基于支持向量机的化工装置电力电子故障诊断

Based on support vector machine for power electronic fault diagnose of chemical plants

采用小波包分析与支持向量机(SVM)对化工装置电力电子故障进行自动识别和诊断,运用变尺度分辨小波包方法对电力电子故障信号进行特征处理。支持向量机能够对小样本数进行模式识别,并且具有良好的分类推广能力。在小波包分析特征基础上,采用分布式多支持向量机(SVM)分类器识别化工装置电力电子故障。结果表明:该方法能准确有效地对化工装置的电力电子故障进行识别和诊断。

The combination of wavelet packet analysis and support vector machine is introduced to solve automatic detection of power e- lectronic fault diagnose. Wavelet packet analysis which holds multi-resolution and multi-scale is introduced to deal with the signal characteris- tics. Support vector machine can carry through the pattern recognition on the small-samples and has well generalized ability. Based on wavelet packet analysis for signal characteristics, the distributed muhi-SVM classifier is utilized to identify the power electronic fault diagnose, the experimental results also show that this method can efficiently identify and diagnose the power electronic fault diagnose of chemical units.