Research On Stochastic Predictive Control Of Wind Power Generation System

Stochastic model predictive control strategy(SMPC)can control the system with strong random disturbance effectively,which broadens the scope of mind for the control of wind power system.Therefore,this paper combines the stochastic model predictive control with the wind power system ensure output power and wind turbine stable by handling random wind speed disturbance in wind power system.Based on the random characteristics and model of wind,the principle of wind energy utilization and the mechanism of the components,the nonlinear model of wind power system is finally established.According to the model and the actual control process,a control target and the constraints to be met by input and output are given.A predictive control problem based on nonlinear model of wind power system is proposed.The appropriate operating point is selected by the characteristics of stochastic wind speed.The model is linearized at the working point to deal with the strong nonlinearity of the original model and to obtain predictable states.The stochastic model of wind power system based on additive disturbance and the stochastic model of wind power system based on mixed disturbance are established respectively according to the randomness of disturbance and the characteristics of input and output constraints.The constraints in stochastic model with additive disturbance are hard.The output in the stochastic model with the mixed disturbance is set as probability constraint.Then the stochastic model predictive control strategies are established respectively for both models.A stochastic model predictive control strategy based on Kalman filter and output feedback is designed.Simplify the output by converting model.Kalman filtering is used to describe the stochastic states in the predicted time domain and to transform the objective functions with random statistical information.The input is designed as output feedback,which leads to more free variables and overcomes the conservativeness of traditional control strategies in handling disturbance.The algorithm regulated the pitch angle and the generator torque so that the output power tracks the desired value well.The simulation of 5MWwind power system shows the effectiveness of the algorithm.A stochastic model predictive control strategy based on multi-step feedback is proposed for the random mixed disturbance model of wind power system with probability constraints.An augmented state space model is proposed of using the observer.The form of the problem and objective function is simplified by using parametric feedback strategy.The model is transformed into the autonomy state space model of expanding dimension by using the information of forecast input sequence.Utilizing the randomness of the disturbance,the statistical performance index in infinite time domain is transformed into the convex function which is only related to the initial state.The multi-step feedback law is used to transform the output probability constraint into a probability constraint on the state,which overcomes the conservativeness of the traditional control strategy for the disturbance.According to the concept of invariance,probability elliptic sets and robust elliptic sets of the original state and the expanded state are given,and the probabilistic constraint is dealt with by the Markov chain model formed by the adjacent moments.The dual-mode control is used to optimize the objective function and to ensure system stability.The simulation of 5MW wind power system of SMPC algorithm and traditional MPC is compared to verify the effectiveness,superiority and adaptability of the SMPC.