In this paper, a practical qualitative+quantitative method named S-ANN is proposed as a forecasting tool, in which the artificial neural network (ANN) of AI is used to handle the quantitative knowledge and the SCENARIO method of systems engineering is used to handle the qualitative knowledge respectively. As a case study, S-ANN method is employed to forecast the ridership of Beijing public transportation, the results show that S-ANN method possesses advantages of feasibility and easily to operate.
GUAN Wei, SHEN Jin-sheng, LI Peng-fei Institute of Systems Engineering, Northern Jiaotong University, Beijing 100044, China
在介绍现有的主要交通流预测方法的基础上,阐述了基于卡尔曼滤波(K a lm an)的预测模型及其具体算法。结合城市环路的交通运行特性,构建了基于卡尔曼滤波的交通流短时预测模型,并根据北京市三环路的实际数据对模型进行验证。实证数据表明,所建立的交通流动态实时预测模型的预测效果比较理想,算法的实时性也满足实际预测系统的要求,可应用于交通流预测及交通智能控制。
从20世纪90年代起,原属交通工程学科中的交通流问题因其特有的复杂性得到了各个学科领域的学者的重视,这同时也使得许多交通科学家、数学家、物理学家和经济学家纷纷加入到交通流理论的研究中来.本文介绍了国际上交通流理论的新发现和新进展,如同步流,交通相变和三相交通流理论,并分析对比了各国不同学者理论之间各自的差异.其中最有影响力的就是德国学者Dirk Helbing和Boris S Kerner;本文中特别讨论了他们两者各自的观点,及其主要分歧所在.
