Eccentricity Fault Detection of Permanent Magnet Synchronous Linear Motor Based on Rainbow Texture Transition and ResNet18 Classification

This paper focuses on eccentricity fault detection of permanent magnet synchronous linear motors (PMLSM). Firstly, based on finite element analysis (FEA), space stray magnetic density signals (SSMDS) of PMSLM under different static and dynamic eccentricity faults are obtained. Secondly, rainbow texture diagram transition (RTDT) method is applied to translate ID SSMDS into 2D colorful images to realize fault feature visualization enhancement. Then, ResNetl8 as an excellent deep learning classification method is used to realize eccentricity fault accurate recognition and classification by fault feature datasheets training and learning. Finally, the superiority and reliability of proposed method is confirmed by comparison experiments with other popular methods.