关于道路交通流准确预测仿真研究  

On Simulation of Accurate Prediction for Road Traffic Flow

作　　者：常峰[1] 旷文珍[1] 刘海峰[1] 宋苏民 

机构地区：[1]兰州交通大学光电技术与智能控制教育部重点实验室,甘肃兰州730070

出　　处：《计算机仿真》2016年第6期111-116,137,共7页Computer Simulation

基　　金：甘肃省青年科技基金计划项目(1308RJYA096);甘肃省青年科技基金计划(145RJYA251);甘肃省高校科研项目(2013A-050)

摘　　要：在预测研究短时交通流问题的研究中,由于传统的短时交通流预测没有考虑交通流的混沌特性,与实际交通流特性状况不符,预测模型存在不准确性和不稳定性等缺陷,针对上述问题,为了提高交通流预测的精确度,考虑交通流量时间序列的混沌特点,提出混沌理论K熵优化小波BP神经网络进行短时交通流预测模型。首先,用小波基函数对BP神经网络的隐含层进行优化;然后,通过引入K熵理论,判断交通流时间序列的混沌特性;最后,通过混沌理论K熵优化小波BP神经网络对短时交通流进行预测,并分析了在各种不同条件下的预测情况。仿真结果表明用该网络模型对交通流时间序列的预测的准确性和稳定性相对于传统的小波BP神经网络有很大提高,分析得出所建立的基于混沌理论K熵优化BP神经网络的短时交通流预测模型,能够对短时交通流进行准确的预测。Traditional short -term traffic flow prediction does not take the chaos of traffic flow characteristics into account, and is not in accord with the actual characteristics of the traffic flow, the prediction model has some defects such as uncertainty and instability. In order to improve the accuracy of traffic flow prediction, the chaotic characteris- tics of the time series of traffic flow are considered, and the chaos theory K - entropy optimization wavelet BP neural network is introduced to predict the short - term traffic flow. Firstly, the BP neural network with wavelet basis func- tion of hidden layer is optimized. And then the chaotic characteristic of traffic flow time series is determined by intro- ducing the theory of K - entropy. Finally, the short - term traffic flow is predicted through the chaos theory K - entro- py optimization wavelet BP neural network, and the prediction under various conditions is analyzed. The simulation results show that the proposed method can improve the prediction accuracy and stability compared with the traditional wavelet BP neural network.

关 键 词：混沌理论 小波变换 神经网络 交通流预测 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]