A Composite Vibration Control Strategy for Active Suspension System Based on Dynamic Event Triggering and Long Short-Term Memory Neural Network
Hui Pang, Mingxiang Wang, Lei Wang, Jibo Luo
Abstract
To improve ride quality and guarantee driving safety of vehicle, a composite vibration control strategy is designed based on dynamic event triggered (DET) mechanism and long short-term memory (LSTM) neural network for active suspension system (ASS) with input dead zone and saturation. First, a quarter-vehicle ASS model is constructed to launch the expected vibration control system containing an appropriate DET controller and a LSTM controller. In which, the DET controller is proposed to reduce the occupancy of communication resources and the inherent Zeno phenomenon of the DET controller can be eliminated. Meanwhile, the LSTM controller is established to make the vertical acceleration of the ASS get closer to zero and thus improve the vehicle ride comfort. The required training data of the LSTM controller is collected by radial basis function neural network-linear quadratic regulator controller. Finally, the effectiveness of the designed vibration controller is demonstrated by the comparative numerical simulations of the ASS, and the results reveal that the proposed controller can enhance the dynamics performances of ASS, compared to existing RNN control and the passive suspension, the vehicle acceleration of ASS with this proposed controller is reduced by 10% and 30%, respectively.