Wind turbine generator output optimization control method considering fatigue load distribution

Abstract

The invention provides a wind turbine generator output optimization control method considering fatigue load distribution, and the method is characterized in that the method comprises the steps: S1, obtaining the historical data of a wind turbine generator; s2, obtaining a weather feature type of the day; s3, standardizing the weather feature type data; s4, establishing a target function of comprehensive benefits of the wind turbine generator; s5, dividing the standardized clustering data in the S3 into a training set and a test set; s6, constructing an LSMT model, and carrying out training verification; and S7, as time goes on, optimizing the LSMT model. An advanced optimization algorithm is adopted to dynamically adjust the power output and the operation strategy of the wind turbine generator set, and it is ensured that optimal energy capture and load distribution are achieved under different wind conditions. By optimizing the blade angle, the generator speed and variable pitch control, the system can effectively reduce the influence of unfavorable load, prolong the service life of the unit and increase the comprehensive benefits.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power, and in particular to a method for optimizing the output control of a wind turbine generator set taking fatigue load distribution into consideration. Background Art

[0002] In the wind power industry, as the trend of large-scale wind turbines becomes increasingly prominent, it is possible to make full use of wind power resources and provide larger capacity of electricity. However, wind turbines will accumulate fatigue loads during operation. When fatigue loads accumulate to a certain extent, they will cause structural damage. In this way, a contradiction is formed between improving the utilization of wind energy and reasonably controlling the fatigue loads of the units. If the two cannot be reasonably controlled, the long-term benefits of the wind farm cannot be brought into play.

[0003] The current wind turbine control strategies are all combined with the active power generated by the unit. The most common control strategies include Maximum Power Point Tracking (MPPT) and Power Reference Tracking (PRT). The MPPT control strategy is the traditional active power control strategy for wind turbines. This strategy can be used when the dispatching system has no power constraints on the wind farm. In this strategy, the wind turbine tracks its maximum wind energy utilization coefficient to maximize the active power of the wind turbine, but it will increase the thrust coefficient of the wind turbine. The PRT control strategy is a modification of the MPPT control strategy. The wind turbine only needs to adjust the wind energy utilization coefficient away from its optimal value. Neither of them takes into account the fatigue load of the unit, that is, the long-term benefits of the wind turbine. Summary of the invention

[0004] The present invention provides a wind turbine output optimization control method considering fatigue load distribution, which adopts advanced optimization algorithm to dynamically adjust the power output and operation strategy of the wind turbine to ensure optimal energy capture and load distribution under different wind conditions.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: A wind turbine output optimization control method considering fatigue load distribution, comprising: S1. Obtain historical data of wind turbines; the historical data includes time, active power, wind speed, turbulence intensity, rotation speed, pitch angle, tip speed ratio, working time and air density and perform preprocessing; S2. Process the wind speed, turbulence intensity and air density in the historical data into daily average, maximum and minimum values ​​to represent the weather characteristic type of the day; S3, standardize the weather feature type data, and use the Kmeans algorithm to cluster the standardized data and divide the data set; S4. Establish a fatigue load model and a wind energy utilization model for wind turbines, and establish an objective function that combines the fatigue load model and the wind energy utilization model as an evaluation of the comprehensive benefits of wind turbines; S5, dividing the standardized cluster data in S3 into a training set and a test set; selecting the corresponding data set in the training set according to the weather type of the day, searching for similar power generation feature types for the day and historical data based on weighted grey correlation analysis, and constructing a new training set; S6. Construct the LSMT model, and use the fatigue load model and wind energy utilization model as input training, use the comprehensive benefit objective function as output training LSMT model, and use the grid search method to adjust the parameters of LSMT to obtain the optimal comprehensive benefit of the wind turbine, and obtain the optimal real-time variable pitch angle and blade tip speed ratio. Verify the effect by comparing with the data of the same weather type in the test set; S7. As time goes by, the daily data is continuously added to the training set to expand the number of training sets and optimize the LSMT model.

