Controller Design For A Class Of Non-Gaussian Systems With Stochastic Delay

The control of stochastic systems has always been one of the hot fronts in the research of control.Most of the current researches are based on Gaussian linear systems.However,actual industrial systems are often non-Gaussian.For non-Gaussian systems,traditional control strategies based on minimization of variance are no longer applicable.At the same time,due to limited conditions,most actual control systems contain random delays.It is generally regarded as random interference and studied in the time domain.Generally these studies leave redundancy.This inevitably results in conservation.Therefore,this paper mainly studies the control problems of non-Gaussian nonlinear systems with stochastic delays,and makes corresponding work in the following aspects in view of the shortcomings of current research.1.Introduce a method to describe random delays in the frequency domain.The output of the model obtained by this method is equivalent to the output expectation of the stochastic time-delay system.Compared with the traditional processing method,it reduces the conservation and complexity,and facilitates the subsequent analysis and research.2.The non-Gaussian stochastic systems are studied,the control methods based on minimum entropy are analyzed,and their deficiencies are pointed out.Based on the improvement of the existing methods,a new adaptive controller is proposed in combination with the gradient descent method to deal with the non-Gaussian nonlinear stochastic systems that cannot be applied by the traditional minimum variance method.3.The proposed adaptive controller is applied to the control of the semiconductor batch production and the maximum wind energy utilization in wind power generation.The adaptability of the adaptive controller is tested and the application field of this control law is expanded.