A method and system for intelligently optimizing power generation of wind turbines based on load control

Abstract

The present invention relates to the field of intelligent control technology, and discloses an intelligent optimization method and system for wind turbine power generation based on load control. The method collects historical wind speed data in the wind turbine, processes it using a data processing method, and constructs a wind speed probability model and a wind power model based on the processed historical wind speed data; and collects various historical parameters in the wind turbine to construct a wind energy utilization coefficient and wind speed relationship model, and selects part of the processed historical wind speed data to dynamically predict the wind power of the wind turbine based on the wind speed probability model and the random simulation method, and construct a dynamic prediction model, and at the same time, based on the constructed dynamic prediction model and the wind energy utilization coefficient and wind speed relationship model, the speed of the wind turbine is controlled, and finally a random sampling method is used to verify the accuracy of the constructed dynamic prediction model, thereby improving the accuracy of the load control of the wind turbine power generation.

Description

Technical Field

[0001] The present invention relates to the field of intelligent control technology, and in particular to a method and system for intelligently optimizing power generation of a wind turbine generator set based on load control. Background Art

[0002] As the proportion of wind power in the power system increases, and because most wind turbines are currently regulated by ensuring the maximum power of wind turbines, the uncertainty of wind speed often does not achieve the expected results by ensuring the maximum power of wind turbines, which not only increases the burden on wind turbines but also makes power generation unstable.

[0003] The existing Chinese patent application CN118192209A, this method constructs a nonlinear mathematical model of the wind turbine, and based on the constructed nonlinear mathematical model of the wind turbine, sets the maximum constraint of the wind turbine output power, the minimum constraint of the fatigue load of the key mechanism and the safe operation constraint of the wind turbine, and establishes an optimization problem of economic model predictive control; obtains the optimal control sequence at the current moment, executes the optimal control operation at the current moment, and obtains the system state parameters at the next moment; but due to ignoring the uncertainty of wind speed, the method of only establishing economic model predictive control has certain limitations. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the deficiencies in the prior art, the present invention provides a method and system for intelligently optimizing the power generation of wind turbines based on load control, which has the advantages of accuracy, real-time, high efficiency and stability, and solves the problem of unstable power generation of wind turbines.

[0006] (II) Technical solution

[0007] In order to solve the technical problem of unstable power generation of the above-mentioned wind turbines, the present invention provides the following technical solutions:

[0008] This embodiment discloses a method for intelligently optimizing power generation of a wind turbine generator system based on load control, which specifically includes the following steps:

[0009] S1. Collecting historical wind speed data in the wind turbine generator set, and processing the collected historical wind speed data in the wind turbine generator set based on a data processing method to obtain processed historical wind speed data;

[0010] S2. constructing a wind speed probability model and a wind power model based on the processed historical wind speed data;

[0011] S3, collecting various historical parameters in the wind turbine generator set and constructing a relationship model between wind energy utilization coefficient and wind speed;

[0012] S4. Select some processed historical wind speed data, dynamically predict the wind power of the wind turbine based on the wind speed probability model and the random simulation method, and build a dynamic prediction model. At the same time, control the speed of the wind turbine based on the built dynamic prediction model and the relationship model between wind energy utilization coefficient and wind speed;

[0013] S5. Use a random sampling method to verify the accuracy of the constructed dynamic prediction model, and optimize the constructed dynamic prediction model based on the verification results.

[0014] The present invention collects historical wind speed data in the wind turbine group, processes the collected historical wind speed data in the wind turbine group based on a data processing method, and constructs a wind speed probability model and a wind power model based on the processed historical wind speed data, and then collects various historical parameters in the wind turbine group, constructs a wind energy utilization coefficient and wind speed relationship model, and at the same time, by selecting part of the processed historical wind speed data, dynamically predicts the wind power of the wind turbine group based on the wind speed probability model and a random simulation method, and constructs a dynamic prediction model, and at the same time, based on the constructed dynamic prediction model and the wind energy utilization coefficient and wind speed relationship model, the speed of the wind turbine group is regulated, and finally a random sampling method is used to verify the accuracy of the constructed dynamic prediction model, thereby improving the accuracy of the wind turbine group's power generation load control.

