Research On Optimal Control Method Of Wind Farm Wake Flow Based On Weighted Graph Portioning

The wake effect can lead to generation loss in a wind farm,which can be mitigated through yaw optimization control.However,most previous studies have used a centralized approach to establish optimal control strategies,which becomes computationally complex as the number of wind turbines increases.In addition,these strategies often do not take into account factors such as wind speed and direction over time,wake expansion properties,and wind turbine control processes,which c leads to unsatisfactory results in actual operations.To address these issues,a new approach is proposed in this paper.First,a wind farm model based on an analytical wake model is developed and validated.Then,a grouping method based on weight graph is proposed to decompose complex highdimensional problems into multiple low-dimensional problems for parallel solution,which effectively improves the calculation efficiency.Finally,a pseudo-dynamic wake model is used to account for the wind direction changes in the simulation.We discuss the effectiveness of the yaw optimization control strategy under dynamic inflow.The main research work is as follows:(1)A wind farm model based on analytical wake model was established.The power loss due to wake effects was analyzed by simulating Horns Rev offshore wind farm with different wind turbine spacing.Based on the measured data,the prediction accuracy of the Jensen wake model and the Gauss wake model are compared.It is shown that the Gauss wake model has a higher prediction accuracy than the Jensen model.(2)The Gauss wake model is employed to simulate the wind farm wake,and the wake weight graph is established based on the wake coupling relationship among wind turbines.The optimization problem is simplified by decoupling the wind farm into several uncoupled subsets of wind turbines.Using a gradient-based distributed optimization method,the total power output of the subsets is maximized by optimizing the wind turbine yaw angle in each subset.The proposed method yields promising results,achieving a power output improvement of less than 1%and demonstrating high computational efficiency,which is approximately 38 times higher than that of the centralized method.(3)Considering wind direction changes over time and wake dynamic characteristics,a yaw optimization control algorithm considering real environment and dynamic operation of wind turbines is established based on the quasi-dynamic wake model,aiming at increasing power generation and reducing yaw times.The effectiveness of the proposed method is analyzed as an example using three aligned wind turbines with measured wind direction.As a result,the energy increment under dynamic inflow is increased by 3.41%,which is 21.43%lower than the estimated yield based on the Gauss wake model.