Recent Developments in Controller Performance Monitoring and Assessment Techniques

摘要

In the past several years there has been considerable commercial and academic interest in methods for analyzing the performance of univariate and multivariate control systems. This focus is motivated by the importance that control systems have in enabling companies to achieve goals related to quality, safety and asset utilization. Control system performance cannot be adequately described by simple statistics, such as the mean and variance of manipulated and controlled variables, the percentage of time that constraints are satisfied, and the on-stream time. Although these are important performance measures, a comprehensive approach for controller performance monitoring usually includes the following elements: i) determination of the capability of the control system, ii) development of suitable statistics for monitoring the performance of the existing system, and iii) development of methods for diagnosing the underlying causes for changes in the performance of the control system. In this paper, recent developments related to these items will be reviewed for both univariate and multivariate systems. Some multivariate time series methods helpful in supporting these controller performance assessment techniques in practice will be discussed, and an industrial example will be provided. Finally, the future direction of commercial applications of controller performance assessment will be briefly discussed, as will the issue of whether controller performance assessment is destined to be offered as a product or a service.