The self-tuning fuzzy sliding mode control method for the suspension system with the LSTM network road identification

Abstract

As an important component of semi-active suspension (SAS) systems, the built-in solenoid valve adjustable damper (BSVAD) is now widely used in the field because of its advantages of fast response, continuously adjustable damping, and lightweight design. However, only a few studies have discussed the BSVAD-modeling method. Hence, this study proposes a method for constructing an agent model using a feedforward neural network (FNN) for an equivalent BSVAD regulation mechanism. Meanwhile, a new control strategy, i.e., a fuzzy sliding mode control method based on road conditions (RC-FSMC), is proposed for calculating the required damping force. The proposed method identifies and analyzes unsprung mass acceleration characteristics using a long short-term memory network to obtain the road condition of the moving vehicle and inputs the output results into the designed fuzzy rules to adjust the sliding mode switching gain and finally realize road condition-based adaptive control. Finally, through simulation test analysis, the results show that the designed RC-FSMC strategy can effectively reduce vehicle body acceleration and vehicle dynamic load under different road conditions to improve the adaptability of the vehicle to different roads.