关键词: 交通流预测, 优化控制, 粗糙集理论, 人工神经网络, 粗糙神经网络

Abstract: Based on the Similarity principle of the fractal theory and on the quantitative analysis of the historical traffic flow measured on path network nodes through rough set theory，this paper discussed the mode structure of the real-time dynamic traffic flow and put forward the optimal control method.First of all，the paper established a data cleaning mode on the basis of the rough set theory；then analyzed the parameter structure in detail，established a forecast mode of the traffic flow with the similarity principle of the fractal theory，described the functioning principle of the optimal control system of the traffic flow and the corresponding network topology structure.Finally，the paper tested the mode with the real data measured from a traffic observation station，the result shows that the mode and the algorithm designed is actually effective.

Key words: forecasting traffic flow, optimal control, rough theory, neural network, rough neural network