摘要
针对核主元分析算法参数设置和主元数量选取问题,提出一种并行优化核函数参数和主元数量的改进核主元分析算法。该算法以类别可分性为准则,应用自适应遗传算法同步对核参数和主元数量进行优化,实现了核参数和主元数量的并行选择。将改进后的核主元分析算法应用于柴油机气阀机构典型故障的特征提取中,结果表明:优化核主元分析能有效降低柴油机气阀机构故障特征向量的维数,提高各类样本的聚类效果。
Aiming at the problems of kernel parameter setting and principal component selection in kernel principal component analysis, an improved kernel principal component analysis （KPCA） algorithm with parallel optimization of kernel function parameters and the number of the principal components was put forward. Based on the principle of category separation, the adaptive genetic algorithm was applied to optimization of the kernel parameters and the number of the principal components synchronously, and the parallel selection of the kernel parameters and the number of the principal components were realized. Then, the improved KPCA algorithm was applied to the feature extraction of typical faults for diesel valve mechanism. The results show that the optimal KPCA can effectively reduce the dimensions of fault feature vectors, and improve clustering effect of various categories of samples.
出处
《噪声与振动控制》
CSCD
2013年第2期19-22,共4页
Noise and Vibration Control
关键词
振动与波
核主元分析
遗传算法
并行优化
特征提取
vibration and wave
kernel principal component analysis
genetic algorithm
parallel optimization
feature extraction
