PID and ANN-based Control of Passive and Semi-Active Suspension System using Different Training Algorithms

Abstract

The study considers Proportional-Integral-Derivative (PID) and Artificial Neural Network (ANN) approaches to control quarter-car passive and semi-active suspensions. The mathematical models have been derived and simulated in MATLAB Simulink. The models were initially controlled using a PID controller, and the results were used to train the ANN model. The conclusions highlight a better response of ANN controller compared to the PID controller. ANN training was performed using the Levenberg-Marquardt (LM), Bayesian Regularisation (BR) and Scaled Conjugate Gradient (SCG) algorithms. The responses of the ANN controllers were compared in terms of settling time and overshoot. The fallouts confirmed better outcomes of ANN controller trained through the BR-algorithm compared to other training algorithms.