A method and system for optimizing control of primary frequency modulation parameters based on big data analysis

Abstract

The present invention provides a primary frequency modulation parameter optimization control method and system based on big data analysis, which relates to the field of wind farm frequency modulation technology. The primary frequency modulation parameter optimization control method based on big data analysis includes the following steps: obtaining historical operation data of wind turbines and preprocessing them, and performing correlation analysis on primary frequency modulation parameters and output power of wind turbines based on binary decision method; establishing a training data set based on the correlation analysis results, and constructing and training a primary frequency modulation parameter prediction model through a vulture search algorithm; using the trained primary frequency modulation parameter prediction model to predict the real-time acquired wind turbine operation data, obtain the optimized primary frequency modulation parameters, and perform frequency modulation control on the wind turbines according to the optimized primary frequency modulation parameters. The present invention improves the intelligence level of wind turbines.

Description

Technical Field

[0001] The present invention relates to the technical field of wind farm frequency regulation, and in particular to a primary frequency regulation parameter optimization control method and system based on big data analysis. Background Art

[0002] With the large-scale access to renewable energy generation facilities such as wind farms, the power system is facing a series of challenges, the most serious of which is the lack of frequency regulation capabilities. This challenge is mainly due to the volatility and uncertainty of renewable energy generation, which puts higher requirements on the frequency stability of the power grid.

[0003] At present, the existing technology still has deficiencies in comprehensively and accurately analyzing the correlation between the primary frequency regulation parameters and output power of wind farms. These deficiencies are mainly reflected in the lack of in-depth understanding of the interaction and influence between the frequency regulation parameters of wind farms and grid demand, which directly affects the optimization and adjustment effect of the frequency regulation parameters, and further affects the overall stability of the grid and the power generation efficiency of wind power.

[0004] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0005] In view of this, the present invention provides a primary frequency modulation parameter optimization control method and system based on big data analysis to solve the above-mentioned problems.

[0006] In order to solve the above problems, the specific technical solutions adopted by the present invention are as follows:

[0007] A primary frequency modulation parameter optimization control method based on big data analysis, the primary frequency modulation parameter optimization control method based on big data analysis comprises the following steps:

[0008] S1. Obtain historical operation data of the wind turbine and perform preprocessing, the historical operation data including primary frequency modulation parameters and output power of the wind turbine;

[0009] S2. Conduct correlation analysis on primary frequency regulation parameters and wind turbine output power based on binary decision method;

[0010] S3. Establish a training data set based on the correlation analysis results, and build and train a frequency modulation parameter prediction model through a vulture search algorithm;

[0011] S4, using the trained primary frequency modulation parameter prediction model to predict the wind turbine operation data obtained in real time, to obtain the optimized primary frequency modulation parameters, and to perform frequency modulation control on the wind turbine according to the optimized primary frequency modulation parameters;

[0012] Preferably, the correlation analysis between the primary frequency modulation parameters and the output power of the wind turbine generator set based on the binary decision method includes the following steps:

[0013] S21, constructing a primary association decision tree of the wind turbine generator set according to the preprocessed historical operation data;

[0014] S22, optimizing the primary association decision tree of the wind turbine generator system by using a logic simplification algorithm to obtain a secondary association decision tree;

[0015] S23, establishing a binary decision diagram of the wind turbine generator using a binary decision method according to the obtained secondary association decision tree;

[0016] S24. According to the binary decision diagram of the wind turbine, a correlation analysis is performed on the primary frequency regulation parameters and the output power of the wind turbine.

[0017] Preferably, obtaining historical operation data of the wind turbine generator set and preprocessing the data comprises the following steps:

[0018] S11. Obtain the primary frequency regulation parameters and output power of the wind turbine using the wind turbine intelligent control platform;

[0019] S12, performing data cleaning processing on the acquired primary frequency modulation parameters and wind turbine output power data;

[0020] S13, normalizing the primary frequency modulation parameters and wind turbine output power data after data cleaning to obtain standardized data.

[0021] Preferably, constructing a primary association decision tree of a wind turbine generator system according to the preprocessed historical operation data comprises the following steps:

[0022] S211, taking the pre-processed wind turbine output power as the top event and identifying the basic event according to the pre-processed primary frequency modulation parameters;

[0023] S212, using time series to analyze the logical relationship between the primary frequency modulation parameter and the output power of the wind turbine generator set, and determining the type of logic gate in the primary association decision tree;

[0024] S213, constructing a primary association decision tree for the wind turbine generator system according to the top events, basic events and logic gate types.

[0025] Preferably, optimizing the primary association decision tree of the wind turbine generator system by using a logic simplification algorithm to obtain a secondary association decision tree includes the following steps:

[0026] S221, traverse the primary association decision tree to find parent and child nodes with the same logic gate type and nodes with only one child node;

[0027] S222, for the parent and child nodes with the same logic gate type, promote the child node of the child node to the parent node, and delete the atomic node and its subtree;

[0028] S223. For a node having only one child node, its child node is promoted to the parent node of the node, and the original node and its subtree are deleted to obtain a secondary associated decision tree.

[0029] Preferably, according to the obtained secondary association decision tree, establishing a binary decision diagram of the wind turbine using a binary decision method comprises the following steps:

[0030] S231, starting from the top event of the secondary association decision tree, counting the occurrence frequencies of all basic events in the secondary association decision tree by traversal;

[0031] S232, sorting the basic events according to the statistical occurrence frequency of the basic events and the logical structure of the secondary associated decision tree to obtain an ordered sequence of basic events;

[0032] S233, based on the ordered sequence of basic events and the logic gate type of the secondary associated decision tree, starting from the top event, gradually build a binary decision diagram.

