Fault diagnosis for power units of cascaded inverters based on combined neural network

In order to improve the accuracy and the stability of the fault diagnosis, a new combined neural network is proposed in this paper. Initial weights and thresholds of the traditional combined network have been optimized by using the genetic algorithm. The network learning method and the convergence are analyzed by using the BP neural network with the negative gradient searching. The combined neural network diagnosis method based on the genetic algorithm optimization is built. The diagnosis method has been applied to the power device fault of the cascaded inverter. The results show that this method used in the power device fault is feasible, and the accuracy of the fault diagnosis can be effectively improved by using this method.