[0006] The specific method of S5 mentioned above includes: S5.1. Construct a sample sequence Xi(k1, k2, k3, k4) for each moment of the data set, where k1, k2, k3, k4 represent the wind speed, turbulence intensity, and air density data at moment i, respectively, and normalize the data to form a comparison series; S5.2. Calculate the weighted correlation degree for each feature, select the feature data at a certain moment as the reference sequence, and perform grey correlation analysis with all moments in the data set. The calculation formula for the sequence correlation coefficient between the comparison sequence and the reference sequence is as follows: ; Xi(k) is the reference sequence, y(k) is the comparison sequence, ξ i Representation elements k Correlation coefficient; min i min k | y ( k )- x i ( k )| is the minimum absolute difference between all comparison sequences and the reference sequence; similarly, max i max k | y ( k )- x i ( k )| is the maximum absolute difference between the sequences; S5.3. Since different weather characteristics have different effects on wind turbine output, weight calculation is introduced when calculating grey correlation. The specific steps are as follows: S5.3.1. Constructing a data matrix using raw dimensionless data ; Where m represents the number of features of each set of sample data, and n represents the sample size of the data; S5.3.2, determine the sample ideal value and negative ideal value; ; in ; S5.3.3. Use the Euclidean formula to calculate the distance between each element in a column of samples and the ideal value; , ; in Represents the Euclidean distance from the sample data to the ideal value, is the Euclidean distance from the data to the negative ideal value; S5.3.4, comprehensive distance calculation; ; Calculate the closeness of each solution to the ideal solution. The larger it is, the more important the element is in the sample data; S5.3.5, vector weight calculation; ; is the weight vector of the required sequence; S5.4, Grey correlation analysis: After calculating the correlation coefficient of each element in Xi(k), the grey correlation r can be calculated by the following formula: i : ; r i >0.7, it is considered as the first level of correlation; 0.5< r i <0.7, identified as the second level of correlation; r i <0.5, the third level of relevance, where α(k) is the weight corresponding to the element; S5.5. i The values ​​are sorted from large to small, and the three largest values ​​in the first-level correlation value range are taken to extract the corresponding power data and construct a new data set.

[0007] The fatigue load model in the above S4 is: The fatigue load factor is used to evaluate the fatigue load level of the wind turbine. Work fatigue caused by wind turbine power generation and turbulence fatigue caused by turbulence on the fan Two parts, as follows: ; ; ; In the formula, For fans Active power at the moment, is the time interval value, is the rated power of the fan, Design life of the fan. is the fan maintenance and repair coefficient, which represents the proportion of maintenance and repair time when the working time is 1. is the turbulence disturbance coefficient, For the fan Effective turbulence at all times; Wind turbines work hard in the process of capturing wind energy. and turbulence fatigue There is a correlation between the two, and the correlation coefficient between them is defined as: ; Therefore, the fan fatigue load factor can be calculated by the following formula: .

[0008] The wind energy utilization model in S4 mentioned above is The mechanical power that the wind wheel obtains from the wind in the direction of the main shaft transmission for: ; in, is the air density, is the radius of the wind rotor blade, for The wind speed at the moment, is the wind energy utilization coefficient, is the pitch angle of the propeller, is the tip speed ratio, The calculation formula is: ; in, for The wind wheel speed at the moment; Wind energy utilization factor is a nonlinear function: .

[0009] The objective function of the comprehensive benefits of the wind turbines in S4 mentioned above is: The comprehensive benefit Z of wind turbines is: ; in, Design life of the fan The total working time of the internal wind turbines, is the unit power gain, The maintenance and repair cost per unit time; The objective function is: ; in, is the final rate of return of the wind turbine.

[0010] After the completion of the above step S6, the fatigue load calculated over time is sent to the fault warning module, compared with the set unit fatigue load data, and different warnings are output according to the comparison results.

[0011] The present invention provides a method for optimizing the output control of a wind turbine generator set considering the distribution of fatigue loads, which aims to improve the power generation efficiency of the wind turbine generator set in a variable wind speed environment, while reducing the structural damage caused by fatigue loads. The method evaluates the fatigue state of each key component of the wind turbine generator set by real-time monitoring of multiple parameters such as wind speed, wind direction, unit vibration and temperature, combined with a fatigue load analysis model. Based on the acquired real-time data, the control system uses advanced optimization algorithms to dynamically adjust the power output and operation strategy of the wind turbine generator set to ensure optimal energy capture and load distribution under different wind conditions. By optimizing the blade angle, generator speed and pitch control, the system can effectively reduce the impact of adverse loads and extend the service life of the unit. In addition, the present invention also includes a fault warning module, which can identify potential structural faults in advance based on fatigue load data analysis, and provide maintenance suggestions to reduce downtime. The power generation efficiency of the wind turbine generator set can be improved through this comprehensive control strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The present invention will be further described below in conjunction with the accompanying drawings and embodiments: Figure 1 It is the inventive flow chart of the present method. DETAILED DESCRIPTION

