如今生产生活中对电子设备的可靠性要求越来越高，因此它们的故障诊断问题也成为重中之重。而不断成熟的计算机仿真技术相较于传统的物理实验有许多优势，如模拟运行的可控性更强、模拟次数无限制等。具有较好分类能力、自学能力的人工神经网络也被发现能大大改善故障诊断的能力。本文以模拟电路为应用背景，在Matlab的simulink中搭建电路模型，利用BP神经网络作为工具对模拟电路中的三类故障进行识别和诊断，仿真结果表明BP网络诊断出这三类故障的正确率达到95%左右，具有良好的故障诊断效果。

 

关键词：故障诊断，BP神经网络，电路模拟与仿真

 

 

Nowadays, requirements of reliability for electronic devices become higher and higher in our daily life and production, so fault diagnosis of them has become a priority. Comparing with traditional physical experiments, maturing computer simulation technology has many advantages, such as simulation runs more controllable and no limitation of simulation times. And artificial neural network has greatly improved the ability of fault diagnosis, who owns better classification and self-learning ability,. This article built the circuit model in Matlab's simulink, and using BP network as a tool for identifying and diagnosing three types of fault in analog circuit, the results show that BP network has high value of research, it can diagnose three fault-categories of the circuit, the correct rate is about 95%.

 

Key words: Fault Diagnosis, BP Neural Network, Circuit Modeling and Simulation.