A LOCAL MODEL NETWORKS BASED MULTIVARIABLE LONG-RANGE PREDICTIVE CONTROL STRATEGY FOR THERMAL POWER-PLANTS

Load-cycling operation of thermal power plants leads to changes in ope rating point right across the whole operating range. This results in n on-linear variations in most of the plant variables. This paper invest igates methods to account for non-linearities without resorting to on- line parameter estimation as done in self-tuning control. A constraine d multivariable long range predictive controller (LRPC), based on gene ralised predictive control (GPC) algorithm, is designed and implemente d in a simulation of 200 MW oil-fired drum-boiler thermal power plant. In order to take into account system non-linearity, the LRPC is evalu ated using two types of predictive models: approximate single global l inear models and local model networks (LMN). As a simpler alternative, single global linear ARIX models were identified off-line with data g enerated by running the plant simulation over a load profile covering the entire operating range along with suitable PRBS signals superimpos ed on controls. For more accurate long-range prediction, networks of d ynamic local linear models, identified after dividing the whole operat ing region into a number of zones, were created. The control strategy gives impressive results, when used in controlling main steam temperat ure and pressure and reheat steam temperature during large rate of loa d changes light across the operating range. The improvements are appar ent in both constant-steam-pressure as well as variable-steam-pressure modes of plant operation. The results obtained with LMNs based LRPC c ompare favourably to the those obtained with global model based LRPC. (C) 1998 Elsevier Science Ltd. All rights reserved.