[0033] Preferably, according to the binary decision diagram of the wind turbine generator set, the correlation analysis of the primary frequency modulation parameter and the output power of the wind turbine generator set comprises the following steps:

[0034] S241. Analyze the minimum cut set that causes the top event based on the binary decision diagram of the wind turbine;

[0035] S242, analyzing the status of the basic events and the top events according to the number of occurrences of the basic events in the minimum cut set, and calculating the structural relevance of each basic event;

[0036] Among them, the state of the basic event and the top event is analyzed according to the number of occurrences of the basic event in the minimum cut set, and the calculation formula for calculating the structural correlation of each basic event is:

[0037] ;

[0038] In the formula, G D ( i ) represents the basic event x structural relevance;

[0039] ∑ D (1 i , x ) represents the basic event x The number of states that occur simultaneously with the top event;

[0040] ∑D (0 i , x ) represents the basic event x The number of states where the top event does not occur and where the top event occurs;

[0041] n represents the number of basic events;

[0042] S243. Sort the basic events according to the structural correlation to obtain the correlation analysis result.

[0043] Preferably, establishing a training data set based on the correlation analysis results, and constructing and training a frequency modulation parameter prediction model by a vulture search algorithm comprises the following steps:

[0044] S31. According to the correlation analysis result, the primary frequency modulation parameters and the output power of the wind turbine generator set that meet the preset correlation threshold are selected as the training set;

[0045] S32, constructing an XGBoost model, and optimizing the parameters of the XGBoost model using the vulture search algorithm to obtain the optimal initial parameter combination;

[0046] S33. According to the optimal initial parameter combination, the XGBoost model is trained using the training set to obtain a frequency modulation parameter prediction model.

[0047] Preferably, constructing an XGBoost model and optimizing the parameters of the XGBoost model using a vulture search algorithm to obtain the optimal initial parameter combination comprises the following steps:

[0048] S321. Build the basic structure of the XGBoost model;

[0049] S322, using the Tent chaos mapping method to initialize the population parameters of the vulture search algorithm;

[0050] S323, using the XGBoost model as the objective function, and calculating the fitness value of each individual in the population parameter;

[0051] S324, sorting the individuals in the population according to the fitness values, and selecting the individual with the highest fitness as the parent generation of the vulture search algorithm;

[0052] S325. Perform crossover and mutation operations using the selected parent individuals to generate a new population, and repeat steps S323 to S324 until a preset number of iterations is reached to obtain the optimal initial parameter combination.

[0053] Preferably, initializing the population parameters of the vulture search algorithm using the Tent chaos mapping method comprises the following steps:

[0054] S3221, initializing the mapping parameters of the Tent chaotic map and the population size of the vulture search algorithm;

[0055] S3222, generating an initial chaotic sequence value using an iterative formula of Tent chaotic mapping;

[0056] S3223, mapping the generated chaotic sequence value to the value range of each population parameter in the vulture search algorithm to obtain the initialized population parameter;

[0057] S3224. Construct the initial population parameters of the vulture search algorithm according to the mapped population parameter values.

[0058] According to another aspect of the present invention, a primary frequency regulation parameter optimization control system based on big data analysis is provided, and the primary frequency regulation parameter optimization control system based on big data analysis includes: a data acquisition processing module, a correlation analysis module, a model building training module and a unit frequency regulation control module;

[0059] A data acquisition and processing module is used to acquire and pre-process the historical operation data of the wind turbine generator set, where the historical operation data includes primary frequency modulation parameters and output power of the wind turbine generator set;

[0060] A correlation analysis module is used to perform correlation analysis on primary frequency regulation parameters and wind turbine output power based on a binary decision method;

[0061] Model building and training module, used to build a training data set based on the correlation analysis results, and to build and train a frequency modulation parameter prediction model through the vulture search algorithm;

[0062] The unit frequency regulation control module is used to use the trained primary frequency regulation parameter prediction model to predict the real-time wind turbine operation data, obtain the optimized primary frequency regulation parameters, and perform frequency regulation control on the wind turbine according to the optimized primary frequency regulation parameters.

[0063] The beneficial effects of the present invention are:

[0064] 1. Through accurate prediction and optimized control, the present invention enables wind turbines to respond more quickly to changes in grid frequency and maintain stable operation of the grid, thereby improving overall operating efficiency. The optimized primary frequency modulation parameters can better adapt to the actual operating conditions of wind turbines and reduce fluctuations and instability caused by parameter mismatch. By reducing unnecessary energy waste and equipment loss, this method helps to reduce the operating costs of wind turbines. Combined with the application of big data analysis and advanced algorithms, this method improves the intelligence level of wind turbines.

[0065] 2. The present invention constructs primary and secondary associated decision trees. The process can deeply explore the complex logical relationship between the primary frequency modulation parameters and the output power of the wind turbine. The binary decision diagram can intuitively reveal the key factors that lead to specific output power changes, that is, the specific settings of the primary frequency modulation parameters. Through correlation analysis, the key factors affecting the output power of the wind turbine can be more accurately identified, which provides a scientific basis for the optimization of the primary frequency modulation parameters, helps to improve the response speed and operating stability of the wind turbine, improve the operating performance of the wind turbine, and promote the development of intelligent operation and maintenance, etc., showing significant advantages.

[0066] 3. The present invention ensures the quality and relevance of training data by selecting data that meets a preset correlation threshold as a training set, laying the foundation for building a high-precision prediction model. The vulture search algorithm is used to optimize the parameters of the XGBoost model, and the optimal initial parameter combination can be found, further improving the prediction performance of the model. The population parameters are initialized by the Tent chaos mapping method, which increases the diversity of the population and helps the algorithm converge to the optimal solution more quickly. By constructing a primary frequency modulation parameter prediction model, accurate prediction and optimal control of the primary frequency modulation parameters of wind turbines can be achieved, which helps to improve the operating efficiency and stability of wind turbines and reduce operating costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0068] Figure 1 is a flow chart of a primary frequency modulation parameter optimization control method based on big data analysis according to an embodiment of the present invention;

[0069] Figure 2 It is a principle block diagram of a primary frequency modulation parameter optimization control system based on big data analysis according to an embodiment of the present invention.

