Research On Natural Circulation Steam Generator Water Level Characteristics And Intelligent Control

As a heat exchanger linking the conventional island and the nuclear island,natural circulation steam generator(SG)is now in common used in PWR nuclear power plants.However in a steam generator,the heat and mass transfer is very complex and the phenomenon of "false water level" can be caused easily under the disturbance conditions.So in the process of the present study,the exact two-phase model of the steam generator secondary side and the automatic control of steam generator water level are the key to be considered.This paper is researched on natural circulation steam generator of the Daya Bay plant,modeling a one-dimensional mathematic model based on the lumped parameter method,the more accurate drift-flux theoretical model has been used in the two-phase flow part of the secondary side,and the non-linear water level equations of steam generator are added in the general mathematical model of the steam generator,which to improve the whole simulation model of steam generator.At 100%,70%,50% and 30% four kinds of power levels,we do steady state simulation referenced to the actual operating parameters,which verifies the accuracy of the simulation model.Based on the computational analysis in steady state,the dynamic simulation for four kinds of power levels has been tested with disturbance of 5%increase in steam flow and 5% increase in water flow.The simulation results show that,under the disturbance conditions,the water level of steam generator exactly emulates the phenomenon of "false water level".Due to the highly complicated,nonlinearity,time-varying and non-minimum phase of the SG,the control of the water level is too hard.Most of SG water level control modulates control system based on traditional PID controller,but it always can't satisfy the requirement of automation control of nuclear power plant under the instantaneous condition.Fuzzy neural network controller based on Takagi-Sugeno model is used in this paper.And there are still some problems to determine the initial parameters,the structure of neural network and the fuzzy rules.To deal with the problems above,dynamic fuzzy neural network(DFNN)is proposed here,it can model,get fuzzy neural network structure parameters and obtain fuzzy rules automatically and the expert knowledge in this field will not need anymore.In this paper,we firstly use DFNN to confirm the structure of the neural network,and then the T-S fuzzyneural PID controller is designed to control the SG water level.The system simulation shows that the intelligent control system successfully reduced the phenomenon of "false water level",and is good for on-line control of the NCSG water level.