Research On Fault Estimation And Control Technology Of Wind Turbine System Based On Doubly-Fed Induction Generator

As a green and renewable resource,wind energy has great potential for improving the energy structure and alleviating the energy crisis.Wind power generation is an effective method to utilize wind energy and has been widely used to generate electricity.With the capacity of wind turbine system increasing and the operation environment depredating,the demands on safety,reliability and high performance become more and more important.The hot concerned research issues are:system power control and fault diagnosis and fault tolerance control.In low wind speed area,the main control objective is to track the optimal power,while in high wind speed area,the objective is to maintain the rated power by adjusting the pitch angle so that it can ensure the operating safety.Fault diagnosis and fault-tolerant control technology can reduce unnecessary downtime,tolerate the component failure and maintain acceptable performance when a fault occurs.Focusing on these two issues,the main study of this thesis includes the following aspects:Firstly,in order to capture the optimal power in low wind speed area,intelligent proportional integral(iPI)controller based on mode free control and iPI sliding mod-e controller have been researched.In the methods,the uncertainty,disturbance and higher-order unknown parts of the system are treated as a lumped disturbance which it can be approached by the extend state observer estimation algorithm.By adding an ex-tra sliding mode module,the estimation error of the lumped disturbance can be reduced and compensated so that it.achieves efficient performance in tracking the optimal power.Considering the pitch angle control in high wind speed area,a intelligent-proportional derivative(iPD)neural network controller have been proposed.In this method,the esti-mation error of the lumped disturbance are reduced by adding a neural network module.These two types of controllers do not depend on a large number of the knowledge of the plant and the structure is simple.The simulation results demonstrate the performance of the proposed methods.Secondly,considering the torque and pitch angle actuator faults in the wind turbine system,a fault estimation based on the unknown input observer and fault tolerant control strategies have been studied.In this method,the wind turbine system are modeling as a linear parameter-varying(LPV)system.As for the LPV system,using the unknown input observer,an actuator fault estimation algorithm is designed.The proposed fault estimation algorithm can be suitable to both the fixed fault and the time-varying fault with first-order derivative bounded.Using the fault estimation,we developed two fault tolerant control strategies:a fault tolerant control strategy based on unknown input observer and a fault tolerant control based on integrator are proposed for wind turbine systems.Thirdly,considering both actuator faults and disturbance in the wind turbine sys-tem,a fault and disturbance estimation algorithm based on Takagi-Sugeno(T-S)fuzzy-logic theory is presented.In this method,the wind turbine system based on doubly fed induction generator(DFIG)is modeled as a T-S fuzzy system.Using a special linear transformation,the original system is decoupled into three independent subsystems:s-tate subsystem(without fault and disturbance),fault subsystem(without disturbance)and disturbance subsystem(without fault).For the state subsystem,a T-S fuzzy observer is designed.The estimation of the fault and disturbance can be obtained by using other subsystem models.The feasibility of the proposed estimation method can be judged by checking the matrix rank.Using the estimation result,an fault tolerant control with disturbance compensation is developed.The simulation results verify the effective per-formance.Finally,considering the limitation of the system measurement variables in the third method(p>l+q).a fault and disturbance estimation algorithms based on sliding mode observer is presented.In this approach,by coordinate transformation,the original system with disturbance and fault are decoupled into two subsystems.For these subsystems,a sliding mode observer is developed and the the actuator fault and disturbance are computed by using equivalent output injection concept.In sensor fault with disturbance case,the sensor faults are treated as new states of an augmented system.These proposed method is not only applicable to the case that the measured variable is equal to the sum of disturbance and fault variables(p≥ l+q),but also applicable to a nonlinear system that can be fuzzified as a liking-linear system.The simulation results proof the performance.