[0013] In order to make the purpose, technical solutions and advantages of the present invention clearer, the following content will systematically and completely describe the specific technical solutions of the present invention in combination with the drawings provided according to the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0014] Embodiment 1: like Figure 1 As shown in , a wind turbine output optimization control method considering fatigue load distribution includes: S1. Obtain historical data of wind turbines; the historical data includes time, active power, wind speed, turbulence intensity, rotation speed, pitch angle, tip speed ratio, working time and air density and perform preprocessing; S2. Process the wind speed, turbulence intensity and air density in the historical data into daily average, maximum and minimum values ​​to represent the weather characteristic type of the day; S3, standardize the weather feature type data, and use the Kmeans algorithm to cluster the standardized data and divide the data set; According to the time information, wind speed level and turbulence, combined with the local historical weather forecast information, the weather feature types are divided into different level types; S4. Establish a fatigue load model and a wind energy utilization model for wind turbines, and establish an objective function that combines the fatigue load model and the wind energy utilization model as an evaluation of the comprehensive benefits of wind turbines; S5, dividing the standardized cluster data in S3 into a training set and a test set; selecting the corresponding data set in the training set according to the weather type of the day, searching for similar power generation feature types for the day and historical data based on weighted grey correlation analysis, and constructing a new training set; S6. Construct the LSMT model, and use the fatigue load model and wind energy utilization model as input training, use the comprehensive benefit objective function as output training LSMT model, and use the grid search method to adjust the parameters of LSMT to obtain the optimal comprehensive benefit of the wind turbine, and obtain the optimal real-time variable pitch angle and blade tip speed ratio. Verify the effect by comparing with the data of the same weather type in the test set; S7. As time goes by, the daily data is continuously added to the training set to expand the number of training sets and optimize the LSMT model.

[0015] Weather characteristic data determines the maximum power that the unit can generate on that day. Substituting other historical data into the fatigue load model, the load model can be converted into active power for calculation. The determining factors of wind energy utilization are the pitch angle and tip speed ratio. Through historical data, the active power that can be obtained by using different solutions under the current weather characteristics can be obtained to obtain fatigue load data. Substituting other data into the wind energy utilization formula can obtain the wind energy utilization of the day. Create an objective function to evaluate the comprehensive benefits of wind turbines. In S5, the active power obtained under different weather factors is calculated to obtain the weight of the weather factor. In addition to being determined by the weather, the actual load data is also determined by the pitch angle adopted and the speed obtained. The objective function established in S4 is continuously adjusted and optimized in the LSMT model to obtain the optimal output optimization. The actual data formed by the newly generated output optimization on that day is then imported into the training set to form new data. The continuous expansion of the data set makes the output optimization more beneficial to the comprehensive benefits of wind turbines.

