中文摘要：

如今的电路越来越复杂，随之而来的就是电路系统的高故障性，所以如何定位故障发生成为一大难题。文中基于提高故障诊断性能的目的，先采用一种改进的局部线性分析算法作为初始数据处理器对输出响应序列进行降维，提取故障特征向量，然后再通过OIF-Elman神经网络搭建故障分类器，对电路进行故障检测。仿真结果表明，将改进过的局部线性分析算法和OIF-Elman神经网络应用到故障诊断中，不仅具有比传统BP神经网络更精确的故障诊断正确率，且整个网络的收敛速度也会有明显提升。

英文摘要：

The circuit is more and more complex and followed by a circuit system failure,so how to locate the fault occurs is a major problem. For improving the performance of fault diagnosis,use an improved locally linear embedding analysis method as the initial data （ raw data） processor for output response sequence to reduce the dimension and extract fault feature vector,and then through the OIF-El-man neural network to build fault classifier for fault detection circuit. Simulation results show that the fault diagnosis method is made of improved LLE and OIF-Elman neural network is not only to have the better diagnosis rate compared with the BP neural network,but also greatly enhance convergence speed for the whole network.