CN-122021291-A - Wake model parameterized optimization method and system based on wind farm historical operation data

CN122021291ACN 122021291 ACN122021291 ACN 122021291ACN-122021291-A

Abstract

The invention relates to the technical field of wind power generation and discloses a wake model parameterized optimization method and a wake model parameterized optimization system based on wind power plant historical operation data, wherein the method comprises the steps of collecting wind power plant historical operation data, preprocessing the collected wind power plant historical operation data to obtain a training set and a testing set, constructing an engineering experience wake model based on an Ishihara single-machine wake model, establishing an optimization problem about a wake growth rate proportionality coefficient in the engineering experience wake model, solving by adopting a particle swarm optimization algorithm, taking average absolute error MAE between a full wind power plant power predicted value and a power actual measurement value in the training set as an adaptability function, and comparing the wake growth rate proportionality coefficient Optimizing the obtained product The method is applied to an engineering experience wake model and is used for power prediction of wind power plant groups. The method can obviously reduce the power prediction error of the wake model, and realize the reduction of the power prediction MAE of the offshore wind farm group and the improvement of annual energy production.

Inventors

ZHAO JIANJIAN
ZHOU FENGFENG
LIU ZUOREN
HONG YIZHOU
XU CHANG
Gong Zhouhao
XUE FEIFEI
XU XIANG
GU YICHENG

Assignees

盛东如东海上风力发电有限责任公司
华能国际电力江苏能源开发有限公司清洁能源分公司
河海大学

Dates

Publication Date: 20260512
Application Date: 20260127

Claims (10)

1. A wake model parameterized optimization method based on wind farm historical operation data is characterized by comprising the following steps: Collecting wind power plant historical operation data, and preprocessing the collected wind power plant historical operation data to obtain a training set and a testing set; Based on an Ishihara single-machine wake model, combining a linear wind speed superposition method based on wind wheel definition and a turbulence intensity superposition method considering wake interaction to construct an engineering experience wake model; Establishing a scaling factor for wake growth rate in engineering experience wake model Solving the optimization problem of (1) by adopting a particle swarm optimization algorithm, taking the average absolute error MAE between the power predicted value and the power measured value of the full wind power plant in the training set as an fitness function, and performing the optimization on Optimizing to obtain optimized ; The optimized product is obtained The method is applied to an engineering experience wake model and is used for power prediction of wind power plant groups.
2. The wake model parameterized optimization method based on wind farm historical operating data according to claim 1, wherein the wake model parameterized optimization method is obtained after optimization After being applied to the engineering experience wake model, the method further comprises the step of inputting data in the test set into the engineering experience wake model to obtain the predicted power of the wind power plant.
3. The wake model parameterized optimization method based on wind farm historical operating data according to claim 1, wherein the creating an engineering experience wake model based on the Ishihara single machine wake model by combining a linear wind speed superposition method based on wind wheel definition and a turbulence intensity superposition method considering wake interaction comprises the following steps: The expression of the linear wind speed superposition method based on wind wheel definition is as follows: ; In the formula, Is the wind speed of the target wind turbine generator, Is the inflow wind speed of the wind farm, The number of the upstream wind turbines; Is the number of the upstream wind turbine generator, Is an upstream wind turbine generator The superposition order is calculated from upstream to downstream according to the windward direction; The expression of the turbulence intensity superposition method of wake interaction is as follows: ; ; In the formula, Is the turbulence intensity; is the intensity of the ambient turbulence; Is wake added turbulence.
4. A wake model parameterized optimization method based on historical operational data of a wind farm according to claim 3, wherein a wake growth rate is introduced Selecting Is a proportional coefficient of (2) To optimize the object, then The calculation method of (1) is as follows: ; In the formula, The thrust coefficient of the yaw wind turbine generator is the corrected parameter; is the intensity of the ambient turbulence; Due to The calculation method of (2) causes the shape of the gaussian distribution to change, while the maximum value is unchanged, so that the velocity loss and the additional turbulence in the wake formula need to be corrected; The correction formula of the speed loss calculation method is as follows: ; In the formula, Is behind the wind turbine generator The standard deviation of the speed of the position is reduced; is the wind speed of the center of the wind wheel; Is wake centerline offset; is the hub height; Representing the diameter of the wind wheel; 、、 is a parameter; Is the radial distance; The correction formula for the additional turbulence calculation method is as follows: ; Wherein, the Is behind the wind turbine generator Standard deviation of velocity of the position; is the intensity of the ambient turbulence; is a turbulence intensity correction term; 、、、、 Is a parameter.
5. The wake model parameterized optimization method based on wind farm historical operating data according to claim 1, wherein the fitness function is expressed as: ; In the formula, Is an actual measurement value corresponding to the actual measurement value of the power of the wind power plant; is a predicted value corresponding to the wind power plant power predicted value and is output after being calculated by the engineering experience wake model; Is the number of samples and corresponds to the number of wind turbines , 。
6. The wake model parameterized optimization method based on historical operating data of a wind farm of claim 1, wherein the pair of Optimizing to obtain optimized Comprising: with the aim of minimizing the average absolute error between the predicted and measured power values Optimizing to obtain optimized The method specifically comprises the following steps: Optimizing parameters The optimization problem is as follows: ; Wherein, the Representing a fitness function; Representing constraint conditions; Is the actual measurement value of the power of the wind power plant; The wind power plant group power calculation method based on the engineering experience wake model is used for calculating the predicted power of the wind power plant.
7. The wake model parameterized optimization system based on wind farm historical operation data is characterized by being used for realizing the wake model parameterized optimization method based on wind farm historical operation data according to any one of claims 1-6, and comprising the following steps: The data acquisition module is configured to acquire wind power plant historical operation data, and preprocess the acquired wind power plant historical operation data to obtain a training set and a testing set; the engineering experience wake model construction module is configured to construct an engineering experience wake model based on an Ishihara single machine wake model by combining a linear wind speed superposition method based on wind wheel definition and a turbulence intensity superposition method considering wake interaction; an optimization solution module configured to build scaling coefficients for wake growth rates in engineering experience wake models Solving the optimization problem of (1) by adopting a particle swarm optimization algorithm, taking the average absolute error MAE between the power predicted value and the power measured value of the full wind power plant in the training set as an fitness function, and performing the optimization on Optimizing to obtain optimized ; A prediction module configured to obtain the optimized result The method is applied to an engineering experience wake model and is used for power prediction of wind power plant groups.
8. A computer readable storage medium having stored thereon a computer program, characterized in that the computer program, when executed by a processor, implements the steps of the wake model parameterized optimization method based on historical operational data of a wind farm as defined in any one of claims 1-6.
9. A computer device, comprising: A memory for storing a computer program; A processor for executing the computer program to implement the steps of the wake model parameterized optimization method based on wind farm historical operational data of any one of claims 1-6.
10. A computer program product comprising a computer program, characterized in that the computer program, when executed by a processor, implements the steps of the wake model parameterized optimization method based on historical operating data of a wind farm according to any one of claims 1-6.

