Computational Intelligence Theory Of Traffic Flow Guidance System

Intelligent transportation system is the best acknowledged method to solve problems in traffic field at present, such as traffic jam, traffic block, traffic accident, traffic pollution. Traffic flow guidance system is the core researching subject in intelligent transportation system and also the best way to avoid urban partial traffic jam. Furthermore, the study of traffic flow guidance theory is based on how well real-time dynamic route choice is researched, about which many researchers did some works in variety of aspects, and as well obtained some important achievements. However, the models established often had many problems such as quantities of computing, long time of optimizing and unfitting for large size traffic network. This dissertation, considering fully about the need of studying intelligent transportation system and its application in China, did some researches in the forecasting dynamic route choice theory based on the computing intelligence theory.First of all, according to urban traffic Time of Day control method and dividing approach of non-linear function theory in ANN application, a kind of idea about multi-interval of time forecasting and guidance was put forward. And then, based on a new research harvest in AI field: Artificial Immune System, some preliminary works were done to traffic network state pattern recognition. Moreover, the author also carried simulation on to city traffic state pattern recognition problem with the established artificial immune algorithm. As a result, the simulation succeeded in realizing automatic partition of traffic intervals of time and overcame some deficiencies of artificial partition or the genetic algorithms method, where a new idea for traffic period of time partition was also presented.In addition, after reviewing some forecasting methods like ARIMA, Kalman filter and ANN, the author proposed a short-term traffic flow forecasting model based on WNN theory. Moreover, with the real traffic data of Jingshi road in Jinan city, a simulating experiment was done to compare with the traditional method. The consequence indicates that this model is obviously better than BP network in terms of both forecasting precision and the network astringency, and it has a good prospect in real-time traffic flow forecasting.