[0070] In the figure:

[0071] 1. Data acquisition and processing module; 2. Correlation analysis module; 3. Model building and training module; 4. Unit frequency regulation control module. DETAILED DESCRIPTION

[0072] In order to enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this application.

[0073] According to an embodiment of the present invention, a primary frequency modulation parameter optimization control method and system based on big data analysis are provided.

[0074] The present invention will now be further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, according to one embodiment of the present invention, a primary frequency modulation parameter optimization control method based on big data analysis is provided, and the primary frequency modulation parameter optimization control method based on big data analysis includes the following steps:

[0075] S1. Obtain historical operation data of the wind turbine and perform preprocessing, the historical operation data including primary frequency modulation parameters and output power of the wind turbine;

[0076] As a preferred implementation, obtaining historical operation data of a wind turbine generator set and preprocessing the data include the following steps:

[0077] S11. Obtain the primary frequency regulation parameters and output power of the wind turbine using the wind turbine intelligent control platform;

[0078] It should be noted that the wind turbine intelligent management and control platform integrates a comprehensive system with data collection, monitoring, analysis and control functions. It can obtain the operating data of wind turbines in real time, including key indicators such as primary frequency regulation parameters and output power. Primary frequency regulation parameters refer to the parameters of wind turbines that automatically adjust their output power in order to maintain grid stability when the grid frequency changes slightly. The output power of wind turbines is an important indicator for measuring the power generation capacity of wind turbines. Real-time acquisition of output power data is necessary for monitoring the operating status of wind turbines, evaluating power generation efficiency, and performing frequency regulation control.

[0079] The wind turbine intelligent management and control platform usually communicates with the control system of the wind turbine to obtain the primary frequency modulation parameters and output power data in real time. These data may be transmitted in the form of digital signals and processed and converted by the platform for subsequent analysis and application.

[0080] S12, performing data cleaning processing on the acquired primary frequency modulation parameters and wind turbine output power data;

[0081] It should be noted that the purpose of data cleaning is to identify and correct errors, anomalies or inconsistencies in the data. Remove duplicate, irrelevant or redundant data. Ensure the accuracy and consistency of the data for subsequent analysis and modeling.

[0082] The steps of data cleaning include:

[0083] Check missing values, identify missing values ​​in the data, and take processing measures such as filling missing values, deleting records with missing values, or retaining missing values ​​as a special category as needed.

[0084] Identify outliers: Identify outliers in the data through statistical analysis, box plots, scatter plots, etc., and correct or delete them based on actual conditions.

[0085] Correct erroneous data, check for errors or inconsistencies in the data like spelling errors, wrong data format, etc. and make corrections accordingly.

[0086] De-duplicate data to identify and delete duplicate records to ensure data uniqueness.

[0087] Process redundant data and remove unnecessary or duplicate features to reduce data complexity and improve analysis efficiency.

[0088] S13, normalizing the primary frequency modulation parameters and wind turbine output power data after data cleaning to obtain standardized data.

[0089] It should be noted that the purpose of normalization is to eliminate the influence of different dimensions and make the values ​​of different features comparable. It can improve the convergence speed and stability of the model, especially when using optimization algorithms such as gradient descent. It can also improve the generalization ability of the model and avoid overfitting. Normalization methods include minimum-maximum normalization, Z-score normalization, etc.

[0090] S2. Conduct correlation analysis on primary frequency regulation parameters and wind turbine output power based on binary decision method;

[0091] As a preferred implementation, the correlation analysis between the primary frequency regulation parameters and the output power of the wind turbine generator set based on the binary decision method includes the following steps:

[0092] S21, constructing a primary association decision tree of the wind turbine generator set according to the preprocessed historical operation data;

[0093] It should be noted that, based on the pre-processed historical operation data, the construction of the primary association decision tree of the wind turbine generator set is based on the logic of the fault decision tree as the core idea.

[0094] As a preferred implementation, constructing a primary association decision tree for a wind turbine generator system based on preprocessed historical operation data includes the following steps:

[0095] S211, taking the pre-processed wind turbine output power as the top event and identifying the basic event according to the pre-processed primary frequency modulation parameters;

[0096] It should be noted that the specific value or change of the wind turbine output power is the core of the decision tree analysis. This means that the actual measured value of the output power will be focused on. The primary frequency modulation parameters are the key factors affecting the output power of wind turbines. Through the pre-processed primary frequency modulation parameters, basic events related to the specific value or change of the output power can be identified. These basic events include specific values ​​of frequency modulation parameters, speed or mode of frequency modulation response, etc.

[0097] S212, using time series to analyze the logical relationship between the primary frequency modulation parameter and the output power of the wind turbine generator set, and determining the type of logic gate in the primary association decision tree;

[0098] Specifically, using time series to analyze the logical relationship between the primary frequency modulation parameters and the output power of the wind turbine generator set and determining the logic gate type in the primary association decision tree includes the following steps:

[0099] Check whether the time series data of primary frequency modulation parameters and wind turbine output power are stable. Non-stationary time series need to be converted into stable series through difference, logarithmic transformation and other methods;

[0100] Use the correlation coefficient of time series (such as Pearson correlation coefficient) to analyze the correlation between primary frequency regulation parameters and wind turbine output power;

[0101] Granger causality test is used to verify whether the change of primary frequency regulation parameters is the cause of the change of wind turbine output power. Granger causality test is a statistical hypothesis test used to determine whether there is a causal relationship between time series.

[0102] According to the results of time series analysis, a dynamic model between the primary frequency regulation parameters and the output power of the wind turbine is established; the model can be a linear regression model, a vector autoregression model (VAR), an autoregression moving average model (ARMA), etc.;

[0103] Based on the results of time series analysis and the dynamic model, the types of logic gates that may be needed in the primary association decision tree are identified. Common logic gate types include "AND gate", "OR gate", "NOT gate", "threshold gate", etc.

