关键词: 电梯曳引机, 轴承故障, 多分量, 多尺度排列熵, 孪生支持向量机

Abstract: This study investigated the multi-component and multi scale diagnostic technology for elevator traction
machine bearing faults. By fully refining the bearing vibration signal and mining the characteristics at different scales of the
signal, more comprehensive and reliable bearing fault diagnosis results could be obtained. Using the multi-component
analysis technique based on local mean decomposition ( LMD) , the vibration signal of the elevator traction machine bearing
was decomposed to obtain multiple product function ( PF ) components. After completing the PF component screening
through cross-correlation coefficients, the PF component was reconstructed. Using a multi scale analysis method based on
multiscale permutation entropy ( MPE) , the permutation entropy of each reconstructed PF component at different scales was
calculated, which was used as a feature for elevator traction machine bearing fault diagnosis. A feature vector was
constructed and input into the fault diagnosis model built by the twin support vector mechanism to obtain the fault diagnosis
results of elevator traction machine bearings. The experimental results showed that this technology can effectively decompose
vibration signals under different fault states to obtain PF components and complete their screening. Accurate diagnosis of
different types of elevator traction machine bearing faults could be obtained.

Key words: elevator traction machine, bearing malfunction, multi-component, multiscale permutation entropy( MPE), twin support vector machine ( TSVM)