[0016] The specific method of S5 mentioned above includes: S5.1. Construct a sample sequence Xi(k1, k2, k3, k4) for each moment of the data set, where k1, k2, k3, k4 represent the wind speed, turbulence intensity, and air density data at moment i, respectively, and normalize the data to form a comparison series; S5.2. Calculate the weighted correlation degree for each feature, select the feature data at a certain moment as the reference sequence, and perform grey correlation analysis with all moments in the data set. The calculation formula for the sequence correlation coefficient between the comparison sequence and the reference sequence is as follows: ; Xi(k) is the reference sequence, y(k) is the comparison sequence, ξ i Representation elements k Correlation coefficient; min i min k | y ( k )- x i ( k )| is the minimum absolute difference between all comparison sequences and the reference sequence; similarly, max i max k | y ( k )- x i ( k )| is the maximum absolute difference between the sequences; the resolution coefficient ρ is 0.4; S5.3. Since different weather characteristics have different effects on wind turbine output, weight calculation is introduced when calculating grey correlation. The specific steps are as follows: S5.3.1. Constructing a data matrix using raw dimensionless data ; Where m represents the number of features of each set of sample data, and n represents the sample size of the data; S5.3.2, determine the sample ideal value and negative ideal value; ; in ; S5.3.3. Use the Euclidean formula to calculate the distance between each element in a column of samples and the ideal value; , ; in Represents the Euclidean distance from the sample data to the ideal value, is the Euclidean distance from the data to the negative ideal value; S5.3.4, comprehensive distance calculation; ; Calculate the closeness of each solution to the ideal solution. The larger it is, the more important the element is in the sample data; S5.3.5, vector weight calculation; ; is the weight vector of the required sequence; S5.4, Grey correlation analysis: After calculating the correlation coefficient of each element in Xi(k), the grey correlation r can be calculated by the following formula: i : ; r i >0.7, it is considered as the first level of correlation; 0.5< r i <0.7, identified as the second level of correlation; r i <0.5, the third level of relevance, where α(k) is the weight corresponding to the element; S5.5. i The values ​​are sorted from large to small, and the three largest values ​​in the first-level correlation value range are taken to extract the corresponding power data and construct a new data set.

[0017] The fatigue load model in the above S4 is: The fatigue load factor is used to evaluate the fatigue load level of the wind turbine. Work fatigue caused by wind turbine power generation and turbulence fatigue caused by turbulence on the fan Two parts, as follows: ; ; ; In the formula, For fans Active power at the moment, is the time interval value, is the rated power of the fan, Design life of the fan. is the fan maintenance and repair coefficient, which represents the proportion of maintenance and repair time when the working time is 1. is the turbulence disturbance coefficient, For the fan Effective turbulence at all times; Wind turbines work hard in the process of capturing wind energy. and turbulent fatigue There is a correlation between the two, and the correlation coefficient between them is defined as: ; Therefore, the fan fatigue load factor can be calculated by the following formula: .

[0018] The wind energy utilization model in S4 mentioned above is The mechanical power that the wind wheel obtains from the wind in the direction of the main shaft transmission for: ; in, is the air density, is the radius of the wind rotor blade, for The wind speed at the moment, is the wind energy utilization coefficient, is the pitch angle of the propeller, is the tip speed ratio, The calculation formula is: ; in, for The wind wheel speed at the moment; Wind energy utilization factor is a nonlinear function: .

[0019] The objective function of the comprehensive benefits of the wind turbines in S4 mentioned above is: The comprehensive benefit Z of wind turbines is: ; in, Design life of the fan The total working time of the internal wind turbines, is the unit power gain, The maintenance and repair cost per unit time; The objective function is: ; in, is the final rate of return of the wind turbine.

[0020] After the completion of the above step S6, the fatigue load calculated over time is sent to the fault warning module, compared with the set unit fatigue load data, and different warnings are output according to the comparison results.

Claims

1. A wind turbine output optimization control method considering fatigue load distribution, characterized in that: include: S1. Obtain historical data of wind turbines; the historical data includes time, active power, wind speed, turbulence intensity, rotation speed, pitch angle, tip speed ratio, working time and air density and perform preprocessing; S2. Process the wind speed, turbulence intensity and air density in the historical data into daily average, maximum and minimum values ​​to represent the weather characteristic type of the day; S3, standardize the weather feature type data, and use the Kmeans algorithm to cluster the standardized data and divide the data set; S4. Establish a fatigue load model and a wind energy utilization model for wind turbines, and establish an objective function that combines the fatigue load model and the wind energy utilization model as an evaluation of the comprehensive benefits of wind turbines; S5, dividing the standardized cluster data in S3 into a training set and a test set; selecting the corresponding data set in the training set according to the weather type of the day, searching for similar power generation feature types for the day and historical data based on weighted grey correlation analysis, and constructing a new training set; S6. Construct the LSMT model, and use the fatigue load model and wind energy utilization model as input training, use the comprehensive benefit objective function as output training LSMT model, and use the grid search method to adjust the parameters of LSMT to obtain the optimal comprehensive benefit of the wind turbine, and obtain the optimal real-time variable pitch angle and blade tip speed ratio. Verify the effect by comparing with the data of the same weather type in the test set; S7. As time goes by, the daily data is continuously added to the training set to expand the number of training sets and optimize the LSMT model.

