind turbine full-characteristic monitoring data set during the fault process, the characteristic variables used for the wind turbine load characteristic correlation analysis are selected as shown in Table 1:

[0121] Table 1 Variable set for correlation analysis of wind turbine load characteristics

[0122] variable name meaning variable name meaning PL Fault voltage amplitude percentage MXT Left and right bending moment at tower top Generator speed Generator speed MYT Bending moment before and after tower top Electrical power Generator power MZT Tower top torque Mean pitch angle Blade pitch angle Tower x acceleration Tower top left and right acceleration Rotating hub Mx Hub x-axis bending moment Tower y acceleration Acceleration before and after the top of the tower Rotating hub My Hub y-axis bending moment Foundation Mx Left and right bending moment at tower bottom Rotating hub Mz Hub z-axis bending moment Foundation My Bending moment before and after tower bottom Blade root Mx Blade root shimmy bending moment Foundation Mz Tower bottom torque Blade root My Blade root swinging moment Generator Torque Generator torque Blade root Mz Blade root torque

[0123] Step 2: Based on the variable set, the characteristic variable time series construction program is applied to process the full characteristic monitoring data set of the wind turbine during the fault process to establish the characteristic variable time series set. The results are as follows: Figure 4 、 Figure 5 as well as Figure 6 shown.

[0124] Step 3: Based on the characteristic variable time series set, the Apriori association analysis algorithm is used to perform a correlation analysis between the load characteristics of the fault ride-through wind turbine and the grid voltage drop depth. The data classification results under the dynamic fuzzy partitioning based on the membership function are shown in Table 2, and the correlation analysis results are shown in Table 3:

[0125] Table 2 Data classification results under dynamic fuzzy partitioning based on membership function

[0126]

[0127] Table 3 Correlation analysis results of wind turbine load characteristics under fault ride-through

[0128]

[0129]

[0130] During fault ride-through, the greater the fault voltage amplitude, that is, the smaller the voltage dip depth, the less severe the wind turbine load change. Rules 5, 7, 8, and 10 indicate that as the voltage dip depth decreases, the maximum loads of the blade root flapping bending moment (blade root coordinate system My), blade root torque (blade root coordinate system Mz), drive chain shafting (hub coordinate system Mx), tower top left-right bending moment (tower top coordinate system Mx), and tower bottom left-right bending moment (tower bottom coordinate system Mx) will decrease. Rules 1, 3, 4, and 6 indicate that as the voltage dip depth decreases, the electromagnetic disturbance experienced by the generator decreases, and the drive chain shafting and pitch control motion also decrease. Rules 9, 12, and 13 indicate that as the voltage dip depth decreases, a decrease in pitch angle will increase the maximum load of the tower bottom torque (tower bottom coordinate system Mz).

[0131] The above association rules and Figure 4 、 Figure 5 as well as Figure 6 The results of the load characteristics of the wind turbines are consistent, which verifies the effectiveness of the wind turbine load characteristics association analysis method based on the Apriori algorithm. In addition, the association rules do not include the relationship between the "PLMin" item and each load characteristic variable. The reason is that the non-zero items in "PL Min" are concentrated in the [0,0.333] interval, and Figure 4 、 Figure 5 as well as Figure 6 It can be seen that when the voltage drop depth exceeds 35%, the change trends of various load characteristic variables are quite different, making it difficult to mine strong association rules.

[0132] The present invention provides a method for analyzing the correlation between the fault ride-through load characteristics of a wind turbine and the depth of the grid voltage drop, as well as a method for implementing component maintenance based on the correlation analysis. The method is suitable for analyzing the correlation between the fault ride-through load characteristics of a wind turbine and the depth of the grid voltage drop, thereby clarifying the correlation between the key operating characteristics of the wind turbine and the grid disturbance parameters, providing a methodological basis for achieving mechanical friendliness of wind turbines for large-scale wind power access to the power system, and is of great significance to improving the safety and stability of wind turbine grid connection.

