Robust Recursive Identification of Multivariable Processes

Abstract

Many industrial processes have the multivariable nature (boiler plant, evaporators, distillation columns, etc.). One of the key requests of production is energy saving and product quality improvement (these two categories are tightly connected). In order to achieve this it is necessary to carefully design process control strategies. Therefore, the quality mathematical model of the process is needed. In this paper it is assumed that the process is described with multivariable ARX (AutoRegressive model with eXogenous input) model. The key assumption, justified with numerous studies of real processes, is that the stochastic disturbance has non-Gaussian distribution. That fact affects on form of recursive identification algorithm. The algorithm becomes nonlinear. Simulations are performed that demonstrate the superiority of the proposed algorithm over the standard algorithms.