[0015] Preferably, the collecting of historical wind speed data in the wind turbine generator set and processing the collected historical wind speed data in the wind turbine generator set based on a data processing method to obtain the processed historical wind speed data comprises the following steps:

[0016] S11, construct a sample set X based on the historical wind speed data collected in the wind turbine, based on the constructed sample set X={ }, z = 1, 2, ..., m, where, represents the zth sample point in the sample set, and each sample point represents a set of historical wind speed data in a wind turbine. The threshold value of abnormal data proportion in the historical wind speed data in the wind turbine is set to And the neighborhood density threshold M of clustering;

[0017] S12. Initialize neighborhood density threshold M=k , calculate each sample point and its k The distance d between the neighborhood sample points, and the distance d The set of D , for each distance in the set d Probability statistics, D The data with cumulative probability in the range of 10% to 90% is recorded as a new set , and put The mathematical expectation of is taken as the neighborhood radius;

[0018] The neighborhood radius calculation formula is as follows:

[0019] ;

[0020] in, For collection The elements in Representing a collection The number of elements in , represents the neighborhood radius, represents mathematical expectation;

[0021] Based on neighborhood radius setting Neighborhood:

[0022] ;

[0023] in, express Neighborhood functions;

[0024] S13, performing clustering by a clustering algorithm based on the calculated neighborhood radius;

[0025] S14. Calculate the proportion of abnormal data in the historical wind speed data of the wind turbine and compare it with the set threshold. Compare, if it is less than or equal to the set threshold , indicating that the filtering is completed. If it is greater than the set threshold ,set up k = k +1, repeat steps S12 and S13;

[0026] S15, setting the historical wind speed data after filtering as the processed historical wind speed data.

[0027] Preferably, clustering by a clustering algorithm based on the calculated neighborhood radius comprises the following steps:

[0028] Mark all sample points as the set that has not been traversed and form the set u;

[0029] After marking is completed, a sample point in set u is randomly selected ;

[0030] Mark sample points Has been traversed;

[0031] When the sample point of There are at least k sample points, a new cluster is created , and the sample points Put into cluster middle;

[0032] Let Q be the sample point of The set of samples in the neighborhood, set point is a sample point in set Q;

[0033] If the sample point is not traversed;

[0034] Then mark the sample point Has been traversed, and traverses the sample points in sequence of Each sample point in the neighborhood;

[0035] When the sample point of There are at least k sample points, then add these points to the set Q;

[0036] When the sample point is not a member of any cluster, then Add to Cluster , and save the cluster ;

[0037] When the sample point of If there is no sample point in the neighborhood, mark the sample point is an outlier point;

[0038] Output the clusters after clustering and the marked abnormal points;

[0039] The abnormal point is set to represent abnormal data in the historical wind speed data within a group of wind turbines.

[0040] The present invention uses the collected historical wind speed data as sample points and calculates each sample point and its k The distance between the neighborhood sample points determines the neighborhood radius. At the same time, clustering is performed based on the determined neighborhood radius through a clustering algorithm to remove abnormal data and ensure the reliability of historical data.

[0041] Preferably, constructing a wind speed probability model and a wind power model based on the processed historical wind speed data comprises the following steps:

[0042] S21, constructing a wind power model based on the processed historical wind speed data;

[0043] The wind power model is as follows:

[0044] ;

[0045] in, Represents the output power of the wind turbine set. The output power of the wind turbine set is the power generated by the wind turbine set. represents the air density, Indicates the rotor radius of the wind turbine. Indicates wind speed, represents the wind energy utilization coefficient, represents the tip speed ratio of the wind turbine. Indicates the pitch angle of the wind turbine;

[0046] S22. Construct a wind speed probability model based on the processed historical wind speed data.

[0047] Preferably, constructing a wind speed probability model based on the processed historical wind speed data comprises the following steps:

[0048] The wind speed probability model is as follows:

[0049] ;

[0050] in, represents the wind speed probability, e represents the natural logarithm, represents the mean of historical wind speed data, represents the standard deviation of historical wind speed data, Indicates wind speed.

[0051] The present invention constructs a wind speed probability model and a wind power model through the processed historical wind speed data, establishes the relationship between wind speed and power, and completes the prediction of wind speed at the same time, thereby improving the efficiency of wind turbine generation calculation.

