基于“透明”神经网络的模拟电路故障诊断

ANALOG CIRCUIT FAULT DIAGNOSIS BASED ON "TRANSPARENT" NEURAL NETWORK

本文通过分析BP神经网络用于模拟电路故障诊断时的缺点,提出了一种新的神经网络,称为"透明"神经网络(TANN).然后将TANN用于模拟电路子网络级故障诊断,并通过实例验证了该诊断方法的有效性.

In this paper,the defects of BP neural network in analog circuit fault diagnosis are discussed and a 'transparent' neural network (TANN) is introduced.The TANN is then used to diagnose analog circuit faults at sub-network level.Validity of this method is testified by experiments.