Research On Coordinated Control System Predictive Control Algorithm Based On Nonparametric Model

The coordinated control system of unit plant plays an important role in responding to the change of power grid load requirements quickly and ensuring generation units safety and economical operation.Traditional control methods based on linear parametric models often can not achieve good control results due to the characteristics of coordinated control system such as multi-variable,strong coupling,nonlinearity,large time delay,and parameters time-varying.Therefore,the application of nonlinear system modeling methods and nonlinear control strategies in coordinated control systems has always been one of the most important research topics.This paper mainly carries out the research work from the above two aspects:(1)Support vector machine(SVM)is a kind of statistical learning theory based on small sample data.By introducing the kernel function and the principle of structural risk minimization,The nonlinear mathematical model obtained by this method can achieve higher fitting accuracy and has good generalization ability.As an improved algorithm,the least square support vector machine(LS-SVM)has some problems in parameter selection and noise sensitivity while simplifying the calculation and improving the convergence speed.Therefore,an online adaptive weighted LS-SVM(OAW-LS-SVM)algorithm is proposed in this paper.In this algorithm,the com-bination of particle swarm optimization and grid search is used to optimize the super parameters,and the robustness of the LS-SVM to noise is improved by the improved weighting function.In addition,this method introduces the concept of rolling time window for real-time model tracking.The simulation results show the effectiveness of the algorithm.Finally,a non-parametric system model of coordinated control system was established successfully.(2)The inverse system method is one of the strategies to solve the nonlinear system control problem.The method realizes the linearization of the nonlinear syst em by solving the inverse system model of the original system.By extending the inverse system method to the MIMO system,the decoupling control of the multivariable non-linear system can be realized.Aiming at the difficulty of obtaining the exact structure and parameters of the actual system model,OAW-LS-SVM algorithm is used to ide-ntify the inverse system model directly.The ?-order purely delayed pseudo-linear system conforming to CARIMA model is obtained by connecting the inverse system to the original system in series.Considering that generalized predictive control has good robustness against noise disturbance and model mismatch,A closed-loop gen-eralized predictive control system based on the inverse system method is designed in this paper.The simulation results show that this method has a good set value tracking and anti-interference ability,which can improve the control quality of the coordinated control system.