[0133] Compared with analyzing the impact of grid voltage faults on unit load characteristics by quantitative simulation or qualitative theory, the present invention clarifies the intrinsic correlation between grid disturbance parameters and unit load characteristics through data-driven means, and can realize the correlation analysis of load characteristics of wind turbines with unknown structures or parameters that may exist during actual operation, thereby improving engineering practicality; using a preset feature variable set to build a program to extract data to construct a feature variable time series data set, it can realize the analysis of large amounts of data, making the correlation analysis results universal and representative; based on the membership function, continuous variables are expanded into three levels of variables: large, medium, and small to intuitively describe their states; and the definition of transaction number calculation in the Apriori algorithm is expanded, which can realize the support and confidence calculation of continuous variables, so that the Apriori algorithm can complete the correlation analysis of continuous variables in load data;

[0134] Because wind turbines face complex grid disturbances and wind excitation during actual operation, traditional theoretical models have difficulty accurately predicting their dynamic responses. This is especially true for turbines with unknown structures or parameters. Data-driven methods can directly extract patterns from operating data without relying on precise physical models, thereby more accurately capturing the complex relationship between grid disturbances and load characteristics. For the output association rules, the higher the support and confidence, the stronger the rule, and the greater the correlation between the two. By finding the association rules between grid disturbances and load characteristics, the present invention can optimize the unit design and control strategy to reduce mechanical component fatigue and damage caused by load fluctuations. The weight is set according to the correlation in the objective function of the optimized control. In terms of preventing mechanical failures, the output association rules can help identify potential mechanical failure risks in advance and reduce the probability of mechanical failure. For example, objects with high correlation are predicted and their failure probability is focused on. Maintenance strategies are optimized based on association rules, reducing unnecessary maintenance costs and extending the service life of the unit. For example, objects with high correlation are focused on maintenance, while objects with low correlation can have the frequency and degree of maintenance reduced. The present invention can solve practical problems such as mechanical failure prevention, operation optimization, and maintenance strategy optimization.

[0135] Figure 7 This is a structural schematic diagram of a component maintenance device based on the analysis of the correlation between the load characteristics of wind turbine fault ride-through and the depth of grid voltage drop provided by the present invention. The component maintenance device based on the analysis of the correlation between the load characteristics of wind turbine fault ride-through and the depth of grid voltage drop includes an extraction unit 1. The extraction unit 1 is used to construct a program using a preset feature variable set to perform classification feature extraction on the full characteristic data set of the grid-connected wind turbine to obtain a feature variable time series data set. The feature variable time series data set includes a grid voltage time series set, a wind turbine output time series set, and a wind turbine load time series set. The working principle of the extraction unit 1 can refer to the aforementioned step 101 and will not be repeated here.

[0136] The component maintenance device based on the correlation analysis between the fault ride-through load characteristics of the wind turbine and the grid voltage drop depth also includes a construction unit 2, which is used to standardize the characteristic variable time series data set to obtain standardized data, and divide the standardized data into three levels of small, medium and large according to preset partition boundaries through a preset membership function to construct a fuzzy transaction set. The working principle of the construction unit 2 can refer to the aforementioned step 102 and will not be repeated here.

[0137] The component maintenance device based on the correlation analysis between the load characteristics of the wind turbine fault ride-through and the depth of the grid voltage drop also includes a processing unit 3, which is used to process the fuzzy transaction set, perform correlation analysis on the load characteristics of the wind turbine fault ride-through and the depth of the grid voltage drop, and obtain association rules for each associated object. The association rules at least include the support and confidence of the associated object. The working principle of the construction unit 3 can refer to the aforementioned step 103 and will not be repeated here.

[0138] The component maintenance device based on the correlation analysis between the fault ride-through load characteristics of the wind turbine and the grid voltage drop depth also includes a determination unit 4, which is used to determine the target maintenance strategy of the associated object from a preset maintenance strategy library according to the association rules of the associated object for any associated object. The working principle of the determination unit 4 can refer to the aforementioned step 104 and will not be repeated here.

