A fuzzy-gain-scheduled neural controller for nuclear steam generators

Malfunction of steam generator water level control is the main cause of unexpected shutdowns in nuclear power plants. Particularly at low powers, it is a difficult task due to shrink and swell phenomena and flow measurement errors. In addition, the steam generator is a highly complex, nonlinear and time-varying system and its parameters vary with operating conditions. Therefore, it seems that design of a suitable controller is a necessary step to enhance plant availability factor. The purpose of this paper is to design, analyze and evaluate a water level controller for U-tube steam generators using neural networks. To attain a good performance in all powers, the proposed system uses a fuzzy system as gain-scheduler. Computer simulations show that the proposed controller improves transient response of steam generator water level through the entire range of operation and demonstrate its superiority to existing controllers.