Description

Wake model parameterized optimization method and system based on wind farm historical operation data Technical Field The invention belongs to the technical field of wind power generation, and relates to a wake model parameterized optimization method and a wake model parameterized optimization system based on wind farm historical operation data, which are particularly suitable for a power prediction and intelligent regulation system of an offshore wind farm group. Background Along with the development of wind power plant construction towards clustering, base and large scale, a large-scale centralized grid-connected wind power plant group gradually becomes one of important forms of Chinese wind power generation, and as the number of wind turbine generators in the wind power plant group is large, the wake flow influence range is wide, and the field-level wake flow influence range of the wind power plant can generally reach 10km. For the offshore wind farm group, the wake flow recovery is slower because the turbulence intensity is lower than that of the offshore wind farm group, the wake flow of the offshore wind farm group not only seriously affects the economic benefit, but also has more complex power characteristics because the offshore meteorological scene is changeable, the seasonal variation is obvious, the space-time difference is obvious. Therefore, researching wake flow of the offshore wind farm group is an important precondition for evaluating the generating capacity of the wind farm group and guiding the wind farm group to operate accurately. The wake research of the wind turbine generator mainly has two hot spots, namely wake development principle research and engineering wake model development and application. In wake development principle research, main characteristics of wake are verified and reconstructed through an outfield experiment, a wind tunnel simulation experiment and a CFD (computational fluid dynamics) method, and an engineering wake model can be developed based on the main characteristics. The application of the wake model in wind power plant engineering refers to providing a method for optimizing wind power plant arrangement or running period indexes or wind turbine generator wake research based on the existing wake model. For wake interference of wind power plant group layers, the current research methods mainly comprise a top-down method and a bottom-up method. The top-down method is to take a single wind power plant as an object, equivalent the single wind power plant to be additional surface roughness or momentum sinking, and change the average wind speed of the incoming flow of the downstream wind power plant by combining the influence of the thermal stability of the atmosphere. The bottom-up method is based on a single-machine wake model, the wind speed at the center of each wind turbine hub is calculated, and the wake distribution of the whole wind power plant group is obtained by applying a wind speed superposition method. Currently in wake regulation research, engineering experience wake models as a basic tool typically use fixed experience parameters given in literature or wind tunnel tests. However, the geographical location, climate environment and turbulence characteristics of different wind farms are greatly different, and the use of a set of fixed parameters can lead to insufficient calculation accuracy of the model in the specific wind farm group, thereby affecting the reliability of subsequent power prediction and regulation optimization. Disclosure of Invention The invention aims to provide a wake model parameterized optimization method and a wake model parameterized optimization system based on wind farm historical operation data, which can obviously reduce wake model power prediction errors, realize the reduction of power prediction MAE of an offshore wind farm group and the improvement of annual energy generation, and are suitable for an intelligent regulation and control system of a million kilowatt-level offshore wind power base. In order to solve the technical problems, the invention is realized by adopting the following technical scheme. In a first aspect, the invention provides a wake model parameterized optimization method based on wind farm historical operation data, which comprises the following steps: Collecting wind power plant historical operation data, and preprocessing the collected wind power plant historical operation data to obtain a training set and a testing set; Based on an Ishihara single-machine wake model, combining a linear wind speed superposition method based on wind wheel definition and a turbulence intensity superposition method considering wake interaction to construct an engineering experience wake model; Establishing a scaling factor for wake growth rate in engineering experience wake model Solving the optimization problem of (1) by adopting a particle swarm optimization algorithm, taking the average absolute error MAE between the po