[0104] For the selected logic gate type, the corresponding parameters are set according to the output results of the time series analysis and dynamic model. For example, the threshold of the "threshold gate" can be set according to the correlation or causal relationship between the frequency modulation parameter and the output power.

[0105] S213, constructing a primary association decision tree for the wind turbine generator system according to the top events, basic events and logic gate types.

[0106] It should be noted that constructing a primary association decision tree of a wind turbine generator system according to top events, basic events and logic gate types includes the following steps:

[0107] According to the logical relationship between the basic event and the top event, determine the logical reasoning path from the basic event to the top event;

[0108] Use graphical tools or hand-drawing to draw a decision tree according to the logical reasoning path. The decision tree should clearly show the connection between basic events, logic gates, and top events;

[0109] For logic gates that require parameters, such as "threshold gates", specific parameter values, such as threshold ranges, are marked on the decision tree.

[0110] S22, optimizing the primary association decision tree of the wind turbine generator system by using a logic simplification algorithm to obtain a secondary association decision tree;

[0111] It should be noted that in an associative decision tree, if a node is directly connected and causes the occurrence of another node, then this node is called the parent node of the other node. For example, T1 is the parent node of X1, X2 and T2. Corresponding to the parent node, if a node is directly caused by another node, then this node is called the child node of the other node. For example, X1, X2 and T2 are child nodes of T1. Nodes with the same parent node are sibling nodes. For example, X1, X2 and T2 are sibling nodes. The starting node of the fault tree, also called the top event, is the object of the entire fault tree analysis. For example, T1 is the root node of the tree.

[0112] As a preferred implementation, optimizing the primary association decision tree of the wind turbine generator system using a logic simplification algorithm to obtain a secondary association decision tree includes the following steps:

[0113] S221, traverse the primary association decision tree to find parent and child nodes with the same logic gate type and nodes with only one child node;

[0114] It should be noted that the traversal of the primary association decision tree starts from the root node of the primary association decision tree, and the root node is regarded as the currently traversed node.

[0115] S222, for the parent and child nodes with the same logic gate type, promote the child node of the child node to the parent node, and delete the atomic node and its subtree;

[0116] For example, in the process of traversing the primary association decision tree, all parent and child nodes with the same logic gate type have been recorded, and each pair of such nodes is determined, where the parent node is recorded as P and the child node is recorded as C;

[0117] For each pair of parent and child nodes P and C with the same logic gate type, check all child nodes of child node C (recorded as C1, C2, ..., Cn). Connect these child nodes (C1, C2, ..., Cn) directly to the parent node P to become new child nodes of P. Keep the relative order and logical relationship between these child nodes (C1, C2, ..., Cn) unchanged.

[0118] After promoting all the children of child node C, the atomic node C and its remaining subtree (if any) are no longer needed. Therefore, the atomic node C and all its remaining child nodes and connections are deleted to ensure the neatness and consistency of the decision tree.

[0119] S223. For a node having only one child node, its child node is promoted to the parent node of the node, and the original node and its subtree are deleted to obtain a secondary associated decision tree.

[0120] For example, in the process of traversing the primary association decision tree, all nodes with only one child node have been recorded. Such a node is determined, where the node is recorded as N and its child node is recorded as M.

[0121] The child node M is promoted to the parent node of node N to become the new child node of its parent node, while the relative order and logical relationship between the child node M and other sibling nodes remain unchanged.

[0122] After promoting the child node M, the original node N and its subtree (i.e., the child node M and its possible child nodes) are no longer needed. Therefore, the original node N and all its child nodes and connections are deleted to ensure the simplicity and consistency of the decision tree.

[0123] S23, establishing a binary decision diagram of the wind turbine generator using a binary decision method according to the obtained secondary association decision tree;

[0124] As a preferred implementation, based on the obtained secondary association decision tree, using the binary decision method to establish a binary decision diagram for the wind turbine generator set includes the following steps:

[0125] S231, starting from the top event of the secondary association decision tree, counting the occurrence frequencies of all basic events in the secondary association decision tree by traversal;

[0126] It should be noted that, starting from the top event of the secondary association decision tree, counting the occurrence frequencies of all basic events in the secondary association decision tree by traversal includes the following steps:

[0127] Create a dictionary or hash table to store basic events and their frequency;

[0128] Determine the top event of the secondary association decision tree and use it as the starting point of the traversal;

[0129] Use depth-first traversal (DFS) or breadth-first traversal (BFS) to traverse the secondary association decision tree. During the traversal, visit each node and check its type;

[0130] If the current node is a basic event (i.e., a leaf node), its frequency is increased by 1;

[0131] If the current node is a logic gate (such as AND gate, OR gate, etc.), continue to traverse its child nodes. When all nodes have been traversed and there are no unprocessed nodes, the traversal ends.

[0132] S232, sorting the basic events according to the statistical occurrence frequency of the basic events and the logical structure of the secondary associated decision tree to obtain an ordered sequence of basic events;

[0133] It should be noted that the basic events are sorted from high to low according to their frequency of occurrence. If two basic events have the same frequency of occurrence, they can be further sorted according to their positions in the secondary association decision tree or other logical structure information (for example, according to their depth in the tree or some logical order).

[0134] S233, based on the ordered sequence of basic events and the logic gate type of the secondary associated decision tree, starting from the top event, gradually build a binary decision diagram.