[0052] Preferably, the collecting of various historical parameters in the wind turbine generator set and constructing a wind energy utilization coefficient and wind speed relationship model comprises the following steps:

[0053] The internal losses of wind turbines are assumed to be negligible;

[0054] The relationship between the tip speed ratio of a wind turbine and the wind speed is shown below:

[0055] ;

[0056] in, Indicates the speed of wind turbine;

[0057] The relationship model between wind energy utilization coefficient and wind speed is shown below;

[0058] ;

[0059] in, Represents the tip speed ratio of the wind turbine at the i-th moment.

[0060] The present invention collects various historical parameters in the wind turbine and constructs a wind energy utilization coefficient and wind speed relationship model to further determine the factors affecting the power generation of the wind turbine, thereby ensuring the stability of the calculation of the power generation of the wind turbine.

[0061] Preferably, the selecting of partially processed historical wind speed data, dynamically predicting the wind power of the wind turbine generator set based on a wind speed probability model and a random simulation method, and constructing a dynamic prediction model, and regulating the speed of the wind turbine generator set based on the constructed dynamic prediction model and a wind energy utilization coefficient and wind speed relationship model comprises the following steps:

[0062] S41. Set the wind power set of the wind turbine group calculated based on the wind speed probability model to S={ , , ..., },in represents the g-th wind power in the wind power regulation process, and the regulation action set A={ , , ..., }in It represents the first control action under the first wind power in the process of wind power control. Each control action represents a set of control parameters of the wind turbine speed under wind power. represents the jth control action at the gth wind power in the wind power control process;

[0063] S42, calculating the initial probability based on the set wind power set and control action set of the wind turbine generator set, and calculating the power change probability at the next moment;

[0064] S43, collecting all possible wind power at the next moment to build a population;

[0065] Set the population set to:

[0066] Among them, B is the population set, is the g-th population;

[0067] S44, setting an objective function of a control parameter of a wind turbine speed based on the constructed population set;

[0068] Set the objective function: ;

[0069] in, The objective function representing the control parameters of the wind turbine speed, Indicates the change in the control parameters of the wind turbine speed;

[0070] S45, setting a stable range of wind power of the wind turbine generator set, iteratively calculating an objective function of a control parameter of a speed of the wind turbine generator set, and selecting an optimal control parameter of a speed of the wind turbine generator set;

[0071] S46. Based on the selected optimal wind turbine speed control parameters, the wind power of the wind turbine at the next moment is controlled, and based on the controlled wind turbine, the wind power of the wind turbine at the next moment is iteratively calculated to construct a dynamic prediction model.

[0072] Preferably, the step of calculating the initial probability based on the set wind power set and control action set of the wind turbine generator set, and calculating the power change probability at the next moment comprises the following steps:

[0073] Set the initial probability of each action in the control action set to be equal, initial probability = 1 / number of control actions that may occur under the current wind power;

[0074] The power change probability at the next moment is calculated as follows:

[0075] The probability distribution of power change at the next moment:

[0076] in, is the probability distribution of power change at the next moment, Indicates the wind power control process wind power After being processed by the reward function, the wind power is reached The probability of represents the reward function, , Indicates wind power Execute control actions The reward value at that time.

[0077] The present invention selects partially processed historical wind speed data, dynamically predicts the wind power of the wind turbine based on a wind speed probability model and a random simulation method, and constructs a dynamic prediction model. At the same time, based on the constructed dynamic prediction model and a wind energy utilization coefficient and wind speed relationship model, the speed of the wind turbine is regulated to complete the prediction of the power generation load control of the wind turbine, thereby improving the real-time performance of the power generation load control prediction of the wind turbine.

[0078] Preferably, the method of using a random sampling method to verify the accuracy of the constructed dynamic prediction model and optimizing the constructed dynamic prediction model according to the verification result comprises the following steps:

[0079] A random sampling method is used to select multiple groups of historical wind speed data at random moments and various parameters in the wind turbines at the corresponding moments, and the selected data are input into the constructed dynamic prediction model for accuracy verification;

[0080] The accuracy verification formula is as follows:

[0081] ;

[0082] Among them, TP represents the number of predictions that are consistent with the actual results; FP represents the number of predictions that are inconsistent with the actual results. is the accuracy rate;

[0083] Set an accuracy threshold. When the dynamic prediction model is less than the accuracy threshold, increase the range of selected data. When the dynamic prediction model is greater than or equal to the accuracy threshold, intelligently control the power generation of the wind turbine based on the current dynamic prediction model.

