关键词: 仿真数据驱动, 长期服役电梯, 水平振动, 子领域自适应, 故障诊断

Abstract: The existing researches of fault diagnosis of elevator guide rails has some problems, such as scarcity of horizontal vibration classification data and large difference in the distribution of training and test data sets. A simulation data-driven fault migration diagnosis method for long-term service elevator guide rails was proposed. Firstly, the horizontal dynamics model of the elevator car was constructed, different types of guide rail fault excitations as system input for simulation and rich horizontal abnormal vibration data of elevator car were obtained. Secondly, the residual network and convolutional attention mechanism were integrated to extract fault features, and the sub-domain adaptive method was used to align the conditional distribution of source domain and target domain in unsupervised scenarios. Finally, the elevator horizontal vibration data under different working conditions were used as the target domain to verify the proposed method. The experimental results show that the proposed method has high fault diagnosis accuracy in unsupervised cross-domain scenarios, which provides a reference for solving the problems of scarcity of fault data for long-term service elevators.

Key words: simulation data-driven, long-term service elevator, horizontal vibration, subdomain adaptation, fault diagnosis