Controller performance monitoring in the presence of uncertainty.

摘要

In order to remain competitive in today's marketplace, efficient manufacturing or processing of high quality products is a prerequisite. Achieving this goal can be materialized with proper design and maintenance of the process. Automated control applications play an important role to achieve process objectives such as safety, quality and optimization. Accomplishing reliable and profitable automated control applications in the process industries requires well designed, tuned and maintained control systems.; Both control theoreticians' and practitioners' main pursuit has been the design and implementation of control algorithms. Although there is a large diversity of sophisticated control algorithms, very few techniques exist for objective measures of controller performance from routine operating process data.; This situation creates a need for tools and methods for the process industry that monitor the performance and diagnose control applications, in order to successfully utilize and maintain control strategies. Controller Performance Monitoring (CPM) has been a newly formed and promising area in the past decade that provides means of diagnosing control loop performance.; First, studied in this research work, is the application of multivariate controller performance monitoring in an industrial snack food frying process. The predicted performance of a minimum variance controller is used as a benchmark standard for the evaluation of controller performance. In order to use this technique, an estimate is needed of the interactor matrix characterizing the process time delay structure of a multivariate process and a closed-loop disturbance model from a set of representative data of the controlled output variables under the current feedback scheme. We report practical experiences and various implementation issues with the use of this technique.; After this objective is accomplished, a more theoretical part follows. First, the validity of the minimum variance benchmark with respect to different kinds of measurement or sensor noise is discussed. Second, several relationships between uncertainty in the process time delay or interactor matrix and its affect on the performance indices are defined and established.