摘要
针对现代机械复杂化、智能化的特点,为快速准确地诊断出设备故障,提出了基于PCA与蚁群算法的机械故障聚类诊断新方法。定义了聚类准确率判别因子,对主元的选取进行自适应调整,利用基于高斯径向基核函数的主元分析方法实现了故障特征提取。以蚁群算法解决旅行商问题为原型,定义了城市圈,改进蚁群算法实现了双重寻优,把故障聚类转化为蚁群算法最擅长的寻求最优解问题,将改进的蚁群算法用于故障特征样本的聚类。实例分析证明了该方法的有效性。
A new method of clustering for mechanical fault diagnosis based on PCA and ant colony algorithm was put forward for modern machinary because of the complexity and intelligence. A cluste- ring accuracy discriminati feature extraction was rea was transformed into find ant colony algorithm. The rithm. The improved ant on factor was defined to adjust principle component. The mechanical fault lized based on Gauss RBF kernel function of the PCA. The fault clustering out optimal solution for the model of traveling salesman problem based on city circle was also defined to realize double optimization by ant colony algo- colony algorithm was used for fault features of the sample clustering. The new method is effective by experiments.
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
2013年第24期3333-3337,3344,共6页
China Mechanical Engineering
基金
国家自然科学基金资助项目(51175169)
湖南省科技计划资助项目(2009FJ4055)
湖南省教育厅重点实验室开放基金资助项目(10K023)
关键词
主元分析
蚁群算法
聚类分析
故障诊断
principle component analysis（PCA）
ant colony algorithm
cluster analysis
fault diag nosis
