Fuzzy logic-based traffic incident detection system with discrete wavelet transform

This paper presents a fuzzy logic-based traffic incident detection system to detect a lane-blocking traffic incident that usually causes of traffic congestion. The proposed system uses fuzzy logic to identify traffic status as normal and abnormal. Macroscopic and microscopic traffic variables, namely, mean speed and standard deviation of inter-arrival time are used as inputs to the fuzzy inference system (FIS). As traffic variables have many fluctuations which are considered as noisy signals, discrete wavelet transform (DWT) as used for de-noising and also extracting features from noisy signals. It is found that the proposed system that uses DWT can give higher detection rate when compared with the system without DWT. Furthermore, the majority voting is also applied to the outputs of FIS in order to increase detection rate. Finally, based on simulation results, the performance of the proposed detection system for lane-blocking traffic incidents will be shown.