Study on the state prediction of electronic device based on the BP neural network of genetic algorithm
Ting An · 2017
The state prediction of electronic device is the foundation and the key of the fault prediction and health management (PHM), it used to use the neural network method. In order to improve the accuracy, the paper studied the problem of using genetic algorithm to improve the BP neural network. First, the genetic algorithm was used to optimize the thresholds and the weights of the BP neural network. Then, the network was trained to obtain optimal solution. In the end, the paper verified the effectiveness of the BP neural network of genetic algorithm by two examples. The results of the first example showed that the average value of relative error of the output voltage predicted by the BP neural network of genetic algorithm is 2.11%, but the BP neural network method is only 5.342%. This shows the genetic algorithm method is superior to the traditional method, and has higher prediction accuracy. The second example drew the similar conclusion.