[0135] Specifically, based on the ordered sequence of basic events and the logic gate type of the secondary associated decision tree, starting from the top event, gradually constructing a binary decision graph includes the following steps:

[0136] Create an empty binary decision graph, which is a graph object or a data structure used to store nodes and edges. Add the top event of the secondary association decision tree as the starting point of the binary decision graph to the graph;

[0137] Take a basic event from the ordered sequence of basic events, find the position of the basic event in the secondary association decision tree, and its relationship with other events or logic gates;

[0138] According to the logic gate type in the secondary association decision tree, add the corresponding logic gate node in the binary decision diagram, add the currently processed basic event as a leaf node to the binary decision diagram, and connect it to the corresponding logic gate node. If the current basic event is not the last one in the sequence, continue to process the next basic event;

[0139] If a logic gate has multiple input events, ensure that the relationship between these input events and the logic gate is correctly represented in the binary decision diagram. Edges or connecting lines can be used to represent the logical relationship between events.

[0140] After adding each new node and edge, update the binary decision diagram to reflect the latest structure. When all basic events have been added to the binary decision diagram, check the completeness of the diagram to ensure that all logical relationships and events have been correctly represented.

[0141] S24. According to the binary decision diagram of the wind turbine, a correlation analysis is performed on the primary frequency regulation parameters and the output power of the wind turbine.

[0142] As a preferred implementation, according to the binary decision diagram of the wind turbine, the correlation analysis of the primary frequency modulation parameters and the output power of the wind turbine includes the following steps:

[0143] S241. Analyze the minimum cut set that causes the top event based on the binary decision diagram of the wind turbine;

[0144] It should be noted that in the binary decision diagram, the minimum cut set is the minimum set of basic events that lead to the top event (i.e., failure or specific state). By traversing the binary decision diagram, all the minimum cut sets that may lead to the top event are identified. By identifying all logical paths from basic events to the top event, and finding those paths that lead to the top event if and only if all events occur.

[0145] S242, analyzing the status of the basic events and the top events according to the number of occurrences of the basic events in the minimum cut set, and calculating the structural relevance of each basic event;

[0146] It should be noted that for each basic event, determine the state vector of its simultaneous occurrence with the top event. This is usually a binary vector, where 1 indicates that the event has occurred and 0 indicates that the event has not occurred. The state vector can be obtained by traversing the binary decision diagram and recording the state of each basic event and the top event;

[0147] For each basic event, calculate the number of times it occurs simultaneously with the top event. This can be achieved by counting the number of 1s in the state vector. The number of simultaneous occurrences reflects the co-occurrence frequency between the basic event and the top event.

[0148] Among them, the state of the basic event and the top event is analyzed according to the number of occurrences of the basic event in the minimum cut set, and the calculation formula for calculating the structural correlation of each basic event is:

[0149] ;

[0150] In the formula, G D ( i ) represents the basic event x structural relevance;

[0151] ∑ D (1 i , x ) represents the basic event x The number of states that occur simultaneously with the top event;

[0152] ∑ D (0 i , x ) represents the basic event x The number of states where the top event does not occur and where the top event occurs;

[0153] n represents the number of basic events;

[0154] S243. Sort the basic events according to the structural correlation to obtain the correlation analysis result.

[0155] S3. Establish a training data set based on the correlation analysis results, and build and train a frequency modulation parameter prediction model through a vulture search algorithm;

[0156] As a preferred implementation, a training data set is established based on the correlation analysis results, and a frequency modulation parameter prediction model is constructed and trained by a vulture search algorithm, including the following steps:

[0157] S31. According to the correlation analysis result, the primary frequency modulation parameters and the output power of the wind turbine generator set that meet the preset correlation threshold are selected as the training set;

[0158] For example, first, a correlation threshold is set, such as 0.5, which means that only those primary frequency regulation parameters whose correlation with the wind turbine output power is greater than or equal to 0.5 are concerned;

[0159] Traverse all primary frequency modulation parameters and check their correlation with output power. For example, if there are three primary frequency modulation parameters A, B and C, their correlations are 0.6, 0.4 and 0.7 respectively;

[0160] According to the correlation threshold, A and C are selected as the input features of the training set because their correlation is greater than 0.5. At the same time, the wind turbine output power corresponding to these parameters is used as the output label of the training set.

[0161] The historical data of the selected primary frequency regulation parameters A and C and their corresponding historical data of wind turbine output power are combined to form a training data set. This data set will be used to subsequently build and train the primary frequency regulation parameter prediction model.

[0162] S32, constructing an XGBoost model, and optimizing the parameters of the XGBoost model using the vulture search algorithm to obtain the optimal initial parameter combination;

[0163] It should be noted that XGBoost (eXtreme Gradient Boosting) is a machine learning algorithm based on the gradient boosting framework, which gradually improves the model by building a series of ordered decision trees. The objective function of XGBoost consists of two parts: the loss function and the regularization term. The loss function is used to measure the difference between the model's predicted value and the actual value, while the regularization term is used to control the complexity of the model to prevent overfitting. XGBoost optimizes the model by minimizing the expected value of the objective function. Specifically, XGBoost uses a mathematical model called a "pseudo-objective function", which approximates the loss function through a second-order Taylor expansion and combines the regularization term to construct the final objective function.

[0164] As a preferred implementation, constructing an XGBoost model and optimizing the parameters of the XGBoost model using a vulture search algorithm to obtain the optimal initial parameter combination includes the following steps:

[0165] S321. Build the basic structure of the XGBoost model;

[0166] It should be noted that the basic structure of building an XGBoost model includes objective function, parameter settings, etc. When building an XGBoost model, you need to set a variety of parameters to control the training process and performance of the model. These parameters can be roughly divided into three categories: general parameters, booster parameters, and learning task parameters.

[0167] S322, using the Tent chaos mapping method to initialize the population parameters of the vulture search algorithm;

[0168] It should be noted that the Tent chaos mapping method is a classic one-dimensional discrete mapping method based on chaos theory, which has unique advantages in describing the behavior of nonlinear systems. Chaos theory reveals a seemingly disordered but regular form of motion. In Tent chaos mapping, the value of the parameter r has an important influence on the dynamic behavior of the system. When the value of r is within a certain range, the evolution of the system will show chaotic phenomena, which are manifested as the unpredictability of the system state and the sensitive dependence on small changes in initial conditions and parameters.

