基于非线性流形学习的磨粒特征提取方法

针对磨粒特征参数多、非线性突出的问题,提出一种基于非线性流形学习的磨粒特征提取方法。该方法将磨粒特征重构到高维相空间中,利用局部线性嵌入算法提取出隐藏其中的低维流形,并根据数据流形的弯曲性和邻域参数的关系,实现高维相空间中局部邻域参数的自适应选取。实验结果表明,该方法有效地克服了主成分分析和核主成分分析方法的不足,提取的磨粒特征敏感性更好,从而提高了磨粒识别的精度。

Wear Particle Feature Extraction Method Based on Nonlinear Manifold Learning

To deal with the problem of a large number of wear particle feature parameters and nonlinear relationship among these parameters,a new feature extraction method based on nonlinear manifold learning was proposed.After embedding the wear particles feature parameters into a high dimensional phase space to reconstruct a dynamical manifold,the locally linear embedding(LLE)algorithm was employed for extracting low dimensional manifold.According to the relationship between the curvature of the manifold and the neighborhood parameter,the adaptive selection of local neighborhood parameters in phase space was implemented.The experimental results show that this approach,compared with the linear principal component analysis(PCA)and nonlinear kernel principal component analysis(KPCA),is more effective to extract the wear particle feature,and enhances the classification ability of wear particles.