Research On Path Optimization Of Distributed Dynamic Route Guidance System Of Intelligent Transportation System

With the rapid advancement of China's economy, the accelebrating process of urbanization, the phenomenon of traffic jam is more and more serious. The urban traffic problems such as traffic environmental pollution has become a hot topic of universal. Since entering new century, the intelligent transportation system draws people's great attention. As the core of the intelligent transportation system, dynamic route guidance system's main function is to make full use of the existing traffic facilities, improving the efficiency of the transportation network. According to the path induced in the control center or the on-board equipment, the dynamic route guidance system is divided into central type dynamic route guidance system and distributed dynamic route guidance system. In this paper, the distributed dynamic route guidance system was studied.This paper expounded the concept and framework of the intelligent transportation system structure, and fully analyzes the functional requirements of distributed dynamic route guidance system, and designs the function of system framework. This paper studied For two main points in distributed dynamic route guidance system: one is the short-term traffic flow prediction; another one is path optimization.The paper analyzed and summarized several classical short-term traffic flow prediction methods, and put forward a kind of weighted combination of short-term traffic flow prediction model based on improved nearest neighbor nonparametric regression and wavelet neural network by combining with the thought of combination forecast. In this paper, the nearest neighbor nonparametric regression forecast method can be improved based on the correlation coefficient, which can increase the effect of prediction accuracy. The way that the wavelet basis function neural network model replaces the transfer function in the neural network model integates in-depth between them, which forms into a feedforward network. The portfolio model, which combines the improved nearest neighbor nonparametric regression forecasting precision with wavelet neural network learning, contains the advantages of algorithm simple and fast convergence speed. Through the algorithm simulation of the traffic flow data and the results of simulation analysis, the combination for ecast model has better prediction precision.Knowing the accurate traffic informations is the key to the dynamic path optimization, Intersection delay is a important part of the whole travel time, and it is analyzed and gives the calculation method of travel time with intersection delay in this paper. This paper referenced to research of previous scholars on vehicle emission model, and structured the light vehicle emission model which was suitable for distributed dynamic route guidance system, and built the double objective optimization model based on travel time and vehicle emissions. In order to adapt to the intersection delay model of transportation network in this paper, the traditional Dijkstra algorithm was improved, and it put forward a kind of interactive double targets of the optimal path algorithm based on K shortest path algorithm. The effectiveness of the proposed algorithm was also verified by numerical examples.