Generating capacity increasing method based on yaw operation data of draught fan

Abstract

The invention provides a generating capacity increasing method based on fan yaw operation data. The generating capacity increasing method comprises the steps that S1, wind power plant data are acquired; s2, establishing a wind power plant unit power model based on wake flow; s3, calculating by using a least square method SLSQP to obtain an optimal yaw angle of each wind turbine generator; s4, establishing a prediction sample according to the wind power plant data in the S1, and forming a training sample set; s5, inputting the input sample data into the support vector machine model; s6, solving a support vector machine model; and S7, verifying the optimal yaw angle, obtained by the least square method and the support vector machine, of each wind turbine generator, and when the difference between the output data of the support vector machine model and the output data of the least square method exceeds a set range, substituting the different data of the least square method into the support vector machine model to form a new training set, and carrying out optimization training on the support vector machine until different value domains of the data are within a set range. The yaw angle estimated by the least square method is checked by using the support vector machine, so that the defects that the least square method is easily influenced by abnormal values and the effect of processing a nonlinear function is poor are avoided, and the calculated result is more accurate and efficient.

Description

Technical Field

[0001] The present invention relates to the field of wind power technology, and in particular to a method for increasing power generation based on wind turbine yaw operation data. Background Art

[0002] As the global demand for renewable energy continues to grow, wind power generation, as an important component of clean energy, has received increasing attention for its technological development and application promotion. As the core equipment of the wind power generation system, the power generation efficiency and operating stability of wind turbines are directly related to the economic and environmental benefits of the entire wind farm. However, in actual operation, wind turbines face challenges from a variety of complex factors, among which the hardware performance differences of the units and the wind field wake effect are one of the key factors affecting power generation performance.

[0003] With the trend of large-scale wind turbines, in order to reduce costs and improve competitiveness, manufacturers and operators are constantly seeking technological innovation to improve the power generation efficiency and operational reliability of wind turbines. Especially in the context of current cost reduction pressure, how to make full use of existing wind turbine operation data to achieve refined management and optimization has become an important means to improve the performance of wind turbines.

[0004] Traditional wind turbine operation management mainly relies on regular maintenance and troubleshooting, which is not only inefficient but also difficult to reflect the actual operating status of the unit in real time. With the development of information technology and data analysis technology, methods based on big data and machine learning have begun to be applied to the field of wind power generation, realizing intelligent management and optimization of the unit through real-time monitoring and analysis of wind turbine operation data.

[0005] During the operation of wind turbines, the yaw angle is one of the key factors affecting power generation performance. The adjustment of the yaw angle can affect the efficiency of the wind rotor in capturing wind energy, and thus affect the power generation of the unit. However, due to differences in the hardware performance of the unit and the influence of the wind field wake effect, the optimal yaw angles of different units in different environments are not the same. Therefore, how to dynamically adjust the yaw angle according to the actual operation data of the unit to maximize the power generation performance has become a technical problem that needs to be solved urgently in the current wind power generation field.

[0006] At present, although there are some studies and practices on yaw control of wind turbines, most methods are still at the theoretical level or experimental stage and lack practical application verification. At the same time, these methods often ignore the influence of unit hardware performance differences and wind field wake effects, resulting in poor results in practical applications. Therefore, there is an urgent need for a method to increase power generation based on actual wind turbine operation data, which can maximize power generation performance by dynamically adjusting the yaw angle in response to the influence of unit hardware performance and wind field wake factors. Summary of the invention

[0007] The present invention provides a method for improving power generation based on wind turbine yaw operation data. In view of the influence of factors such as the hardware performance of the wind turbine and the wind field wake, the optimal yaw angle is found through the actual operation data of the wind turbine, which can effectively improve the power generation performance of the wind turbine throughout its life cycle.

