基于遗传BP网络的模拟电路故障诊断

Fault Diagnosis Based on Genetic Algorithm-optimized BP Networks

在模拟电路故障诊断中,BP(back propagation)神经网络得到了广泛的应用并取得了不错的效果。但是BP神经网络在训练时仍然存在网络学习收敛速度慢、不易获得全局最优解、网络结构不确定等缺点。采用Levenberg-Marquardt算法进行网络训练,并用遗传算法对BP神经网络结构、初始连接权值和阈值进行全局优选,可以有效克服BP网络存在的缺陷。以Leap Frog Filter滤波器电路的故障诊断为例,仿真实验表明,优化后的BP网络能够快速有效的诊断电路中存在的故障,并且具有更高的诊断精度。

Back Propagation（ BP） networks have been widely used in analog circuit diagnosis and have achieved some success. However,there are some inherent disadvantages in traditional BP neural network,such as the low speed of error convergency,easily falling into local minimum and the uncertainty structure of networks. Therefore,a new BP network method optimized by genetic algorithms（ GA） and Levenberg-Marquardt（ LM） algorithm is proposed. In this method,the structure of BP network is optimized by GA. Then LM algorithm is used to train the BP network. The training results can be used to diagnose the faults of analog circuits,which is able to overcome the inherent disadvantages of traditional BP network. The fault diagnosis of the Leap Frog Filter circuit is taken as the example and the simulation results demonstrate the effectiveness and applicability of the proposed method。