Circuit Failure Diagnosis Analysis and Research Based On Neural Network

For the better analysis of failure diagnosis problems, firstly, using MATLAB software to simulate failure problems such as the emergence of regular power, short circuit and broken circuit. Combining the failure practical example, neural network analysis is introduced for diagnosis. Failure diagnosis using two kinds of neural networks: the BP network and Elman network training. Trainings choose rational network design parameters, training functions and the simulation realizes practical results analysis. Then, through the comparative analysis of two network training simulation and testing results, the results conclude that Elman network training avoiding the slower pace shortcoming in BP network training. But there are still existing deficiencies. At last, the improved methods and measures provided better parameters set for the power failure diagnosis is presented.