Rapid development of the number of motor vehicles has caused a series of problems, e.g. traffic jam, road safety, environmental pollution and energy consumption, while the effective traffic flow prediction can be used in real-time traffic management and road condition analysis. Traffic flow has randomness and dynamism, as well as regularity in a long term, and it can be affected by working days and external conditions. In this paper, it studies on accuracy of the improved genetic algorithm in traffic flow prediction analysis on the premise of intelligent prediction theory, and seeks for the effective traffic flow data prediction method. The study results indicate that: the improved genetic algorithm model has better prediction performance in the application of traffic flow prediction analysis, and it can realize the minimum relative error with the original traffic flow. In condition that network structure is determined, the application of the improved genetic algorithm can obtain better weight and threshold via the use of objective function. Prediction accuracy of the improved genetic algorithm model is higher than BP neural network algorithm model and regression neural network algorithm, and the improved genetic algorithm of BP neural network may expense more time. The study in this paper can provide real-time traffic management and guidance with instant information, and relieve the pressure of traffic jam effectively.
Keywords: traffic flow; traffic management; intelligent prediction theory; improved genetic algorithm; prediction