Capturing Temporal Information with LSTM to Stabilize A Rotating Machine

Yaoyao Wei, Weiming Qu, Tao Zhang, Yian Deng, Fan Hu, Zhang Xiangqi, Tianlin Liu, Paul Bolchover, Tao Yu, Guishui Zheng, Ping Zhang, Qing Liu, Dingsheng LuoICCV 2022 · DOI

Abstract

Rotor stability, which indicates the performance of a rotor keeping the system in good working order without transverse vibration, is one of the important research contents in rotor dynamics area. Actually, in a real working environment, it is difficult to stabilize a rotating machine because violent transverse vibrations often happen due to the influence of air resistance, ground friction, errors in manufacture of the rotor, and other factors. The vibrations not only cause instability of the system, but also greatly limit the increase of the rotational speed. We think that the temporal information of the rotating machine, which records the history of its motion, is useful to help reduce the vibrations and stabilize the rotating machine. A recurrent neural network architecture, Long Short-Term Memory (LSTM) is employed to capture the complex and changeable temporal information. So, in this research, an intelligent control approach, that is deep reinforcement learning coupled with LSTM capturing temporal information, is proposed to stabilize a rotating machine. Experimental results in simulation platform demonstrate the effectiveness and robustness of the approach.
BibTeX
@inproceedings{wei2022capturing,
  author = {Yaoyao Wei and Weiming Qu and Tao Zhang and Yian Deng and Fan Hu and Zhang Xiangqi and Tianlin Liu and Paul Bolchover and Tao Yu and Guishui Zheng and Ping Zhang and Qing Liu and Dingsheng Luo},
  title = {{Capturing Temporal Information with LSTM to Stabilize A Rotating Machine}},
  booktitle = {IEEE/CVF International Conference on Computer Vision},
  year = {2022},
  doi = {10.1109/cvidliccea56201.2022.9825467},
}

Loading related content...

Figure