论文部分内容阅读
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。
In order to accurately estimate the average speed of road sections, an average speed fusion method based on BP neural network and D-S evidence theory is proposed. The BP neural network is trained to estimate the probability density function value, then the data fusion is carried out by the D-S evidence theory, which integrates the characteristics of BP neural network self-learning and D-S evidence theory reasoning. The framework of the fusion method is proposed and a concrete calculation model is given. The fusion method is validated by the measured data of floating car, microwave detector data and license plate recognition data on the Beijing-Tibet Expressway, and the robustness of the fusion method is analyzed when the microwave detector fails. The analysis results show that the average absolute error percentage of the fusion data increases by 7.90% and 20.72% respectively compared with the data of the floating car or the microwave detector, and the fusion method can obtain good results. When the microwave detector fails, the fusion accuracy decreases, but the error of the fusion data is still less than the error of using only the floating car data, indicating that the fusion method has a certain robustness.