[0139] Compared with analyzing the impact of grid voltage faults on unit load characteristics by quantitative simulation or qualitative theory, the present invention clarifies the intrinsic correlation between grid disturbance parameters and unit load characteristics through data-driven means, and can realize the correlation analysis of load characteristics of wind turbines with unknown structures or parameters that may exist during actual operation, thereby improving engineering practicality; using a preset feature variable set to build a program to extract data to construct a feature variable time series data set, it can realize the analysis of large amounts of data, making the correlation analysis results universal and representative; based on the membership function, continuous variables are expanded into three levels of variables: large, medium, and small to intuitively describe their states; and the definition of transaction number calculation in the Apriori algorithm is expanded, which can realize the support and confidence calculation of continuous variables, so that the Apriori algorithm can complete the correlation analysis of continuous variables in load data;

[0140] Since wind turbines face complex grid disturbances and wind excitation in actual operation, traditional theoretical models are difficult to accurately predict their dynamic responses, especially for turbines with unknown structures or parameters. Data-driven methods can directly extract rules from operating data without relying on precise physical models, thereby more accurately capturing the complex relationship between grid disturbances and load characteristics. For the output association rules, the higher the support and confidence, the stronger the rule, and the greater the correlation between the two. By finding the association rules between grid disturbances and load characteristics, the present invention can optimize the unit design and control strategy to reduce fatigue and damage of mechanical components caused by load fluctuations, and set weights according to the correlation in the objective function of the optimized control. In terms of preventing mechanical failures, the output association rules can help identify potential mechanical failure risks in advance and reduce the probability of mechanical failures, such as predicting objects with greater correlation and focusing on their failure probability. Maintenance strategies are optimized based on association rules to reduce unnecessary maintenance costs and extend the service life of the unit. For example, objects with greater correlation are focused on maintenance, while objects with less correlation can reduce the number and degree of maintenance. The present invention can solve practical problems such as mechanical failure prevention, operation optimization, and maintenance strategy optimization.

[0141] Figure 8 Schematic diagram of the structure of the electronic device provided by the present invention. Figure 8 As shown, the electronic device may include: a processor (processor) 110, a communication interface (Communications Interface) 120, a memory (memory) 130 and a communication bus 140, wherein the processor 110, the communication interface 120, and the memory 130 communicate with each other through the communication bus 140. The processor 110 can call the logic instructions in the memory 130 to execute a component maintenance method based on the correlation analysis between the load characteristics of the wind turbine fault ride-through and the depth of the grid voltage drop, the method including: performing classification feature extraction on the full characteristic data set of the grid-connected wind turbine to obtain a characteristic variable time series data set, the characteristic variable time series data set including the grid voltage time series set, the wind turbine output time series set and the wind turbine load time series set; standardizing the characteristic variable time series data set to obtain standardized data, dividing the standardized data into three levels of small, medium and large according to preset partition boundaries through a preset membership function, and constructing a fuzzy transaction set; processing the fuzzy transaction set, performing correlation analysis on the load characteristics of the wind turbine fault ride-through and the depth of the grid voltage drop, and obtaining association rules for each associated object, the association rules including at least the support and confidence of the associated object; for any associated object, determining the target maintenance strategy of the associated object from a preset maintenance strategy library according to the association rule of the associated object.

[0142] In addition, the logic instructions in the above-mentioned memory 130 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0143] On the other hand, the present invention also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute a component maintenance method based on the correlation analysis of the fault ride-through load characteristics of the wind turbine set and the grid voltage drop depth provided by the above methods. The method includes: performing classification feature extraction on the full characteristic data set of the grid-connected wind turbine set to obtain a characteristic variable time series data set, the characteristic variable time series data set including a grid voltage time series set, a wind turbine output time series set and a wind turbine load time series set; standardizing the characteristic variable time series data set to obtain standardized data, dividing the standardized data into three levels of small, medium and large according to preset partition boundaries through a preset membership function, and constructing a fuzzy transaction set; processing the fuzzy transaction set, performing correlation analysis on the fault ride-through load characteristics of the wind turbine set and the grid voltage drop depth, and obtaining association rules for each associated object, the association rules at least including the support and confidence of the associated object; for any associated object, determining the target maintenance strategy of the associated object from a preset maintenance strategy library according to the association rule of the associated object.

[0144] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the component maintenance method based on the correlation analysis between the fault-ride-through load characteristics of the wind turbine and the grid voltage drop depth provided by the above-mentioned methods, the method comprising: performing classification feature extraction on the full characteristic data set of the grid-connected wind turbine to obtain a characteristic variable time series data set, the characteristic variable time series data set comprising a grid voltage time series set, a wind turbine output time series set and a wind turbine load time series set; standardizing the characteristic variable time series data set to obtain standardized data, dividing the standardized data into three levels of small, medium and large according to preset partition boundaries through a preset membership function, and constructing a fuzzy transaction set; processing the fuzzy transaction set, performing correlation analysis on the fault-ride-through load characteristics of the wind turbine and the grid voltage drop depth, and obtaining association rules for each associated object, the association rules comprising at least the support and confidence of the associated object; for any associated object, determining the target maintenance strategy of the associated object from a preset maintenance strategy library according to the association rule of the associated object.

