Abstract: Aiming at the problem of too high dimension of sample data leading to too much SVM studies in the researches of fault diagnose，This paper proposes a new SVM fault diagnose method based on DCT，using the “energy compaction” and “high frequency suppression” features of DCT，which used in the process of image compression.At first，reduce the dimension of fault samples in light of the method of DCT.Secondly，train the main coefficients of DCT instead of the original samples using SVM so as to restrain the effect of noise for fault classification greatly，and reduce the computation in the process of diagnosis.Finally，the validity of the algorithm is proved by experiment simulation.

Key words: fault diagnose, Support Vector Machine（SVM）, Discrete Cosine Transform（DCT）, high frequency suppression

摘要： 针对故障诊断研究中，样本数据维数过高导致故障模式分类时SVM学习强度太大的问题，利用DCT方法在降噪处理时体现出“能量集中”和“高频抑制”的特性，提出一种基于DCT的SVM故障诊断方法。先对故障样本进行DCT降维，再利用SVM方法对主要维离散余弦系数进行模式训练来代替对故障样本的直接训练，从而大大抑制了噪声对故障分类的影响，同时也减少了诊断运算量，最后通过实验仿真验证了算法的有效性。

关键词: 故障诊断, 支持向量机（SVM）, 离散余弦亦换（DCT）, 高频抑制