Adaptive Fuzzy Logic Control of an anti-lock braking system

Over the last few decades digital computer technology has matured to the point that the engineering world has witnessed a revolution in the way in which mathematical problems are solved. Problems too complex to be solved analytically can now be solved in a feasible manner, by employing a computer system which emulates the human learning and decision-making process. Areas such as neural networks, expert systems, and fuzzy logic systems implement `human-like' capabilities. This work focuses on the design of Adaptive Fuzzy Logic Control of an anti-lock braking system. This controller will never know the exact plant model, but will know input-output relations of the plant (training data). The controller will initially employ a priori training data to control the braking system, but will continue to train on-line while continuously updating confidence parameters and placement of fuzzy sets by employing optimization algorithms. Old data will be slowly forgotten while up-to-date training data are acquired. Thus, changes in road conditions or in the plant itself will be learned.