[0145] The device embodiments described above are merely illustrative. 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 may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0146] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, or of course, by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the existing technology can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or certain parts of the embodiments.

[0147] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A component maintenance method based on the correlation analysis of wind turbine fault ride-through load characteristics and grid voltage drop depth, characterized in that: include: Classification feature extraction is performed on the full characteristic data set of grid-connected wind turbines. Obtaining a characteristic variable time series data set, wherein the characteristic variable time series data set includes a grid voltage time series set, a wind turbine output time series set, and a wind turbine load time series set; Standardizing the characteristic variable time series data set to obtain standardized data, dividing the standardized data into three levels of small, medium, and large based on preset partition boundaries using a preset membership function, and constructing a fuzzy transaction set; Processing the fuzzy transaction set, performing correlation analysis on the load characteristics of the fault ride-through wind turbine and the grid voltage drop depth, and obtaining an association rule for each associated object, wherein the association rule at least includes a support degree and a confidence degree of the associated object; For any associated object, a target maintenance strategy for the associated object is determined from a preset maintenance strategy library according to the association rule of the associated object.

2. The component maintenance method based on the deep correlation analysis of wind turbine fault ride-through load characteristics and grid voltage drop according to claim 1 is characterized in that: The classification feature extraction of the full characteristic data set of the grid-connected wind turbine generator system to obtain the characteristic variable time series data set is determined according to a preset characteristic variable set construction program; The preset feature variable set construction procedure includes an information recognition step, a feature matching step, a useless elimination step, and a sequence construction step; The program is constructed by using the preset feature variable set to extract classification features from the full characteristic data set of the grid-connected wind turbine generator set. Get the feature variable time series data set, including: In the information identification step, different categories of data in the full characteristic data set of the grid-connected wind turbine are identified in sequence according to the data header information, and the data are extracted and stored in categories to obtain grid voltage data, wind turbine output data, and wind turbine load data; In the feature matching step, whether the grid voltage data, the wind turbine output data, and the wind turbine load data are required data is determined based on the preset fault type and the load feature variable name; if so, the retained grid voltage data, the retained wind turbine output data, and the retained wind turbine load data are retained and obtained; In the useless elimination step, data that fails to pass feature matching is eliminated; In the sequence construction step, time series are constructed for the retained grid voltage data, retained wind turbine output data, and retained wind turbine load data according to the timestamps in the data files, and saved as the characteristic variable time series data set.

3. The component maintenance method based on the deep correlation analysis of wind turbine fault ride-through load characteristics and grid voltage drop according to claim 1 is characterized in that: The method of dividing the normalized data into three levels of small, medium, and large based on preset partition boundaries by a preset membership function to construct a fuzzy transaction set includes: Among them, u min 、u med and u max These are the membership functions corresponding to small, medium, and large levels, r1, r2, r3, and r4 are the boundary values ​​of the preset partition boundaries, and x is the standardized data.

4. The component maintenance method based on the correlation analysis between the fault ride-through load characteristics of the wind turbine and the grid voltage drop depth according to claim 3 is characterized in that: Before constructing the fuzzy transaction set, the boundary values ​​r1, r2, r3, and r4 of the preset partition boundaries are determined based on the sample probability distribution function: Calculate the standardized sample data corresponding to the cumulative probability value of 0.2 as r1; Calculate the standardized sample data corresponding to the cumulative probability value of 0.4 as r2; Calculate the standardized sample data corresponding to the cumulative probability value of 0.6 as r3; Calculate the standardized sample data corresponding to the cumulative probability value of 0.8 as r4.

5. The component maintenance method based on the deep correlation analysis of wind turbine fault ride-through load characteristics and grid voltage drop according to claim 1 is characterized in that: The full characteristic data set of the grid-connected wind turbine generator system includes electrical data and load data; The electrical data includes generator power, generator speed and grid voltage amplitude; The load data includes load data in the global coordinate system, tower coordinate system, nacelle coordinate system, blade coordinate system, hub coordinate system, and yaw coordinate system in the GL specification.

