Research On Fault Diagnosis Of A Condensation And Feedwater System Based On Genetic Algorithm For Ship's Nuclear Power Plant

Recent researches have done a great work in improving the ability of fault diagnosis(FD) method for nuclear power plants (NPP) among which the Expert system (ES) and Neural network (NN) are more practical. But the more complex a dynamic system is, the harder for it to form a full backup of samples which is the key to assure the precision of NN and ES method. And when a system like NPP's is studied, the FD results based on the above methods even can not be proved in theory. Taken the NPP's FD characteristic into consideration, a novel diagnosis method is proposed based on probabilistic causal-effect model and genetic algorithm, which takes advantage of calculating probability function from the causal-effect model as the fitness function instead of the original complex system. The combination study of the FD method and genetic algorithm (GA) is drawing world's attention. The GA can figure out a global optimization in a bionic process, providing a general frame to solve the optimization problem of the large complex system's FD, and it's robust.The research has proved in a simulation that the new method has done well with GA's parallel and global searching characteristic and it is more efficient and high precise on FD. The FD model in this paper is built up based on the original model and a lot of work has been done to train and simulate the examples, the NPP FD system is created. The simulation of the ship's NPP FD based on GA and probabilistic causal-effect model says an adaptability of the new method in dealing with uncertain factors and rise in precision to analysis single fault and multi faults as well. The FD result matches the fact and prove to be of higher reliability and practicability in project.