Research On Steam Generator Water Level Characteristic And Intelligent Control

One of the major components in a PWR plant is U-tube steam generator (SG) that provides necessary steam for generating electricity power. Properly controlling the water level of SG is very important in order to secure the sufficient cooling source of nuclear reactor and prevent the damage of turbine blades. There are factors such as nonlinearity, high complexity, time varying, swell-and-shrink phenomena and non-minimum phase dynamics which make water level control problem of SG very challenging. Whats more, SG parameters undergo large changes according to different operational conditions. The SG cannot work satisfactorily with conventional fixed PID gains over full operating range.In this paper, based on adequately analyzing SG water level dynamic characteristics, several intelligent techniques such as Genetic Algorithms, Fuzzy Logic control theory, and Fuzzy Neural Networks are applied to SG water level control.The whole research work of this paper includes three parts. First, a SG water level PID control system of which parameters are optimized by GA is designed based on the powerful capability of GA for searching the global optimal resolution. Second, a fuzzy control system is successfully designed and it can control the water level effectively because the fuzzy control technology doesn't require the precise model of control object. Further more, the parameters of the fuzzy control system are optimized by GA, so the water level can be controlled better. Third, three kinds of self-turning FNN water level controller are designed through integrating the fuzzy technique and neural network. The parameters of the first two FNN controllers are learned through BP algorithm and GA, respectively. The third FNN is called CFNN, and it is composed of many compensatory neural cells which have fast learning speed.All the intelligent SG water level control systems mentioned above are programmed and simulated with VC++ and MATLAB. Simulation result shows that the performances are better than conventional PID control system. Therefore, the proposed methods play an important role in guiding the research work of SG water level intelligent control to a certain extent.