摘要： 提出一种基于PCA和神经网络的故障诊断/识别方法，利用主元分析法提取故障样本集的主元，实现故障样本的最优压缩，简化故障诊断中神经网络分类器的结构，提高神经网络的分类速度和测试精度。仿真结果表明，该方法可以有效减少输入层神经元个数，提高神经网络模型的学习效率和诊断的准确性，具有良好的故障识别能力。

关键词: 主元分析, 神经网络, 故障诊断

Abstract: A new method of fault diagnosis distinguishing based on PCA-neural network is raised. It uses PCA theory to extract the main element from the fault sample data, realizes optimum compression of fault sample data, simplifies structure of neural network classifier in fault diagnosis, and enhances classification speed and precision. The results of power electronic circuit experiment show that the method can decrease the number of the network input nerve cells effectively, and enhance study efficiency and diagnosis accuracy. The way has very good fault distinguishing ability and vast prospect.

Key words: Primary Cell Analysis(PCA), neural network, fault diagnosis

中图分类号: 

• TP18