[0084] This embodiment also discloses a wind turbine generator set power generation intelligent optimization system based on load control, comprising: a data acquisition module, a data processing module, a wind power model construction module, a wind energy utilization coefficient analysis module, a dynamic prediction model construction module and an intelligent control module;

[0085] The data acquisition module is used to collect various historical parameter data in the wind turbine generator set, and transmit the collected various parameter data to the data processing module;

[0086] The data processing module is used to process the received parameter data and transmit the processed data to the wind power model building module;

[0087] The wind power model building module is used to build a wind power model according to the processed wind speed data;

[0088] The wind energy utilization coefficient analysis module is used to calculate the wind energy utilization coefficient under the current wind speed;

[0089] The dynamic prediction model building module is used to establish a dynamic prediction model based on historical data and output control parameters;

[0090] The intelligent control module is used to control the wind turbine generator set according to the control parameters output by the dynamic prediction model.

[0091] (III) Beneficial effects

[0092] Compared with the prior art, the present invention provides a method and system for intelligently optimizing the power generation of a wind turbine generator system based on load control, which has the following beneficial effects:

[0093] 1. The invention collects historical wind speed data in the wind turbine, processes the collected historical wind speed data in the wind turbine based on a data processing method, and constructs a wind speed probability model and a wind power model based on the processed historical wind speed data; then collects various historical parameters in the wind turbine, constructs a wind energy utilization coefficient and wind speed relationship model, and selects part of the processed historical wind speed data to dynamically predict the wind power of the wind turbine based on the wind speed probability model and a random simulation method, and constructs a dynamic prediction model. At the same time, based on the constructed dynamic prediction model and the wind energy utilization coefficient and wind speed relationship model, the speed of the wind turbine is regulated. Finally, a random sampling method is used to verify the accuracy of the constructed dynamic prediction model, thereby improving the accuracy of the wind turbine load control.

[0094] 2. The invention uses the collected historical wind speed data as sample points and calculates the relationship between each sample point and its k The distance between the neighborhood sample points determines the neighborhood radius. At the same time, clustering is performed based on the determined neighborhood radius through a clustering algorithm to remove abnormal data and ensure the reliability of historical data.

[0095] 3. The invention constructs a wind speed probability model and a wind power model through the processed historical wind speed data, establishes the relationship between wind speed and power, and completes the prediction of wind speed, thereby improving the efficiency of wind turbine power generation calculation.

[0096] 4. The invention collects various historical parameters in the wind turbine and constructs a model of the relationship between wind energy utilization coefficient and wind speed, thereby further determining the factors that affect the power generation of the wind turbine and ensuring the stability of the calculation of the power generation of the wind turbine.

[0097] 5. The invention selects some processed historical wind speed data, dynamically predicts the wind power of the wind turbine based on the wind speed probability model and the random simulation method, and constructs a dynamic prediction model. At the same time, based on the constructed dynamic prediction model and the wind energy utilization coefficient and wind speed relationship model, the wind turbine speed is regulated to complete the prediction of the wind turbine power generation load control, thereby improving the real-time performance of the wind turbine power generation load control prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0098] Figure 1 This is a schematic diagram of the structure of the intelligent optimization process of wind turbine power generation under load control of the present invention. DETAILED DESCRIPTION

[0099] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0100] Example 1

[0101] See also Figure 1 This embodiment discloses a method for intelligently optimizing power generation of a wind turbine generator system based on load control, which specifically includes the following steps:

[0102] S1. Collecting historical wind speed data in the wind turbine generator set, and processing the collected historical wind speed data in the wind turbine generator set based on a data processing method to obtain processed historical wind speed data;

[0103] Collecting historical wind speed data in a wind turbine generator set, and processing the collected historical wind speed data in the wind turbine generator set based on a data processing method to obtain the processed historical wind speed data includes the following steps:

[0104] S11, construct a sample set X based on the historical wind speed data collected in the wind turbine, based on the constructed sample set X={ }, z = 1, 2, ..., m, where, represents the zth sample point in the sample set, and each sample point represents a set of historical wind speed data in a wind turbine. The threshold value of abnormal data proportion in the historical wind speed data in the wind turbine is set to And the neighborhood density threshold M of clustering;

