SYNTHESIS OF A NOVEL ROBUST INTELLIGENT CASCADED REACTIVITY CONTROLLER FOR CANDU REACTOR

Abstract

In this paper, a Novel Intelligent Cascaded Reactivity Controller (NICRC) is synthesized for a CANadian Deuterium Uranium (CANDU) -type Nuclear Power Plant (NPP) operating in Pakistan. The designed NICRC is a cascaded configuration of reactor power and moderator level controllers composed of five sub-controllers. The proposed NICRC is designed using intelligent soft computing technique. The original reactivity controller of CANDU nuclear power plant is a networked controller implemented on a Programmable Logic Controller (PLC). The new NICRC is designed based on two Intelligent Distributed Cascaded Power Controller (IDCPC) and Intelligent Cascaded Moderator Level Controller (ICMLC) for moderator level control in CANDU reactor core. The IDCPC is composed of three neural sub-controllers while ICMLC is composed of two neural sub-controllers. The proposed NICRC is designed using Adaptive Back Propagation Feedforward Neural Network (ABPFNN). The proposed controller is synthesized in a distributed parallel computing environment using MATLAB. The proposed NICRC is formulated in a highly complex multi-objective neural form and evaluated against full power operation of CANDU nuclear power plant from cold start-up to high power. The performance of highly robust NICRC is tested and evaluated for a typical transient providing complete coverage of moderator level, low log and steam pressure modes of CANDU reactor and found excellent within the desired control bands.