Non-linear generalised predictive control of a jacketed well mixed tank as applied to a batch process—A polymerisation reaction

Abstract

The purpose of this study is to control the temperature of a well mixed jacketed tank under set-point changes with the use of non-linear generalised predictive control (NLGPC) system. Firstly, perturbations were given to the cooling flow rates so as to investigate the dynamic behaviour of the process. The emphasis was put on the use of polynomial non-linear autoregressive moving average with external input (NARIMAX) model related with tank temperature and heat input for non-linear control. The second part of the study deals with the presentation of identification algorithm for the construction of polynomial NARIMAX and autoregressive moving average with external input (ARIMAX) models. A pseudo random binary sequence (PRBS) signal was utilised as a forcing function in order to determine the parameters of the models. Levenberg–Marquardt algorithm was applied in the estimation of relevant parameters of NARIMAX model. Similar studies were performed for the ARIMAX model using Bierman algorithm. Next linear and non-linear models with generalised predictive control (GPC) algorithm were applied to test the performance of the available control in response to disturbances. Finally the results from NLGPC were compared with linear generalised predictive control (LGPC) to see the efficiency of the batch control system.