[0169] As a preferred embodiment, the population parameters of the vulture search algorithm are initialized using the Tent chaos mapping method, including the following steps:

[0170] S3221, initializing the mapping parameters of the Tent chaotic map and the population size of the vulture search algorithm;

[0171] It should be noted that the Tent chaos map is a simple nonlinear map, and its formula is:

[0172] ;

[0173] In the formula, F ( x ) represents the mapping function, which is the input x Output under the action of the function; x Represents the input value of the mapping, which is a real number defined in the interval [0,1].

[0174] Among them, the population size of the vulture search algorithm is the number of individuals N in the population, which is also the length of the chaotic sequence generated by iteration.

[0175] S3222, generating an initial chaotic sequence value using an iterative formula of Tent chaotic mapping;

[0176] It should be noted that by selecting a random number in the interval [0,1] as the initial value, using the iterative formula of the Tent chaotic map, starting from the initial value, iterating N times, each iteration calculates the next value and adds it to the sequence. After N iterations, a chaotic sequence with a length of N will be obtained.

[0177] S3223, mapping the generated chaotic sequence value to the value range of each population parameter in the vulture search algorithm to obtain the initialized population parameter;

[0178] It should be noted that mapping the generated chaotic sequence value to the value range of each population parameter in the vulture search algorithm to obtain the initialized population parameter includes the following steps:

[0179] For each population parameter in the vulture search algorithm, determine its value range;

[0180] For each value in the chaotic sequence, it is mapped by a linear mapping formula until initialization values ​​are generated for all parameters in the vulture search algorithm. It is checked whether the generated initialization population parameters are within the predetermined range. If the conditions are not met, the mapping formula or the chaotic sequence generation method needs to be adjusted.

[0181] The calculation formula of the linear mapping formula is:

[0182] ;

[0183] In the formula, X new Represents the new value after mapping, which will be used as the initial value of a parameter in the vulture search algorithm; X min and X max Indicates the minimum and maximum values ​​of the parameter in the eagle search algorithm; X chaotic Represents a random value in a chaotic sequence; b Represents a constant close to 0, which is used to ensure that the value of the chaotic sequence does not map outside the parameter range.

[0184] S3224. Construct the initial population parameters of the vulture search algorithm according to the mapped population parameter values.

[0185] It should be noted that the above mapping steps need to be performed for each population parameter in the vulture search algorithm. In the end, an initial population consisting of mapped parameter values ​​will be obtained. In this population, each individual will be specifically represented by a set of mapped parameter values. If the vulture search algorithm involves multiple parameters, then each individual will be a vector composed of multiple mapped parameter values. These mapped parameter values ​​need to be integrated into a complete individual in the order of the parameters in the vulture search algorithm.

[0186] S323, using the XGBoost model as the objective function, and calculating the fitness value of each individual in the population parameter;

[0187] Specifically, in the vulture search algorithm (or any similar optimization algorithm), the XGBoost model can be used as the objective function to evaluate the fitness of each individual in the population. Fitness values ​​are usually used to measure the performance of individuals in optimization problems, that is, how close they are to the optimal solution. Here are the steps on how to use the XGBoost model to calculate the fitness value of each individual in the population parameters:

[0188] Define the XGBoost model, including its parameters (such as the number of trees, learning rate, maximum depth, etc.). This model will be used to evaluate the performance of each individual in the population.

[0189] Encodes each individual in the population as a set of parameters that will be used as input or configuration for the XGBoost model. For example, an individual may contain different hyperparameter combinations for XGBoost.

[0190] For each individual in the population, the XGBoost model is configured using its encoded parameters. Based on the performance of the model, a fitness value is calculated for each individual. The higher the value, the better the individual performs in the optimization problem.

[0191] S324, sorting the individuals in the population according to the fitness values, and selecting the individual with the highest fitness as the parent generation of the vulture search algorithm;

[0192] S325. Perform crossover and mutation operations using the selected parent individuals to generate a new population, and repeat steps S323 to S324 until a preset number of iterations is reached to obtain the optimal initial parameter combination.

[0193] It should be noted that for each pair of selected parent individuals, a crossover operation is performed to generate offspring. The crossover operation usually involves exchanging or combining some parameters or features of the parent individuals. For example, in the vulture search algorithm, you can randomly select part of the parameters of the parent individuals and exchange them to generate new individuals. Perform a mutation operation on the offspring individuals generated after the crossover operation. The mutation operation introduces diversity by randomly changing some parameters of the individuals. This helps the algorithm explore new areas in the search space and may find better solutions. Repeat the crossover and mutation operations until a sufficient number of new individuals are generated to form a new population. The new population will contain offspring individuals after the crossover and mutation operations.

[0194] S33. According to the optimal initial parameter combination, the XGBoost model is trained using the training set to obtain a frequency modulation parameter prediction model.

[0195] S4, using the trained primary frequency modulation parameter prediction model to predict the wind turbine operation data obtained in real time, to obtain the optimized primary frequency modulation parameters, and to perform frequency modulation control on the wind turbine according to the optimized primary frequency modulation parameters;

[0196] Specifically, using the trained primary frequency modulation parameter prediction model to predict the real-time wind turbine operation data to obtain optimized primary frequency modulation parameters, and performing frequency modulation control on the wind turbine according to the optimized primary frequency modulation parameters includes the following steps:

[0197] Obtain the operating data of wind turbines in real time, including factors that may affect primary frequency modulation parameters such as wind speed, power, and temperature, and pre-process the data to ensure that the format, range, and missing value processing of the data are consistent with those during model training;

[0198] The preprocessed real-time data is input into the trained primary frequency modulation parameter prediction model, and the model outputs the predicted wind turbine output power based on the input data.