[0008] In order to solve the above technical problems, the technical solution adopted by the present invention is: A method for increasing power generation based on wind turbine yaw operation data, comprising: S1. Obtain wind farm data: Obtain unit operation data from the wind farm SCADA system server, including wind speed, wind direction, temperature, air pressure, air density, impeller radius, yaw angle, and active power; S2. Establish a wind farm unit power model based on wake; S3, taking the maximum total power of the wind farm as the goal, taking the range of the maximum yaw angle and the minimum yaw angle as constraints, using the least square method SLSQP to calculate the optimal yaw angle of each wind turbine; S4, establishing prediction samples based on the wind farm data of S1 to form a training sample set; S5, inputting the input sample data into the support vector machine model; S6, solving the support vector machine model to obtain the optimal yaw angle of each wind turbine under the support vector machine model; S7. Verify the optimal yaw angle of each wind turbine obtained by the least squares method and the support vector machine. When the output data of the support vector machine model differs from the output data of the least squares method and exceeds the set range, substitute the different data of the least squares method into the support vector machine model to form a new training set, and perform optimization training of the support vector machine until the different value ranges of the data are within the set range.

[0009] The process of establishing the wind farm unit power model based on wake in S2 above is as follows: 3D wind speed The analytical expression is: ; ; ; ; In the formula, is the velocity difference at the center of the wake; is the free flow speed; for the wake deflection; is the fan hub height; is the width of the wake in the y direction; is the width of the wake in the z direction; is the thrust coefficient; The subscript “0” is the initial value at the beginning of the far wake, which depends on the ambient turbulence intensity. and thrust coefficient ; In addition to wind speed loss, wake deflection should also include the phenomenon of yaw deviation when the wind turbine is turning; the wake deflection angle caused by yaw deviation can be described as: ; In the formula, is the crew yaw angle; initial wake deflection Defined as: ; In the formula, is the length of the near wake; the total deflection of the wake caused by the yaw deviation is: ; It can be expressed as: ; Due to yaw misalignment, wake recovery is added, namely: = ; ; ; In the formula, is the effective wind speed of the unit; is the effective yaw wake speed of the crew; is the average speed; is the effective yaw angle; the total yaw angle is equal to the sum of the actual yaw offset of the turbine and the effective yaw angle; Wind turbine power for: ; In the formula, is the radius of the wind wheel; is the air density; is the wind energy utilization coefficient; is the wind speed; is the axial induction factor; there are: ; The above wind power coefficient can be corrected using a correction factor to account for the yaw deviation angle The impact of Indicates other losses of wind power coefficient; wind farm Typhoon generator output power for: ; in, is a constant.

[0010] The specific process of S3 mentioned above is: The optimization goal is to make the wind farm output power Maximum, that is: ; The optimal power of each wind turbine is set to H (MW), then the output power of the entire wind farm is The maximum value that can be converted into , set the yaw angle range of the wind turbine to [ ], within this range, multiple yaw angles are taken at uniform intervals and substituted into , then the yaw angle with the smallest value is the optimal one.

[0011] The specific process of S4 mentioned above is: Take wind speed, wind direction, temperature, air pressure, air density, impeller radius and active power as input quantities, and assume that the input and output sample data sets are (k=1,2,…,N), N is a natural number, where, is an n-dimensional vector, , represents a multi-digit real quantity, is the yaw angle of the wind turbine, , the linear equation in the feature space can be expressed as: ; in, : A mapping function that maps the input space to a high-risk feature space, is the weight vector of the hyperplane, ; b is the bias vector.

[0012] The specific process of S5 mentioned above is: Selection Error The sum of squares is the loss function, which is optimized as follows: ; in, is the objective function, is a constraint condition, N is a natural number, is the penalty coefficient factor, is the error, is the kth error, which is used to control the role of error adjustment. It can find a compromise between training error and model complexity so that the desired function has good generalization ability, and The larger the value, the smaller the regression error of the model.

