Study On Controller Performance Assessment

With the increasing complexity of control structures and the sheer number of controllers in modern process plants, controller performance assessment is becoming a key issue in industry. This thesis focuses on improvements of two most popular controller performance assessment indexes, minimum variance index and relative variance index. Controller performance assessment software is developed based on the relative variance index. In this thesis, the main research contributions are described in three aspects as follows:1. Based on the research of minimum variance index, a new method of performance assessment for system subject to multi-disturbance dynamics was proposed and the algorithm of the exact output variance lower limit of the structured closed-loop response for each dynamics was given. Also, the method takes consideration to the weights and variance limits of the closed-loop response for each dynamics, which made the result of the performance assessment more suitable for practical application.2. An improved RVI performance assessment method based on internal model control was achieved. Internal model control is suitable to be benchmark of RVI, for the characteristics of turnoff the tracking performance and robust performance, having minimum variance of process output and eliminating the effects of stochastic disturbances effectively. At the same time, the improved RVI performance assessment method need no disturbance model and can provide reasonable and robust optimal controller for references.3. GUI of MATLAB software is used to achieve an interface of the three kinds of performance assessment algorithms based on relative variance index. Once the model of process or the data of historical operation is known, the output of process, performance assessment index and the optimal controller parameters can be obtained automatically and displayed on the screen using the developed software.