APPLICATION OF FUZZY-LOGIC AND NEURAL NETWORKS TO AUTOMATICALLY DETECT FREEWAY TRAFFIC INCIDENTS

To date, efforts to manage freeway congestion have been seriously impe ded by the inability to promptly and reliably detect the presence of t raffic incidents. Traditional incident-detection algorithms distinguis h between congested and uncongested operation by comparing measured tr affic-stream parameters with predefined threshold values. Given the ra nge of possible operating conditions in the traffic stream, selecting a single threshold value, and the suitability of that selected thresho ld, is full of uncertainty. This inherent uncertainty makes fuzzy logi c a promising approach to incident detection. A Fuzzy Logic Incident P atrol System (FLIPS) is proposed to solve many of the problems inheren t in traditional incident-detection algorithms. The FLIPS combines fuz zy logic with the learning capabilities of neural networks to form a c onnectionist model. The system can be constructed automatically from t raining examples to find the optimal input/output membership functions . Threshold values, implicitly obtained by fuzzy-logic rules and membe rship functions, are treated as dependent variables, which change acco rding to prevailing traffic-stream parameters measured by detectors. T he FLIPS avoids the rule-matching time of the inference engine in the traditional fuzzy-logic system. The potential effectiveness of the FLI PS is evaluated using an empirical database collected in Toronto, Cana da. Future refinement to the FLIPS are also discussed in this paper.