[0013] The specific process of S6 mentioned above is: Determine the regression parameters of the support vector machine model = and b, introduce the Lagrangian function to solve: ; in, is the Lagrange multiplier; the optimal and b are obtained by the Carruth-Kuhn-Tucker t condition, namely: ; By eliminating the variables in the above formula and , the optimization problem is transformed into solving the following linear equations: ; In the formula, ; ; ; is the identity matrix; For square array; According to the Mercer condition, there exists a mapping and kernel function K( ) makes: ; The Substitute the values ​​of and b into the kernel width using the adaptive selection method. and the penalty factor , that is, the support vector machine model is obtained; Find After and b, the support vector machine model can be obtained as: ; The kernel function selected is the radial basis RBF function, that is, ; Substitute the input sample data set to obtain the output sample set of the support vector machine model , which is the optimal yaw angle of the wind turbine.

[0014] The present invention provides a method for improving power generation based on wind turbine yaw operation data, establishes a wake power generation model of a wind farm, uses the least squares method to obtain the yaw angle of maximum field total power, and then uses a support vector machine to verify the value of the yaw angle, thereby obtaining a more accurate yaw control value. Under the defect that the least squares method is not convenient for nonlinear function calculation, the support vector machine is used to make up for the nonlinear model optimization method of the model, so that a more suitable yaw control can be obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] 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

[0016] 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.

[0017] Embodiment 1: like Figure 1 As shown in , a method for improving power generation based on wind turbine yaw operation data includes: S1. Obtain wind farm data: Obtain unit operation data from the wind farm SCADA system server, including wind speed, wind direction, temperature, air pressure, air density, impeller radius, yaw angle, and active power; S2. Establish a wind farm unit power model based on wake; S3, taking the maximum total power of the wind farm as the goal, taking the range of the maximum yaw angle and the minimum yaw angle as constraints, using the least square method SLSQP to calculate the optimal yaw angle of each wind turbine; S4, establishing prediction samples based on the wind farm data of S1 to form a training sample set; S5, inputting the input sample data into the support vector machine model; S6, solving the support vector machine model to obtain the optimal yaw angle of each wind turbine under the support vector machine model; S7. Verify the optimal yaw angle of each wind turbine obtained by the least squares method and the support vector machine. When the output data of the support vector machine model differs from the output data of the least squares method and exceeds the set range, substitute the different data of the least squares method into the support vector machine model to form a new training set, and then perform optimization training of the support vector machine until the different value ranges of the data are within the set range.

[0018] When the least squares method is used in the wind farm unit power model established by the present invention, when there are outliers in the data, it will be affected by the outliers. The least squares method finds the best fit by minimizing the square of the error. The outliers will significantly increase the total error, causing the calculation results to deviate from the true value. In addition, since the model is not a pure linear function, the least squares method is not effective in processing. At this time, a support vector machine is used to compensate for it. The Lagrange multiplier of the support vector machine is used to avoid being affected by the outliers, and its feature of being good at processing nonlinear functions is used to compensate for the least squares method.

[0019] The process of establishing the wind farm unit power model based on wake in S2 above is as follows: 3D wind speed The analytical expression is: ; ; ; ; In the formula, is the velocity difference at the center of the wake; is the free flow speed; for the wake deflection; is the fan hub height; is the width of the wake in the y direction; is the width of the wake in the z direction; is the thrust coefficient; The subscript “0” is the initial value at the beginning of the far wake, which depends on the ambient turbulence intensity. and thrust coefficient ; In addition to wind speed loss, wake deflection should also include the phenomenon of yaw deviation when the wind turbine is turning; the wake deflection angle caused by yaw deviation can be described as: ; In the formula, is the crew yaw angle; initial wake deflection Defined as: ; In the formula, is the length of the near wake, which is about 3D; D is the diameter of the wind rotor; the total deflection of the wake caused by the yaw deviation is: ; It can be expressed as: ; Due to yaw misalignment, wake recovery is added, namely: = ; ; ; In the formula, is the effective wind speed of the unit; is the effective yaw wake speed of the crew; is the average speed; is the effective yaw angle; the total yaw angle is equal to the sum of the actual yaw offset of the turbine and the effective yaw angle; Wind turbine power for: ; In the formula, is the radius of the wind wheel; is the air density; is the wind energy utilization coefficient; is the wind speed; is the axial induction factor; there are: ; The above wind power coefficient can be corrected using a correction factor to account for the yaw deviation angle The impact of Indicates other losses of wind power coefficient; wind farm Typhoon generator output power for: ; in, is a constant.