[0199] According to the prediction results and actual needs, it may be necessary to optimize or adjust the primary frequency regulation parameters to determine whether the optimized primary frequency regulation parameters meet the operating requirements and frequency regulation control strategy of the wind turbine.

[0200] If the optimized primary frequency regulation parameters meet the requirements, they will be applied to the frequency regulation control system of the wind turbine to adjust the operating state of the wind turbine according to the new primary frequency regulation parameters to achieve more stable and efficient power output.

[0201] like Figure 2 As shown, according to another embodiment of the present invention, a primary frequency regulation parameter optimization control system based on big data analysis is provided, and the primary frequency regulation parameter optimization control system based on big data analysis includes: a data acquisition processing module 1, a correlation analysis module 2, a model building training module 3 and a unit frequency regulation control module 4;

[0202] Data acquisition and processing module 1, used to acquire and pre-process historical operation data of the wind turbine generator set, the historical operation data including primary frequency modulation parameters and output power of the wind turbine generator set;

[0203] A correlation analysis module 2 is used to perform correlation analysis on the primary frequency regulation parameters and the output power of the wind turbine generator set based on a binary decision method;

[0204] Model building training module 3, used to establish a training data set based on the correlation analysis results, and to build and train a frequency modulation parameter prediction model through a vulture search algorithm;

[0205] The unit frequency regulation control module 4 is used to predict the wind turbine unit operation data acquired in real time by using the trained primary frequency regulation parameter prediction model to obtain optimized primary frequency regulation parameters, and perform frequency regulation control on the wind turbine unit according to the optimized primary frequency regulation parameters.

[0206] In summary, with the help of the above technical solutions of the present invention, the present invention can respond to the frequency change of the power grid more quickly and maintain the stable operation of the power grid through accurate prediction and optimization control, thereby improving the overall operation efficiency. The optimized primary frequency modulation parameters can better adapt to the actual operation of the wind turbine, reduce the fluctuation and instability caused by parameter mismatch, and reduce unnecessary energy waste and equipment loss. The method helps to reduce the operating cost of the wind turbine. Combined with the application of big data analysis and advanced algorithms, the method improves the intelligence level of the wind turbine. The present invention constructs primary and secondary association decision trees. The process can deeply explore the complex logical relationship between the primary frequency modulation parameters and the output power of the wind turbine. By using the binary decision diagram, the key factors that cause the specific output power change, that is, the specific setting of the primary frequency modulation parameters, can be intuitively revealed. Through correlation analysis, the key factors that affect the output power of the wind turbine can be more accurately identified, which provides a scientific basis for the optimization of the primary frequency modulation parameters, helps to improve the response speed and operation stability of the wind turbine, improves the operation performance of the wind turbine, and promotes the development of intelligent operation and maintenance. It shows significant advantages. The present invention ensures the quality and relevance of training data by selecting data that meets a preset correlation threshold as a training set, laying a foundation for building a high-precision prediction model. By using the vulture search algorithm to optimize the parameters of the XGBoost model, the optimal initial parameter combination can be found, further improving the prediction performance of the model. The population parameters are initialized by the Tent chaos mapping method, which increases the diversity of the population and helps the algorithm converge to the optimal solution more quickly. By constructing a primary frequency modulation parameter prediction model, accurate prediction and optimal control of the primary frequency modulation parameters of wind turbines can be achieved, which helps to improve the operating efficiency and stability of wind turbines and reduce operating costs.

[0207] It should be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, optical storage, etc.) containing computer-usable program code.

[0208] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A primary frequency modulation parameter optimization control method based on big data analysis, characterized in that: The primary frequency modulation parameter optimization control method based on big data analysis includes the following steps: S1. Obtaining historical operation data of the wind turbine generator set and preprocessing the data, wherein the historical operation data includes primary frequency modulation parameters and output power of the wind turbine generator set; S2. Conduct correlation analysis on primary frequency regulation parameters and wind turbine output power based on binary decision method; S3. Establish a training data set based on the correlation analysis results, and build and train a frequency modulation parameter prediction model through a vulture search algorithm; S4, using the trained primary frequency modulation parameter prediction model to predict the wind turbine operation data obtained in real time, to obtain the optimized primary frequency modulation parameters, and to perform frequency modulation control on the wind turbine according to the optimized primary frequency modulation parameters; The correlation analysis of the primary frequency modulation parameters and the output power of the wind turbine generator set based on the binary decision method includes the following steps: S21, constructing a primary association decision tree of the wind turbine generator system according to the preprocessed historical operation data; S22, optimizing the primary association decision tree of the wind turbine generator system by using a logic simplification algorithm to obtain a secondary association decision tree; S23, establishing a binary decision diagram of the wind turbine generator using a binary decision method according to the obtained secondary association decision tree; S24, performing correlation analysis on the primary frequency regulation parameters and the output power of the wind turbine generator set according to the binary decision diagram of the wind turbine generator set; The construction of a primary association decision tree for a wind turbine generator system according to the pre-processed historical operation data comprises the following steps: S211, taking the pre-processed wind turbine output power as the top event and identifying the basic event according to the pre-processed primary frequency modulation parameters; S212, using time series to analyze the logical relationship between the primary frequency modulation parameter and the output power of the wind turbine generator set, and determining the type of logic gate in the primary association decision tree; S213, constructing a primary association decision tree for the wind turbine generator set according to the top events, basic events and logic gate types; The method of optimizing the primary association decision tree of the wind turbine generator system by using a logic simplification algorithm to obtain a secondary association decision tree comprises the following steps: S221, traverse the primary association decision tree to find parent and child nodes with the same logic gate type and nodes with only one child node; S222, for the parent and child nodes with the same logic gate type, promote the child node of the child node to the parent node, and delete the atomic node and its subtree; S223, for a node with only one child node, promote its child node to the parent node of the node, and delete the original node and its subtree to obtain a secondary association decision tree; The method of establishing a binary decision diagram of a wind turbine generator using a binary decision method according to the obtained secondary association decision tree comprises the following steps: S231, starting from the top event of the secondary association decision tree, counting the occurrence frequencies of all basic events in the secondary association decision tree by traversal; S232, sorting the basic events according to the statistical occurrence frequency of the basic events and the logical structure of the secondary associated decision tree to obtain an ordered sequence of basic events; S233, based on the ordered sequence of basic events and the logic gate type of the secondary associated decision tree, starting from the top event, gradually build a binary decision diagram; The correlation analysis of the primary frequency modulation parameters and the output power of the wind turbine generator set according to the binary decision diagram of the wind turbine generator set comprises the following steps: S241. Analyze the minimum cut set that causes the top event based on the binary decision diagram of the wind turbine; S242, analyzing the status of the basic events and the top events according to the number of occurrences of the basic events in the minimum cut set, and calculating the structural relevance of each basic event; The formula for analyzing the state of the basic events and the top events according to the number of occurrences of the basic events in the minimum cut set and calculating the structural correlation of each basic event is: ； In the formula, G D ( i ) represents the basic event x structural relevance; ∑ D (1 i , x ) represents the basic event x The number of states that occur simultaneously with the top event; ∑ D (0 i , x ) represents the basic event x The number of states where the top event does not occur and where the top event occurs; n represents the number of basic events; S243, sorting the basic events according to the structural relevance to obtain a relevance analysis result; The method of using the trained primary frequency modulation parameter prediction model to predict the real-time acquired wind turbine operation data to obtain the optimized primary frequency modulation parameters, and performing frequency modulation control on the wind turbine according to the optimized primary frequency modulation parameters includes the following steps: Obtain wind turbine operating data in real time and perform pre-processing; The pre-processed real-time data is input into the trained primary frequency modulation parameter prediction model, and the predicted wind turbine output power is output according to the input data; According to the prediction results and actual needs, the primary frequency regulation parameters are optimized, and it is determined whether the optimized primary frequency regulation parameters meet the operation requirements and frequency regulation control strategy of the wind turbine; If the requirements are met, the optimized primary frequency regulation parameters are applied to the frequency regulation control system of the wind turbine generator set, and the operating state of the wind turbine generator set is adjusted.

