Performance assessment of model predictive control in industrial distillation column

A synthetical method of multivariable control system performance assessment is proposed in this paper, which uses multivariable minimum variance control (MVC) benchmark to determine the stochastic performance, and normalized multivariate impulse response (NMIR) curve as an alternative measure of performance to test the dynamic performance, and with the help of auto-correlation function (ACF) and cross-correlation function (CCF) to analyse if there are oscillations exist. The method is applied to assess the performance of multivariable predictive control system of industrial distillation column.