Automated Monitoring and Assessment of Controller Performance

The automation of methods for monitoring and assessment of controller performance can permit their wide spread and frequent use. Minimum variance control provides a useful benchmark for assessing the controller performance and cross correlation analysis can be used to identify the cause(s) of poor performance. Discrimination between plant model mismatch and poor controller tuning, isolation of poor feedback control or poor feedforward control is possible by analyzing normal operating data and data based on small set-point perturbations. The assessment procedure is automated by utilizing a portmanteau test. The data to be used in automated performance analysis has to be screened for abnormalities and information content. Data screening, performance monitoring and diagnosis are implemented by software developed in Matlab.