中文摘要：

为克服传统分形理论不包括信号细节成分的缺陷，提出了一种基于小波分形算法的旋转机械故障特征提取算法．该算法将小波函数和分形维数2种故障诊断方法结合起来，既考虑信号细节成分，也注重其局部与整体的关系．利用转子试验台系统模拟了3种故障工况下的旋转机械振动信号，并分别用传统分形维数算法和所提出的小波分形算法对其进行了特征提取．结果表明，2种算法提取得到的特征均有良好效果，但小波分形算法具有较高的准确性，为准确提取旋转机械振动信号故障特征提供了一种快速有效的新方法．

英文摘要：

To overcome the shortcoming of traditional fractal theory＇s lacking of signal detail components, a fault diagnosis method based on wavelet fractal algorithm is brought up in this article. The algorithm combines wavelet function and fractal theory together, so it considers both the signal detail components and the relationship between local and whole. An experimental machinery system is adopted to simulate ro- tating machinery vibration signals under three different conditions~ and they are feature extracted by both traditional fractal algorithm and the proposed wavelet fractal algorithm. The results show that both algo- rithms are effective in feature extraction, while the wavelet fractal algorithm provides a more accurate fea- ture extraction of rotating machinery signals.