2. The method for optimizing and controlling primary frequency modulation parameters based on big data analysis according to claim 1 is characterized in that: The acquisition of historical operation data of the wind turbine generator set and preprocessing includes the following steps: S11. Obtain the primary frequency regulation parameters and output power of the wind turbine using the wind turbine intelligent control platform; S12, performing data cleaning processing on the acquired primary frequency modulation parameters and wind turbine output power data; S13, normalizing the primary frequency modulation parameters and wind turbine output power data after data cleaning to obtain standardized data.

3. The method for optimizing and controlling primary frequency modulation parameters based on big data analysis according to claim 1 is characterized in that: The method of establishing a training data set based on the correlation analysis results and constructing and training a frequency modulation parameter prediction model by using a vulture search algorithm comprises the following steps: S31. According to the correlation analysis result, the primary frequency modulation parameters and the output power of the wind turbine generator set that meet the preset correlation threshold are selected as the training set; S32, constructing an XGBoost model, and optimizing the parameters of the XGBoost model using the vulture search algorithm to obtain the optimal initial parameter combination; S33. According to the optimal initial parameter combination, the XGBoost model is trained using the training set to obtain a frequency modulation parameter prediction model.

4. The method for optimizing and controlling primary frequency modulation parameters based on big data analysis according to claim 3 is characterized in that: The construction of the XGBoost model and the use of the vulture search algorithm to optimize the parameters of the XGBoost model to obtain the optimal initial parameter combination include the following steps: S321. Build the basic structure of the XGBoost model; S322, using the Tent chaos mapping method to initialize the population parameters of the vulture search algorithm; S323, using the XGBoost model as the objective function, and calculating the fitness value of each individual in the population parameter; S324, sorting the individuals in the population according to the fitness values, and selecting the individual with the highest fitness as the parent generation of the vulture search algorithm; S325. Perform crossover and mutation operations using the selected parent individuals to generate a new population, and repeat steps S323 to S324 until a preset number of iterations is reached to obtain the optimal initial parameter combination.

5. The method for optimizing and controlling primary frequency modulation parameters based on big data analysis according to claim 4 is characterized in that: The method of initializing the population parameters of the vulture search algorithm using the Tent chaos mapping method comprises the following steps: S3221, initializing the mapping parameters of the Tent chaotic map and the population size of the vulture search algorithm; S3222, generating an initial chaotic sequence value using an iterative formula of Tent chaotic mapping; S3223, mapping the generated chaotic sequence value to the value range of each population parameter in the vulture search algorithm to obtain the initialized population parameter; S3224. Construct the initial population parameters of the vulture search algorithm according to the mapped population parameter values.

6. A primary frequency modulation parameter optimization control system based on big data analysis, used to implement the primary frequency modulation parameter optimization control method based on big data analysis as described in any one of claims 1 to 5, characterized in that: The primary frequency regulation parameter optimization control system based on big data analysis includes: a data acquisition and processing module, a correlation analysis module, a model building and training module, and a unit frequency regulation control module; The data acquisition and processing module is used to acquire and pre-process the historical operation data of the wind turbine generator set, wherein the historical operation data includes primary frequency modulation parameters and output power of the wind turbine generator set; The correlation analysis module is used to perform correlation analysis on the primary frequency regulation parameters and the output power of the wind turbine generator set based on a binary decision method; The model building training module is used to establish a training data set based on the correlation analysis results, and to build and train a frequency modulation parameter prediction model through a vulture search algorithm; The unit frequency regulation control module is used to predict the wind turbine unit operation data acquired in real time using the trained primary frequency regulation parameter prediction model to obtain optimized primary frequency regulation parameters, and perform frequency regulation control on the wind turbine unit according to the optimized primary frequency regulation parameters.

Images