Application of a New Scheme for Adaptive Unfalsified Control to a CSTR with Noisy Measurements

Tanet Wonghong and Sebastian Engell
Process Dynamics and Operations Group


Abstract

In this paper, a new scheme for adaptive unfalsified control with non-ideal measurements is presented and demonstrated for a well-known example of a nonlinear plant, the continuous stirred tank reactor (CSTR) with the van-der-Vusse reaction scheme. In our adaptive control algorithm, there are two adaptation mechanisms: 1. Switching of the active controller in a fixed set of candidate controllers by the epsilon-hysteresis switching algorithm. 2. Adaptation of the set of controllers performed by a population-based evolutionary algorithm. In this paper, the effect of measurement errors on the adaptive control scheme is investigated. The total least squares method is used to perform the deconvolution of noisy signals.