Application of machine learning method to control the vibration of the car's suspension system

Abstract

A machine learning method (MLM) developed based on fuzzy logic control (FLC) has been proposed to control the vibration of the car's suspension systems. To apply the MLM, a quarter car model with two degrees of freedom has been established via MATLAB/Simulink. The reduction of the vertical car body's acceleration response is the objective function. Through FLC's logic rule and data maps of the different road surfaces, the MLM of ANFIS has been trained via its self-learning process to control the car's suspension system. The result indicates that the vibration response of the car's suspension system controlled by both FLC and MLM is reduced in comparison without the control. Particularly, the MLM has an obvious effect on reducing the acceleration response of the car body under different excitations of the road surface in comparison with the FLC. Consequently, the car's suspension system controlled by the MLM could further enhance the car's ride comfort compared to the traditional control methods.