[0020] The specific process of S3 mentioned above is: The optimization goal is to make the wind farm output power Maximum, that is: ; The optimal power of each wind turbine is set to H (MW), then the output power of the entire wind farm is The maximum value that can be converted into , set the yaw angle range of the wind turbine to [ ], within this range, multiple yaw angles are taken at uniform intervals and substituted into , then the yaw angle with the smallest value is the optimal one.

[0021] The specific process of S4 mentioned above is: Take wind speed, wind direction, temperature, air pressure, air density, impeller radius and active power as input quantities, and assume that the input and output sample data sets are (k=1,2,…,N), N is a natural number, where, is an n-dimensional vector, , represents a multi-digit real quantity, is the yaw angle of the wind turbine, , the linear equation in the feature space can be expressed as: ; in, : A mapping function that maps the input space to a high-risk feature space, is the weight vector of the hyperplane, ; b is the bias vector.

[0022] The specific process of S5 mentioned above is: Selection Error The sum of squares is the loss function, which is optimized as follows: ; in, is the objective function, is a constraint condition, N is a natural number, is the penalty coefficient factor, is the error, is the kth error, which is used to control the role of error adjustment. It can find a compromise between training error and model complexity so that the desired function has good generalization ability, and The larger the value, the smaller the regression error of the model.

[0023] The specific process of S6 mentioned above is: Determine the regression parameters of the support vector machine model = and b, introduce the Lagrangian function to solve: ; in, is the Lagrange multiplier; the optimal and b are obtained by the Carruth-Kuhn-Tucker t condition, namely: ; By eliminating the variables in the above formula and , the optimization problem is transformed into solving the following linear equations: ; In the formula, ; ; ; is the identity matrix; For square array; According to the Mercer condition, there exists a mapping and kernel function K( ) makes: ; The Substitute the values ​​of and b into the kernel width using the adaptive selection method. and the penalty factor , that is, the support vector machine model is obtained; Find After and b, the support vector machine model can be obtained as: ; The kernel function selected is the radial basis RBF function, that is, ; Substitute the input sample data set to obtain the output sample set of the support vector machine model , which is the optimal yaw angle of the wind turbine.

Claims

1. A method for increasing power generation based on wind turbine yaw operation data, characterized in that: include: S1. Obtain wind farm data; Obtain unit operation data from the wind farm SCADA system server, including wind speed, wind direction, temperature, air pressure, air density, impeller radius, yaw angle, and active power; S2. Establish a wind farm unit power model based on wake; S3, taking the maximum total power of the wind farm as the goal, taking the range of the maximum yaw angle and the minimum yaw angle as constraints, using the least square method SLSQP to calculate the optimal yaw angle of each wind turbine; S4, establishing prediction samples based on the wind farm data of S1 to form a training sample set; S5, inputting the input sample data into the support vector machine model; S6, solving the support vector machine model to obtain the optimal yaw angle of each wind turbine under the support vector machine model; S7. Verify the optimal yaw angle of each wind turbine obtained by the least squares method and the support vector machine. When the output data of the support vector machine model differs from the output data of the least squares method and exceeds the set range, substitute the different data of the least squares method into the support vector machine model to form a new training set, and perform optimization training of the support vector machine until the different value ranges of the data are within the set range.

