SVM-Based Fuzzy Control For Wind Energy Conversion System

While the MW level variable-speed constant-frequency wind power generation systems widely used, more efficient control systems are needed. Variable speed wind power generation system is a typical multivariable nonlinear system. The traditional PID control system needs an effective system model, however, as the uncertainty of aerodynamics and the complexity of power electronic devices, it is very difficult to obtain accurate models. Coupled with the increased sediment on the blades, the weather condition changes, the unit mechanical changes caused by aging, even if the effective system models are established, the control systems are only applicable to the specific working conditions and operating periods. So the more reasonable and effective control system must be designed. TSK fuzzy model is used widely. It can approximate the nonlinear function by arbitrary precision. The nonlinear systems will be analysised as a number of linear subsystems, thus the control theorys of the linear systems also can be used in the nonlinear systems, too. SVM is based on the principle of structural risk minimization, with good generalization ability. Therefore, a new fuzzy controllor for maximum energy extraction from variable speed wind turbines is presented in this paper, and the Takagi–Sugeno–Kang (TSK) fuzzy model based on Support Vector Machine (SVM) is used in this controller. To identify this model, fuzzy clustering methods for partitioning the input–output space, and Least Squares Support Vector Machine (LS-SVM) methods for the consequent parameters of fuzzy rule adaptation are used. The simulation result on variable-speed wind generation system confirms the aim of maximum energy extraction completed perfectly. In addition, this controller exhibits strong real-time, fault tolerance, learning capability and high speed of computation.