Design of fuzzy control system for chemical injection system retrofit using neural network model in thermal power plant

Many conventional thermal power plants were started up and shut down infrequently and the operating loads were constant at the normal operation. In contrast, in recent years, with the participation of nuclear power generation and from the varied pattern of load consumption, it has been required that the plants should be operated efficiently. Thus, the DSS (daily startup shutdown) and WSS (weekly startup shutdown) have been carried out, and the plant startup, shutdown and load variations have become very frequent. This tendency is reflected also on water conditioning systems. The Poryoung thermal power plant designed with the concept of DSS operation, and also, the CIS (chemical injection system) as one of water conditioning systems have automatic operation mode to maintain constant water quality with conventional control system. Unfortunately, the control system does not maintain the set point of water quality. The engineer revise the control concept with F(x) function, but the results of test operation do not show satisfactory performance. This paper discussed the parameter identification of CIS using neural network model and suggested a fuzzy control system for CIS to improve the control performance and showed the validation of suggested system using simulation