Original scientific paper
An incident detection method considering meteorological factor with fuzzy logic
Min Zhao
; Key Laboratory of Cyber Physical Social Dependable Service Computation, Chongqing, PR China; School of Automation of Chongqing University, Congqing, PR China
Xi Chen
; Key Laboratory of Cyber Physical Social Dependable Service Computation, Chongqing, PR China; School of Automation of Chongqing University, Congqing, PR China
Dihua Sun
Abstract
To improve the performance of automatic incident detection algorithm under extreme weather conditions, this paper introduces an innovative method to quantify the relationship between multiple weather parameters and the occurrence of traffic incident as the meteorological influencing factor, and combines the factor with traffic parameters to improve the effect of detection. The new algorithm consists of two modules: meteorological influencing factor module and incident detection module. The meteorological influencing factor module based on fuzzy logic is designed to determine the factor. On the basis of learning vector quantization (LVQ) neural network, the new incident detection module uses the factor and traffic parameters to detect incidents. The algorithm is tested with data collected from a typical freeway in Chongqing, China. Also, the performance of the algorithm is evaluated by the common criteria of detection rate (DR), false alarm rate (FAR) and mean time to detection (MTTD). The experiments conducted on the field data study the influence of different algorithm architectures exerted on the detection performance. In addition, comparative experiments are performed. The experimental results have demonstrated that the proposed algorithm has higher DR, lower FAR than the contrast algorithms, and the proposed algorithm has better potential for the application of freeway automatic incident detection.
Keywords
automatic incident detection; meteorological factor; fuzzy logic learning vector quantization (LQV)
Hrčak ID:
192860
URI
Publication date:
25.1.2018.
Visits: 925 *