ANN based on an improved PSO algorithm and its application to fault diagnosis in NPP
Mao We · Journal of Naval University of Engineering · 2014
As the traditional BP ANN and standard PSO-ANN tend to cause a low convergence rate and are prone to focus on the local optimization points in the course of fault diagnosis,a new ANN training method based on the improved PSO(IM-PSO)algorithm has been proposed.This method has advantages of fast convergence rate and not being prone to remain at local optimization points,so that the method is used for the fault diagnosis of the condensation and feed water system in nuclear power plants.The diagnosis results prove that this method is in performance superior to the traditional BP ANN and standard PSO-ANN,thus it can be applied to the fault diagnosis field of a complex system.