Fault diagnosis of analogue circuits based on improved genetic algorithm and neutral-network

Fault diagnosis of analogue circuits based on improved genetic algorithm and neutral-network

Wang Mingfang1, Wang Jie1, Zhao Xuejun2, Yuan Xiujiu2

COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(12C) 89-95

1The College of Missile Air Force Engineering University, Xian, Shaanxi 710051, PR China

2The College of Science Air Force Engineering University, Xian, Shaanxi 710051, PR China

This paper proposes a novel method for optimization of BP neutral network using improved genetic algorithm to diagnose the circuit fault in power-supply system. First, BP neutral-network's structure is determined so that its threshold values and weight values can be optimized by GA. Second, stable threshold values and weight values are obtained via the calculation of GA's operators. Finally, the values are utilized in BP neutral network as initial parameters to conduct sample iteration training. The results show that, during fault diagnosis, BP neural network and genetic algorithm combined with each other to achieve complementary advantages between the two methods.