基于多块相对变换独立主元分析的故障诊断方法

针对复杂工业过程中故障诊断技术存在数据可分性差、噪声干扰、故障定位困难的问题,提出一种基于多块相对变换独立主元分析(MBRTICA)的故障诊断方法.为了使所提取的故障特征具有可分性,采用相对变换原理与FastICA算法融合的方式构建相对变换独立主元分析方法(RTICA)用于检测故障的发生.通过引入多块理论,将高维数据分成多个子块单元,并在每个子块单元内分别进行RTICA处理,确定故障发生的位置.最后用电主轴轴承裂纹故障的实验对所提方法进行验证,实验结果表明,基于MBRTICA的故障诊断方法可提高数据的可分性,能够有效减少噪声,同时提高故障检测的精度, 实现故障定位功能, 全面地对故障进行分析.

Fault diagnosis approach based on relative transformation ICA of multiblock

A fault diagnosis algorithm based on relative transformation independent component analysis(ICA) of multiblock(MBRTICA) is proposed to solve the problems caused by unsatisfactory data separability, noise interference and difficulties in tracing faults during the industrial complicated processes. In order to improve the separability of extracted fault features, relative transformation ICA method is used to detect the occurrence of faults by combining the relative transformation principle and FastICA algorithm. By introducing the multi-block theory, the high-dimensional data is divided into multiple sub-blocks, and the RTICA algorithm is performed in each sub-block unit to determine the location of the fault. The proposed algorithm is validated by the spindle experiment with bearing crack fault. The experimental results show that, the fault diagnosis algorithm based on MBRTICA can improve the separability of data and effectively reduce the noise, which can improve the fault detection precision, and realize the fault location function to carry out comprehensive fault analysis.