高速列车转向架是否故障及故障种类对列车运行的安全性有重要的影响。为了及时高效的对转向架关键部件进行故障诊断，本文选用高速列车转向架典型故障振动信号，提出了运用功率谱与主成分分析相结合的方法提取信号特征，先对样本数据进行功率谱估计，构造包含所有工况的特征频点数组，将这些频率点在每个样本的功率谱中对应的幅值作为特征向量，再通过主成分分析降维处理，去除冗余的特征项，最后经过支持向量机分类判断出故障种类。用加速度和位移传感器选取5个测点，取得了满意的识别结果，准确率在90%以上，验证了该方法的有效性。%The bogie faults and fault types of high-speed trains have an important impact on the safety of train opera-tion. In order to diagnose the bogie fault efficiently, a novel method for feature extraction was proposed by combination of power spectrum with principal component analysis. First of all, the power spectrum evaluation was done using the sample da-ta of the typical fault signals of the bogie, and an array including feature frequency points of a variety of faults was construct-ed. Then, the power spectrum amplitudes of these points were used to constitute a feature vector, and the redundant features were removed by principal component analysis. Finally, the fault type was identified by SVM. As an example, 5 different test points were selected for the testing and satisfactory recognition results were obtained. The correct recognition rate was up to 90%, which verified the effectiveness of this method.
展开▼