[0105] S12. Initialize neighborhood density threshold M=k , calculate each sample point and its k The distance d between the neighborhood sample points, and the distance d The set of D , for each distance in the set d Probability statistics, D The data with cumulative probability in the range of 10% to 90% is recorded as a new set , and put The mathematical expectation of is taken as the neighborhood radius;

[0106] The neighborhood radius calculation formula is as follows:

[0107] ;

[0108] in, For collection The elements in Representing a collection The number of elements in , represents the neighborhood radius, represents mathematical expectation;

[0109] Furthermore, based on the neighborhood radius setting Neighborhood:

[0110] ;

[0111] in, express Neighborhood functions;

[0112] S13, performing clustering by a clustering algorithm based on the calculated neighborhood radius;

[0113] Mark all sample points as the set that has not been traversed and form the set u;

[0114] After marking is completed, a sample point in set u is randomly selected ;

[0115] Mark sample points Has been traversed;

[0116] When the sample point of There are at least k sample points, a new cluster is created , and the sample points Put into cluster middle;

[0117] Furthermore, let Q be the sample point of The set of samples in the neighborhood, set point is a sample point in set Q;

[0118] If the sample point is not traversed;

[0119] Then mark the sample point Has been traversed, and traverses the sample points in sequence of Each sample point in the neighborhood;

[0120] When the sample point of There are at least k sample points, then add these points to the set Q;

[0121] When the sample point is not a member of any cluster, then Add to Cluster , and save the cluster ;

[0122] When the sample point of If there is no sample point in the neighborhood, mark the sample point is an outlier point;

[0123] Furthermore, the clusters after clustering and the marked abnormal points are output;

[0124] The abnormal point is set to represent abnormal data in the historical wind speed data within a group of wind turbines;

[0125] S14. Calculate the proportion of abnormal data in the historical wind speed data of the wind turbine and compare it with the set threshold. Compare, if it is less than or equal to the set threshold , indicating that the filtering is completed. If it is greater than the set threshold ,set up k = k +1, repeat steps S12 and S13;

[0126] S15, setting the historical wind speed data after filtering as the processed historical wind speed data;

[0127] S2. constructing a wind speed probability model and a wind power model based on the processed historical wind speed data;

[0128] S21, constructing a wind power model based on the processed historical wind speed data;

[0129] The wind power model is as follows:

[0130] ;

[0131] in, Represents the output power of the wind turbine set. The output power of the wind turbine set is the power generated by the wind turbine set. represents the air density, Indicates the radius of the wind turbine rotor. Indicates wind speed, represents the wind energy utilization coefficient, represents the tip speed ratio of the wind turbine. Indicates the pitch angle of the wind turbine;

[0132] S22, constructing a wind speed probability model based on the processed historical wind speed data;

[0133] The wind speed probability model is as follows:

[0134] ;

[0135] in, represents the wind speed probability, e represents the natural logarithm, represents the mean of historical wind speed data, represents the standard deviation of historical wind speed data, Indicates wind speed;

[0136] S3, collecting various historical parameters in the wind turbine generator set and constructing a relationship model between wind energy utilization coefficient and wind speed;

[0137] The internal losses of wind turbines are assumed to be negligible;

[0138] The relationship between the tip speed ratio of a wind turbine and the wind speed is shown below:

[0139] ;

[0140] in, Indicates the speed of wind turbine;

[0141] The relationship model between wind energy utilization coefficient and wind speed is shown below;

[0142] ;

[0143] in, represents the tip speed ratio of the wind turbine at the i-th moment;

[0144] S4. Select some processed historical wind speed data, dynamically predict the wind power of the wind turbine based on the wind speed probability model and the random simulation method, and build a dynamic prediction model. At the same time, control the speed of the wind turbine based on the built dynamic prediction model and the relationship model between wind energy utilization coefficient and wind speed;

[0145] S41. Set the wind power set of the wind turbine group calculated based on the wind speed probability model to S={ , , ..., },in represents the g-th wind power in the wind power regulation process, and the regulation action set A={ , , ..., }in It represents the first control action under the first wind power in the process of wind power control. Each control action represents a set of control parameters of the wind turbine speed under wind power. represents the jth control action at the gth wind power in the wind power control process;

[0146] S42, calculating the initial probability based on the set wind power set and control action set of the wind turbine generator set, and calculating the power change probability at the next moment;

