为实现滚动轴承故障的智能诊断,提出一种基于经验模式分解和免疫参数自适应支持向量机相结合的滚动轴承故障诊断方法。使用经验模式分解将故障信号分解为若干个本征模函数之和,然后通过定义的故障特征频率筛选函数,自动地从各本征模函数的包络谱中提取出包含轴承外圈、内圈及滚动体故障的特征向量。在特征提取的基础上,将改进的免疫克隆选择算法和K折交叉验证方法相结合,实现了支持向量机参数的自适应优化选取,并进一步训练得到免疫参数自适应支持向量机分类器。通过SKF6203滚动轴承数据实验表明,该方法能获得较高的故障诊断识别率。
To realize intelligent fault diagnosis of rolling bearings, a novel fault diagnosis approach based on Empmcal Mode Decomposition (EMD) and Immune Parameter Adaptive Support Vector Machine (IPA-SVM) was proposed. The fault signal was decomposed into the sum of a number of Intrinsic Mode Functions (IMFs) by EMD. Through fault characteristic frequency filter function, the fault feature vectors were extracted from the envelope spectrum of IMFs automatically. On the basis of feature extraction, an improved immune clonal selection algorithm was introduced to optimize parameters of SVM with K-fold cross validation method, and the IPA-SVM classifier was obtained after sample training. SKF6203 rolling bearing data experiments demonstrated that the superior fault recognition rate could be acquired by proposed method.