Research On Application Of Stochastic Model Predictive Control In Wind Power Generation Systems

Control theory has been applied to control wind turbine systems, and has made great achievements. However, most studies have focused on tracking rated speed and power, and have not emphasized on the optimization of fatigue load problems. In this thesis, we study the optimization of a fatigue load problem on a wind farm level.As a new class of computer control algorithms, model predictive control has remarkable effects in solving complex problems of industrial process optimization control. We address a possibilities analysis for model predictive control in a wind farm fatigue optimization problem. The wind farm model in this thesis is obtained by four steps. Firstly we linearize the wind turbine model, secondly we construct a linear wind model, thirdly we incorporate the wind model into our wind turbine model, and finally we concatenate N of these turbine models to obtain the model of an entire wind farm model. For this stochastic linear model of the wind farm system, we will use stochastic LQ and stochastic MPC algorithms to deal with the fatigue load optimization problem. These algorithms are realized by simulations, and the performances are analyzed.