中文摘要：

介绍了EMD和排列熵异常检测的原理，通过EMD分解提取轴承异常信号，采用互信息和伪近邻法确定排列熵算法的嵌入维数和延迟时间，利用排列熵算法模型对轴承全寿命数据进行了分析，有效地检测出轴承运行的正常、异常与故障状态的发生时刻。该方法计算效率高、适应性好、结果直观明显，具备在线异常检测的良好应用前景。

英文摘要：

Firstly, abnormal detection principle based on EMD and PE are introduced, and then abnormal signal is extracted by EMD, embedding dimension and delay time in the permutation entropy algorithm are determined by mutual information and false nearest neighbor （FNN） , finally the whole life data of a rolling bearing is taken as an example, the moments corresponding to normal, abnormal and fauh condition are successfully distinguished by the proposed model. The proposed method shares a high efficiency on computation, good adaptive characteristics and intuitionistic result, which gives a good application prospect for on - line abnormality detection.