摘要
短时交通流预测是实现交通流诱导的关键技术之一.针对目前短时交通混沌预测模型预测结果差异较大的问题,归纳了4种基于混沌理论的短时交通流预测模型:RBF神经网络模型、最大Lyapunov指数模型、局域线性模型和Volterra滤波器自适应预测模型,并对这4种预测模型进行了比较研究.应用4种预测模型对几个典型的非线性系统进行预测,验证了算法的准确性.然后用这4种预测模型对微观实测交通流的时间序列进行实证分析.仿真结果表明,4种预测模型对典型混沌时间序列具有很好的预测效果;而对实测交通流预测,其预测精度和稳定性较差,但可以满足实时交通流预测的需要.
The prediction of short-term traffic flow is one of the key to realize traffic flow guidance. Aiming at tile prediclfion precision problem using chaotic prediction model for short-time traffic flow, the 4 prediction model in short-term traffic flow base on chaos theory, such as radial basis function neural network (BRF) model, Lyapunov exponent model, local- region prediction model and Volterra filter model are introduced. Besides, tile 4 prediction models were comparative studied. Firstly, the time series of several typical nonlinear systems are predicted by the 4 prediction model in order to confirm the veracity of results of the algorithm. Secondly, tinm series of real traffic flow are researched with them. The simulation results show that the proposed 4 prediction model has effective prediction results for typical nonlinear systems. And its forecasting precision and stability are usually poor when the real traffic flow was predicted with them, but they can meet real-time prediction of traffic flow completely.
出处
《数学的实践与认识》
CSCD
北大核心
2011年第17期106-114,共9页
Mathematics in Practice and Theory
基金
国家自然科学基金(50478088)
关键词
交通工程
交通流预测
混沌理论
相空间重构
traffic engineering
traffic flow prediction
chaos theory
phase space reconstruc-tion