Stochastic Performance Assessment of the Controller with Complex Deterministic Trend Output and its Application

摘要

There is a certain type controller that has unknown and complex deterministic trend output of changing with time slowly in actual industrial processes. And the traditional minimum variance criterion can't be used to evaluate the stochastic performance of the controller directly. Hence this paper presents a method to address the difficulties in such case, which combines wavelet technology and minimum variance criterion for stochastic performance assessment. Firstly, wavelet technology (WA method) is adopted to separate stochastic portion from the output time series. And then the stochastic portion is modeled and analyzed. Therefore, stochastic p erformance of the controller is assessed based on traditional minimum variance criterion. The effectiveness of the proposed method is verified through simulations. This method can avoid over-differencing when the stochastic portion is extracted by using difference method (DI method), and realizes accurate and reliable assessment for stochastic performance of the controller.