DESIGN OF A MODEL BASED MULTIVARIABLE ROBUST INTELLIGENT POWER CONTROLLER FOR LOAD FOLLOWING OPERATION OF NUCLEAR POWER PLANT

In this paper, a new Multi-Input Single-Output Robust Intelligent Power Controller (MISO-RIPC) is designed for a Pressurized Heavy Water Reactor (PHWR)-type Nuclear Power Plant (NPP) operating in High Power Steam Pressure Mode (HPSPM) in Pakistan. The proposed MISO-RIPC is a highly nonlinear intelligent controller synthesized based on Adaptive Feed forward Artificial Neural Network (AFANN) and has a 3-20-1 topology with high degree of robustness. An optimization procedure is performed for the selection of optimum number of neurons for highly nonlinear AFANN. A proposed multi-layer neuro controller is evolved as an optimization problem that resolves the nonlinear issues of complex control structure of PHWR-type nuclear power plant based on Real Time Dynamic Transient Data (RTDTD) of steam flows through north and south headers and steam pressure in the steam generator. The training of an intelligent controller is performed by soft computing and parallel tracking of Conventional Reactor Power Controller operating in Steam Pressure Mode (CRPC-SP). A data driven model of PHWR-type nuclear power plant is developed in high power mode using Controlled Auto-Regressive and Integrated Moving Average (CARIMA) technique. The proposed MISORIPC is validated against RTDTD. All design and simulation work is carried out in MATLAB. The dynamic behavior of the proposed MISO-RIPC is evaluated using a very special RTDTD. The performance of proposed intelligent controller is found highly smooth with excellent tractability features. The proposed intelligent controller is found robust based on reverse engineering approach.