Fuzzy inference approach in traffic congestion detection

One of the major tasks within the concept of an intelligent transportation system is the immediate indication of traffic breakdowns. A conventional approach evaluates a traffic condition by classifying (1) traffic volume and (2) vehicles average speed. This mathematical approach is acceptable and leads to good results as long as the analyzed data correctly represents the observed situation. However, both traffic situations and behaviour of individual drivers cannot be foreseen. In such circumstances, ‘crisp’ computational models cannot deal effectively with accompanied ambiguities and uncertainties. An alternative approach is to apply fuzzy logic systems, which enable knowledge-based analysis for effective and efficient traffic congestion detection. In this paper, traffic flow and density are inputs for the proposed fuzzy inference model and the output comes in form of detected levels of congestion (ranging from ‘congestion free’ to ‘extreme congestions’ conditions). The results show that fuzzy logic inference model for congestion detection might be highly suitable for transportation planning, management and security assessment.