摘要
针对雨雾环境下的行车安全问题,构建动态控速模型,旨在合理确定行车速度,减小不安全因素对行车安全的影响.利用小波降噪法对行车数据进行处理,通过长短期记忆神经网络(LSTM)方法对动态速度数据训练,并根据对应路段的历史车速预测新的车速,以预测车速的 85%位车速作为参考,设置对应路段的限速值.为了检验所确定限速值的实际效果,以加减速力度、踩踏油门踏板及制动踏板的频率作为检测因子.实验结果表明,LSTM模型在处理对时间具有依赖性的交通速度预测问题上具有很好的优越性,在设定的新限速值下,驾驶员在行车过程中加、减速力度明显减弱,急加速和急减速频率减少,实验路段高速行驶车辆数量减少,交通流趋于平稳.
Abstract
Aiming at the driving safety problems under rain and fog environment,a dynamic speed control model was established in this paper to determine driving speed reasonably and reduce the influence of unsafe factors on driv-ing safety.The wavelet de-noising method was used to process the driving data,and the LSTM method was adopted to train the dynamic speed data,and then the speed was predicted according to the historical speed of the corresponding road section.The speed limit value of the corresponding road section was set with 85%speed of the predicted speed as reference.In order to test actual effect of the speed limit value determined,acceleration and deceleration force,as well as pedaling frequency of accelerator and brake pedal were used as testing means.Experimental results show that the LSTM model has the superiority in dealing with the traffic speed prediction problem dependent on time;Under the new speed limit set,the driver's acceleration and deceleration efforts are significantly reduced during driving,and the frequency of rapid acceleration and deceleration is reduced.The number of vehicles driving at high speeds on the ex-perimental section is reduced,and the traffic flow tends to stabilize.
NETL
NSTL
维普
万方数据