[0147] Set the initial probability of each action in the control action set to be equal, initial probability = 1 / number of control actions that may occur under the current wind power;

[0148] The power change probability at the next moment is calculated as follows:

[0149] The probability distribution of power change at the next moment:

[0150] in, is the probability distribution of power change at the next moment, Indicates the wind power control process wind power After being processed by the reward function, the wind power is reached The probability of represents the reward function, , Indicates wind power Execute control actions The reward value at time

[0151] S43, collecting all possible wind power at the next moment to build a population;

[0152] Set the population set to:

[0153] Among them, B is the population set, is the g-th population;

[0154] S44, setting an objective function of a control parameter of a wind turbine speed based on the constructed population set;

[0155] Set the objective function: ;

[0156] in, The objective function representing the control parameters of the wind turbine speed, Indicates the change in the control parameters of the wind turbine speed;

[0157] S45, setting a stable range of wind power of the wind turbine generator set, iteratively calculating an objective function of a control parameter of a speed of the wind turbine generator set, and selecting an optimal control parameter of a speed of the wind turbine generator set;

[0158] S46, regulating the wind power of the wind turbine at the next moment based on the selected optimal wind turbine speed regulation parameter, and iteratively calculating the wind power of the wind turbine at the next moment based on the regulated wind turbine, to construct a dynamic prediction model;

[0159] S5. Use a random sampling method to verify the accuracy of the constructed dynamic prediction model, and optimize the constructed dynamic prediction model according to the verification results;

[0160] A random sampling method is used to select multiple groups of historical wind speed data at random moments and various parameters in the wind turbines at the corresponding moments, and the selected data are input into the constructed dynamic prediction model for accuracy verification;

[0161] The accuracy verification formula is as follows:

[0162] ;

[0163] Among them, TP represents the number of predictions that are consistent with the actual results; FP represents the number of predictions that are inconsistent with the actual results. is the accuracy rate;

[0164] Furthermore, an accuracy threshold is set. When the dynamic prediction model is constructed less than the accuracy threshold, the selected data range is increased. When the dynamic prediction model is constructed greater than or equal to the accuracy threshold, the power generation of the wind turbine is intelligently controlled based on the current dynamic prediction model.

[0165] Example 2

[0166] This embodiment also discloses a wind turbine generator set power generation intelligent optimization system based on load control, comprising: a data acquisition module, a data processing module, a wind power model construction module, a wind energy utilization coefficient analysis module, a dynamic prediction model construction module and an intelligent control module;

[0167] The data acquisition module is used to collect various historical parameter data in the wind turbine generator set, and transmit the collected various parameter data to the data processing module;

[0168] The data processing module is used to process the received parameter data and transmit the processed data to the wind power model building module;

[0169] The wind power model building module is used to build a wind power model according to the processed wind speed data;

[0170] The wind energy utilization coefficient analysis module is used to calculate the wind energy utilization coefficient under the current wind speed;

[0171] The dynamic prediction model building module is used to establish a dynamic prediction model based on historical data and output control parameters;

[0172] The intelligent control module is used to control the wind turbine generator set according to the control parameters output by the dynamic prediction model.