2. A method for increasing power generation based on wind turbine yaw operation data according to claim 1, characterized in that: The process of establishing the wind farm unit power model based on wake in S2 is as follows: 3D wind speed The analytical expression is: ； ； ； ； In the formula, is the velocity difference at the center of the wake; is the free flow speed; for wake deflection; is the fan hub height; is the width of the wake in the y direction; is the width of the wake in the z direction; is the thrust coefficient; The subscript "0" is the initial value at the beginning of the far wake, which depends on the ambient turbulence intensity. and thrust coefficient ; In addition to the wind speed deficit, the wake deflection should also include the phenomenon that the wind turbine yaws when performing a turn. The wake deflection angle caused by the yaw deviation can be described as: ； In the formula, is the crew yaw angle; initial wake deflection Defined as: ； In the formula, is the length of the near wake; the total deflection of the wake caused by the yaw deviation is: ； It can be expressed as: ； Due to yaw misalignment, wake recovery is added, namely: = ； ； ； In the formula, is the effective wind speed of the unit; is the effective yaw wake speed of the crew; is the average speed; is the effective yaw angle; the total yaw angle is equal to the sum of the actual yaw offset of the turbine and the effective yaw angle; Wind turbine power for: ； In the formula, is the radius of the wind wheel; is the air density; is the wind energy utilization coefficient; is the wind speed; is the axial induction factor; there are: ； The above wind power coefficient can be corrected using a correction factor to account for the yaw deviation angle The impact of Indicates other losses of wind power coefficient; wind farm Typhoon generator output power for: ； in, is a constant.

3. A method for increasing power generation based on wind turbine yaw operation data according to claim 2, characterized in that: The specific process of S3 is as follows: The optimization goal is to make the wind farm output power Maximum, that is: ； The optimal power of each wind turbine is set to H (MW), then the output power of the entire wind farm is The maximum value that can be converted into , set the yaw angle range of the wind turbine to [ ], within this range, multiple yaw angles are taken at uniform intervals and substituted into , then the yaw angle with the smallest value is the optimal one.

4. A method for increasing power generation based on wind turbine yaw operation data according to claim 3, characterized in that: The specific process of S4 is as follows: Take wind speed, wind direction, temperature, air pressure, air density, impeller radius and active power as input quantities, and assume that the input and output sample data sets are (k=1,2,…,N), N is a natural number, where, is an n-dimensional vector, , represents a multi-digit real quantity, is the yaw angle of the wind turbine, , the linear equation in the feature space can be expressed as: ； in, : A mapping function that maps the input space to a high-risk feature space, is the weight vector of the hyperplane, ; b is the bias vector.

5. A method for increasing power generation based on wind turbine yaw operation data according to claim 4, characterized in that: The specific process of S5 is as follows: Selection Error The sum of squares is the loss function, which is optimized as follows: ； in, is the objective function, is a constraint condition, N is a natural number, is the penalty coefficient factor, is the error, is the kth error, which is used to control the role of error adjustment. It can find a compromise between training error and model complexity so that the desired function has good generalization ability, and The larger the value, the smaller the regression error of the model.

6. A method for increasing power generation based on wind turbine yaw operation data according to claim 5, characterized in that: The specific process of S6 is as follows: Determine the regression parameters of the support vector machine model = and b, introduce the Lagrangian function to solve: ； in, is the Lagrange multiplier; the optimal and b are obtained by the Carruth-Kuhn-Tucker t condition, namely: ； By eliminating the variables in the above formula and , the optimization problem is transformed into solving the following linear equations: ； In the formula, ; ; ; is the identity matrix; For square array; According to the Mercer condition, there exists a mapping and kernel function K( ) makes: ； The Substitute the values ​​of and b into the kernel width using the adaptive selection method. and the penalty factor , that is, the support vector machine model is obtained; Find After and b, the support vector machine model can be obtained as: ； The kernel function selected is the radial basis RBF function, that is, ； Substitute the input sample data set to obtain the output sample set of the support vector machine model , which is the optimal yaw angle of the wind turbine.

Images