DEVELOPMENT OF A FUZZY-EXPERT SYSTEM FOR INCIDENT DETECTION AND CLASSIFICATION

This research presents the exploratory results of using fuzzy-expert s ystems for incident detection and classification. The proposed system functions to detect not only the occurrence of incidents, but also the ir located lanes, and the resulting type of severity. With such inform ation; the traffic control center can better advise drivers to take ne cessary lane changes and take timely actions for minimizing the incide nt impacts on traffic conditions. Although the current research remain s exploratory in nature, as both the fuzzy membership functions and ke y parameters were determined empirically rather than fine-tuned with a neural network or genetic algorithm, the preliminary results have con firmed the promising properties of the proposed system. (C) 1998 Elsev ier Science Ltd. All rights reserved.