[0173] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent optimization of wind turbine power generation based on load control, characterized in that: The following steps are involved: S1. Collecting historical wind speed data in the wind turbine generator set, and processing the collected historical wind speed data in the wind turbine generator set based on a data processing method to obtain processed historical wind speed data; S2. constructing a wind speed probability model and a wind power model based on the processed historical wind speed data; S3, collecting various historical parameters in the wind turbine generator set and constructing a relationship model between wind energy utilization coefficient and wind speed; S4. Select some processed historical wind speed data, dynamically predict the wind power of the wind turbine based on the wind speed probability model and the random simulation method, and build a dynamic prediction model. At the same time, control the speed of the wind turbine based on the built dynamic prediction model and the relationship model between wind energy utilization coefficient and wind speed; S5. Use a random sampling method to verify the accuracy of the constructed dynamic prediction model, and optimize the constructed dynamic prediction model according to the verification results; The method of selecting a portion of the processed historical wind speed data, dynamically predicting the wind power of the wind turbine generator set based on the wind speed probability model and the random simulation method, and constructing a dynamic prediction model, and regulating the speed of the wind turbine generator set based on the constructed dynamic prediction model and the wind energy utilization coefficient and wind speed relationship model includes the following steps: S41, setting the wind power set of the wind turbine group calculated based on the wind speed probability model to S = {s1, s2, ..., s g }, where s g represents the g-th wind power in the wind power regulation process, and the regulation action set A = {a 11 , a 12 , ..., a gj }where a 11 represents the first control action under the first wind power in the wind power control process. Each control action represents a set of control parameters of the wind turbine speed under the wind power. gj represents the jth control action at the gth wind power in the wind power control process; S42, calculating the initial probability based on the set wind power set and control action set of the wind turbine generator set, and calculating the power change probability at the next moment; S43, collecting all possible wind power at the next moment to build a population; Set the population set to: B = {b1, b2, ..., b g } Among them, B is the population set, b g is the g-th population; S44, setting an objective function of a control parameter of a wind turbine speed based on the constructed population set; Set the objective function: Among them, Fit represents the objective function of the control parameters of the wind turbine speed, and Δω represents the change of the control parameters of the wind turbine speed; S45, setting a stable range of wind power of the wind turbine generator set, iteratively calculating an objective function of a control parameter of a speed of the wind turbine generator set, and selecting an optimal control parameter of a speed of the wind turbine generator set; S46. Based on the selected optimal wind turbine speed control parameters, the wind power of the wind turbine at the next moment is controlled, and based on the controlled wind turbine, the wind power of the wind turbine at the next moment is iteratively calculated to construct a dynamic prediction model.

2. The method for intelligent optimization of wind turbine power generation based on load control according to claim 1 is characterized in that: The collecting of historical wind speed data in the wind turbine generator set and processing the collected historical wind speed data in the wind turbine generator set based on the data processing method to obtain the processed historical wind speed data comprises the following steps: S11, constructing a sample set X based on the historical wind speed data collected in the wind turbine, based on the constructed sample set X = {x1, x2, ..., x z }, z = 1, 2, ..., m, where x z represents the zth sample point in the sample set, and each sample point represents a group of historical wind speed data in a wind turbine. The threshold of abnormal data proportion in the historical wind speed data in the wind turbine is set to σ and the neighborhood density threshold of clustering is set to M. S12, initialize the neighborhood density threshold M=k, calculate the distance d between each sample point and its k neighboring sample points, and record the set of distances d as D, perform probability statistics on each distance d in the set, and record the data in D with a cumulative probability in the range of 10% to 90% as the new set D * , and D * The mathematical expectation of is taken as the neighborhood radius; The neighborhood radius calculation formula is as follows: in, For the set D * The elements in the set D * The number of elements in , ε represents the neighborhood radius, and E represents the mathematical expectation; Set the ε neighborhood based on the neighborhood radius: N ε (x1)={x z ∈X|d(x1,x z )<ε}; Among them, N ε represents the ε neighborhood function; S13, performing clustering by a clustering algorithm based on the calculated neighborhood radius; S14, calculate the proportion of abnormal data in the historical wind speed data in the wind turbine, and compare it with the set threshold value σ. If it is less than or equal to the set threshold value σ, it means that the filtering is completed. If it is greater than the set threshold value σ, set k=k+1 and repeat steps S12 and S13; S15, setting the historical wind speed data after filtering as the processed historical wind speed data.

3. The method for intelligent optimization of wind turbine power generation based on load control according to claim 2 is characterized in that: The clustering based on the calculated neighborhood radius by using a clustering algorithm comprises the following steps: Mark all sample points as the set of untraversed samples u; After marking is completed, a sample point y is randomly selected from the set u; Mark sample point y as traversed; When there are at least k sample points in the ε neighborhood of the sample point y, a new cluster is created And put the sample point y into the cluster middle; Let Q be the set of samples in the ε neighborhood of sample point y, and set point y' as the sample point in set Q; If the sample point y' is not traversed; Then mark the sample point y' as traversed, and traverse each sample point in the ε neighborhood of the sample point y' in turn; When there are at least k sample points in the ε neighborhood of the sample point y', these points are added to the set Q; When the sample point y' is not a member of any cluster, add y' to the cluster And save the cluster When there is no sample point in the ε neighborhood of sample point y, the sample point y is marked as an outlier; Output the clusters after clustering and the marked abnormal points; The abnormal point is set to represent abnormal data in the historical wind speed data within a group of wind turbines.

