Control Performance Assessment Of A Class Of Stochastic Distribution Control System

There exist a large number of random behaviors in the industrial system.The random behavior may not obey Gaussian distribution;and even under the Gaussian noise,the nonlinear characteristics of system will also make the system output obey to the non-Gaussian distribution.It is particularly obvious in the control and filtering problems.For this case only using the standard deviation as the reference tool and standard cannot precisely measure the performance of the control loop or filter.To solve these performance assessment problems of the nonlinear non-gaussian system,people try to use the minimum entropy to assess the performance of the control loop.This dissertation mainly aims at performance evaluation benchmark of control loop of the non-gaussian system and nonlinear system of performance evaluation for the following research.1.According to the entropy and variance in the gaussian signal has a one-to-one correspondence principle;this paper corrects the performance evaluation benchmark calculation method of non-gaussian signal system.On this basis,for the non-uniqueness of the minimum entropy,this paper introduces the evaluation benchmark of control circuit and its calculation method based on the minimum entropy of the general stochastic signal system.For correction it is verified by the simulation that calculation method of the minimum entropy shows has better adaptability2.A Sliding Mode Variable Structure Control mechanism based on PDF shaping is realized.On this basis,the corresponding control loop performance is acquired by linear approximation method to assess benchmark.It is verified by the simulation that the effectiveness of the control loop performance evaluation benchmark is positive,and the sliding mode variable structure control based on PDF shaping of show better control performance.3.Further discussed is given on filter design and the corresponding performance evaluation methods.An Unscented Kalman Filter(UKF)and an Extended Kalman Filter(EKF)both based on PDF shaping are given.On this basis,the linear approximation method is used to collect the corresponding performance benchmark.The effectiveness of the filter performance evaluation benchmark is verified by simulation and the filter based on PDF shaping show better filter performance.