POSSIBLE APPLICATION OF ARTIFICIAL NEURAL NETWORKS TO VEHICLE DYNAMICS AND CONTROL - A LITERATURE-REVIEW

It is a well-known fact that difficulties are generally encountered wh en mathematical modelling is performed on vehicle systems and sub-syst ems that are inherently non-linear, Conventional modelling techniques - simulation, where non-linearities are approximated by simple mathema tical formulas or look-up tables; finite element approximation, where the system is replaced by a linear model if the range of variables is small; and numerical techniques - have been very successful in modelli ng complex, non-linear systems. Under certain conditions, however, the approximations and hypotheses developed for each of these models fail to reflect the true behaviour of the system. The non-linearity of veh icle systems and sub-systems is particularly severe when the vehicle i s pushed toward its performance limits. The result is that conventiona l modelling techniques become progressively more inaccurate. Recent de velopments in the area of artificial neural networks (ANNs) may provid e an alternative approach to the modelling of vehicular dynamics, part icularly for highly non-linear systems that near their performance lim its. The objective of this paper is to direct attention to the possibl e applications of ANNs to vehicle system dynamics and control. A liter ature search of work related to the application of ANNs to vehicle sys tems is presented and an attempt made to propose possible applications of ANNs to vehicle dynamics and control.