4. The method for intelligent optimization of wind turbine power generation based on load control according to claim 1 is characterized in that: The construction of a wind speed probability model and a wind power model based on the processed historical wind speed data comprises the following steps: S21, constructing a wind power model based on the processed historical wind speed data; The wind power model is as follows: Among them, P w represents the output power of the wind turbine, which is the power generation of the wind turbine, ρ represents the air density, R represents the radius of the wind turbine rotor, v represents the wind speed, C P (λ,γ) represents the wind energy utilization coefficient, λ represents the tip speed ratio of the wind turbine, and γ represents the pitch angle of the wind turbine; S22. Construct a wind speed probability model based on the processed historical wind speed data.

5. The method for intelligent optimization of wind turbine power generation based on load control according to claim 4 is characterized in that: The process of constructing a wind speed probability model based on the processed historical wind speed data comprises the following steps: The wind speed probability model is as follows: Among them, f(v) represents the wind speed probability, e represents the natural logarithm, α represents the mean of the historical wind speed data, β represents the standard deviation of the historical wind speed data, and v represents the wind speed.

6. The method for intelligent optimization of wind turbine power generation based on load control according to claim 1, characterized in that: The collecting of various historical parameters in the wind turbine generator set and the construction of a wind energy utilization coefficient and wind speed relationship model comprises the following steps: The internal losses of wind turbines are assumed to be negligible; The relationship between the tip speed ratio of a wind turbine and the wind speed is shown below: Among them, ω represents the speed of wind turbine; The relationship model between wind energy utilization coefficient and wind speed is shown below; Among them, λ i Represents the tip speed ratio of the wind turbine at the i-th moment.

7. The method for intelligent optimization of wind turbine power generation based on load control according to claim 1, characterized in that: The method of calculating the initial probability based on the set wind power set and control action set of the wind turbine generator set, and calculating the power change probability at the next moment includes the following steps: Set the initial probability of each action in the control action set to be equal, initial probability = 1 / the number of control actions that may occur under the current wind power; The power change probability at the next moment is calculated as follows: The probability distribution of power change at the next moment: W:S×H,W(s1,a 11 ,s2):s1×H(s1,a 11 )→s2 Among them, W is the probability distribution of power change at the next moment, W(s1,a 11 ,s2) represents the probability of wind power s1 reaching wind power s2 after being processed by the reward function during the wind power control process. H represents the reward function, H = S × A, H (s1, a 11 )=s1×a 11 ,H(s1,a 11 ) indicates that wind power s1 performs control action a 11 The reward value at that time.

8. The method for intelligent optimization of wind turbine power generation based on load control according to claim 1 is characterized in that: The method of using a random sampling method to verify the accuracy of the constructed dynamic prediction model and optimizing the constructed dynamic prediction model according to the verification result includes the following steps: A random sampling method is used to select multiple groups of historical wind speed data at random moments and various parameters in the wind turbines at the corresponding moments, and the selected data are input into the constructed dynamic prediction model for accuracy verification; The accuracy verification formula is as follows: Among them, TP represents the number of predictions that are consistent with the actual results; FP represents the number of predictions that are inconsistent with the actual results, and F is the accuracy rate; Set an accuracy threshold. When the dynamic prediction model is less than the accuracy threshold, increase the range of selected data. When the dynamic prediction model is greater than or equal to the accuracy threshold, intelligently control the power generation of the wind turbine based on the current dynamic prediction model.

9. A system for implementing the method for intelligent optimization of wind turbine power generation based on load control as described in any one of claims 1 to 8, characterized in that: include: Data acquisition module, data processing module, wind power model building module, wind energy utilization coefficient analysis module, dynamic prediction model building module and intelligent control module; The data acquisition module is used to collect various historical parameter data in the wind turbine generator set, and transmit the collected various parameter data to the data processing module; The data processing module is used to process the received parameter data and transmit the processed data to the wind power model building module; The wind power model building module is used to build a wind power model according to the processed wind speed data; The wind energy utilization coefficient analysis module is used to calculate the wind energy utilization coefficient under the current wind speed; The dynamic prediction model building module is used to establish a dynamic prediction model based on historical data and output control parameters; The intelligent control module is used to control the wind turbine generator set according to the control parameters output by the dynamic prediction model.

Images