SVR with hybrid chaotic genetic algorithm for short-term traffic flow forecasting

Abstract

Accurate forecast of traffic flow is crucial to effective and proactive traffic management systems in the context of intelligent transportation systems and dynamic traffic assignment. This paper presents an application of a supervised statistical learning technique called support vector regression (SVR) with hybrid chaotic genetic algorithm (CGAs) for urban short-term traffic flow forecasting. With the increase of complexity and the larger scale of traffic flow forecast demand, genetic algorithms (GAs) are often faced with the problems of premature convergence, slowly reaching the global optimal solution or trapping into a local optimum. The proposed algorithm is used to overcome premature local optimum in determining three parameters of the SVR model. The predictive performance is compared to other models and the results show the algorithm can not only overcome the premature of GA but also can increase its robustness, and at the same time reduce the error of traffic flow forecasting, raise the forecast precision.