2. A wind turbine output optimization control method considering fatigue load distribution according to claim 1, characterized in that: The specific method of S5 includes: S5.

1. Construct a sample sequence Xi(k1, k2, k3, k4) for each moment of the data set, where k1, k2, k3, k4 represent the wind speed, turbulence intensity, and air density data at moment i, respectively, and normalize the data to form a comparison series; S5.

2. Calculate the weighted correlation degree for each feature, select the feature data at a certain moment as the reference sequence, and perform grey correlation analysis with all moments in the data set. The calculation formula for the sequence correlation coefficient between the comparison sequence and the reference sequence is as follows: ； Xi(k) is the reference sequence, y(k) is the comparison sequence, ξ i Representation elements k Correlation coefficient; min i min k | y ( k )- x i ( k )| is the minimum absolute difference between all comparison sequences and the reference sequence; similarly, max i max k | y ( k )- x i ( k )| is the maximum absolute difference between the sequences; S5.

3. Since different weather characteristics have different effects on wind turbine output, weight calculation is introduced when calculating grey correlation. The specific steps are as follows: S5.3.

1. Constructing a data matrix using raw dimensionless data ； Where m represents the number of features of each set of sample data, and n represents the sample size of the data; S5.3.2, determine the sample ideal value and negative ideal value; ； in ; S5.3.

3. Use the Euclidean formula to calculate the distance between each element in a column of samples and the ideal value; ， ； in Represents the Euclidean distance from the sample data to the ideal value, is the Euclidean distance from the data to the negative ideal value; S5.3.4, comprehensive distance calculation; ； Calculate the closeness of each solution to the ideal solution. The larger it is, the more important the element is in the sample data; S5.3.5, vector weight calculation; ； is the weight vector of the required sequence; S5.4, Grey correlation analysis: After calculating the correlation coefficient of each element in Xi(k), the grey correlation r can be calculated by the following formula: i : ； r i >0.7, it is considered as the first level of correlation; 0.5< r i <0.7, identified as the second level of correlation; r i <0.5, the third level of relevance, where α(k) is the weight corresponding to the element; S5.

5. i The values ​​are sorted from large to small, and the three largest values ​​in the first-level correlation value range are taken to extract the corresponding power data and construct a new data set.

3. A wind turbine output optimization control method considering fatigue load distribution according to claim 2, characterized in that: The fatigue load model in S4 is: The fatigue load factor is used to evaluate the fatigue load level of the wind turbine. Work fatigue caused by wind turbine power generation and turbulence fatigue caused by turbulence on the fan Two parts, as follows: ； ； ； In the formula, For fans Active power at the moment, is the time interval value, is the rated power of the fan, Design life of the fan. is the fan maintenance and repair coefficient, which represents the proportion of maintenance and repair time when the working time is 1. is the turbulence disturbance coefficient, For the fan Effective turbulence at all times; Wind turbines work hard in the process of capturing wind energy. and turbulent fatigue There is a correlation between the two, and the correlation coefficient between them is defined as: ； Therefore, the fan fatigue load factor can be calculated by the following formula: 。 4. A wind turbine output optimization control method considering fatigue load distribution according to claim 3, characterized in that: The wind energy utilization model in S4 is The mechanical power that the wind wheel obtains from the wind in the direction of the main shaft transmission for: ； in, is the air density, is the radius of the wind rotor blade, for The wind speed at the moment, is the wind energy utilization coefficient, is the pitch angle of the propeller, is the tip speed ratio, The calculation formula is: ； in, for The wind wheel speed at the moment; Wind energy utilization factor is a nonlinear function: 。 5. A wind turbine output optimization control method considering fatigue load distribution according to claim 4, characterized in that: The objective function of the comprehensive benefits of the wind turbines in S4 is: The comprehensive benefit Z of wind turbines is: ； in, Design life of the fan The total working time of the internal wind turbines, is the unit power gain, The maintenance and repair cost per unit time; The objective function is: ； in, is the final rate of return of the wind turbine.

6. A wind turbine output optimization control method considering fatigue load distribution according to claim 5, characterized in that: After the step S6 is completed, the fatigue load calculated over time is sent to the fault warning module, compared with the set unit fatigue load data, and different warnings are output according to the comparison results.

Images