基于BP神经网络与D-S证据理论的路段平均速度融合方法

发布时间：2018-04-15 15:10

本文选题：智能交通系统 + 路段平均速度　；参考：《交通运输工程学报》2014年05期

【摘要】：为精确估计路段平均速度,提出了基于BP神经网络与D-S证据理论的路段平均速度融合方法。通过训练完成的BP神经网络估计概率密度函数值,进而通过D-S证据理论进行数据融合,整合了BP神经网络自学习的特点与D-S证据理论推理的能力。提出了融合方法的框架,给出了具体的计算模型。利用京藏高速公路上的实测浮动车数据、微波检测器数据、车牌识别数据对融合方法进行了验证,并分析了当微波检测器失效时融合方法的鲁棒性。分析结果表明:融合数据的平均绝对误差百分率比仅使用浮动车数据或微波检测器数据分别提高了7.90%、20.72%,融合方法能够得到较好的效果。微波检测器失效的情况下,融合精度有所下降,但融合数据的误差仍然小于仅使用浮动车数据的误差,说明融合方法具有一定的鲁棒性。
[Abstract]:In order to estimate the average speed of road section accurately, a method of road average speed fusion based on BP neural network and D-S evidence theory is proposed.The BP neural network is trained to estimate the value of probability density function, and then the data fusion is carried out by D-S evidence theory, which integrates the self-learning characteristics of BP neural network and the reasoning ability of D-S evidence theory.The frame of fusion method is presented, and the concrete calculation model is given.The fusion method is verified by the measured floating vehicle data, microwave detector data and license plate recognition data on the Beijing-Tibet Expressway. The robustness of the fusion method is analyzed when the microwave detector fails.The results show that the average absolute error percentage of fusion data is 7.90% higher than that of floating vehicle data or microwave detector data, respectively, and the fusion method can get better results.In the case of microwave detector failure, the fusion accuracy is decreased, but the error of fusion data is still smaller than that of only using floating car data, which shows that the fusion method has some robustness.
【作者单位】：清华大学交通研究所;清华大学土木工程系;
【基金】：国家自然科学基金项目(71361130015)
【分类号】：U491.1

【参考文献】