6. The component maintenance method based on the deep correlation analysis of wind turbine fault ride-through load characteristics and grid voltage drop according to claim 1 is characterized in that: The fuzzy transaction set is processed to perform correlation analysis on the load characteristics of the fault ride-through wind turbine and the grid voltage drop depth, and an association rule for each associated object is obtained, including: Repeat the following steps: Scan all items in the fuzzy transaction set, calculate the support of the objects, generate a frequent 1-item set L1 and set k=1; To connect, set k to increase by 1, and L k-1 Generate candidate set C k ; Prune the candidate set C according to the minimum support k Generate frequent k-itemset L k ; Until L is satisfied k If it is empty, the iteration is terminated, the confidence of the object is calculated, and the association rules of each associated object are generated based on the minimum confidence.

7. The component maintenance method based on the analysis of the correlation between the fault ride-through load characteristics of the wind turbine and the grid voltage drop depth according to claim 6 is characterized in that: The association rules of each association object include: When the grid voltage drops deeper, the changing trends of generator speed, generator power, generator torque, blade pitch angle, left and right acceleration of the tower top, and front and back acceleration of the tower top; As well as the changing trends of the maximum loads of the hub x-axis bending moment, hub y-axis bending moment, hub z-axis bending moment, blade root swing bending moment, blade root flapping bending moment, blade root torque, tower top left and right bending moment, tower top fore-and-aft bending moment, tower top torque, tower bottom left and right bending moment, tower bottom fore-and-aft bending moment, and tower bottom torque.

8. The component maintenance method based on the correlation analysis between the fault ride-through load characteristics of wind turbines and the grid voltage drop depth according to claim 1 is characterized in that: The determining the target maintenance strategy of the associated object from a preset maintenance strategy library according to the association rule of the associated object includes: If the support degree of the associated object is greater than 20% and the confidence degree of the associated object is greater than or equal to 15%, the target maintenance strategy is determined to be a high-priority maintenance plan. The high-priority maintenance plan includes setting the unit to automatically trigger a protection action and generate an emergency maintenance work order when a power grid fault occurs, and checking related associated objects within a preset time period; If the support degree of the associated object is greater than 12% and the confidence degree of the associated object is greater than or equal to 10%, determining the target maintenance strategy as a medium-priority maintenance plan, the medium-priority maintenance plan includes incorporating the associated object into the monthly maintenance plan, monitoring the load parameters of the associated object, and determining the adjustment of the control strategy after manual review; When the support degree of the associated object is greater than 8% and the confidence degree of the associated object is greater than ≥3%, the target maintenance strategy is determined to be a low-priority maintenance plan, which includes monitoring and recording the operating data of the associated object without triggering active maintenance, and recording quarterly summary analysis.

9. A device using the component maintenance method based on correlation analysis of wind turbine fault ride-through load characteristics and grid voltage drop depth as described in any one of claims 1 to 8, characterized in that: include: An extraction unit is used to extract classification features from a full characteristic data set of a grid-connected wind turbine generator set. Obtaining a characteristic variable time series data set, wherein the characteristic variable time series data set includes a grid voltage time series set, a wind turbine output time series set, and a wind turbine load time series set; A construction unit, the construction unit being configured to standardize the characteristic variable time series data set to obtain standardized data, and to divide the standardized data into three levels of small, medium, and large based on preset partition boundaries using a preset membership function, thereby constructing a fuzzy transaction set; a processing unit configured to process the fuzzy transaction set, perform correlation analysis on the load characteristics of the fault ride-through wind turbine and the grid voltage drop depth, and obtain an association rule for each associated object, wherein the association rule includes at least a support and a confidence of the associated object; A determination unit is configured to determine, for any associated object, a target maintenance strategy for the associated object from a preset maintenance strategy library according to an association rule of the associated object.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the component maintenance method based on the deep correlation analysis between the fault ride-through load characteristics of the wind turbine and the grid voltage drop as described in any one of claims 1 to 8 is implemented.

11. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a calculation program. When the calculation program is executed by a processor, the processor implements the component maintenance method based on the correlation analysis between the fault ride-through load characteristics of the wind turbine and the grid voltage drop depth as described in any one of claims 1 to 8.

Images