Modelling and control of an intelligent antilock braking system

Abstract

This thesis describes an intelligent approach to control an Antilock Braking System (ABS) employing a Gradient Descent Method for on-line adaptation of Fuzzy-Sliding Mode Controller (on-line FSMC). This is a new method that is applied into the ABS which includes the estimation of learning parameters to minimize the prediction error between the desired input and the actual output in the feedback control loop. The fuzzy system is developed using Adaptive Neuro-Fuzzy Inference System (ANFIS) by training the input-output data of the ABS based on Sliding Mode Control (SMC). This study also describes another new approach to robust control ABS employing Active Force Control (AFC) scheme. The main objective of the proposed control strategies is to control the optimum wheel slip value of the ABS that include nonlinearities, parametric uncertainties and disturbances in order to prevent the controlled wheel from becoming fully locked.. A modelling and simulation study was rigorously carried-out on a simplified quarter car braking model in the straightforward speed (straight line braking) with the applied control schemes taken into account. A comparative study between a number of control strategies was also performed to demonstrate the differences in performance. The simulation result showed that the on-line FSMC strategy demonstrated its superiority in controlling the ABS than the SMC strategy. The simulation results showed also that the AFC scheme demonstrated its robustness even in the presence of disturbance