摘要
针对城市短时交通流随机波动性强、可靠性低、预测精度差等问题,将变分模态分解(VariationalMode Decomposition,VMD)和改进麻雀搜索算法(ImproveSparrowSearchAlgorithm,ISSA)与长短期记忆(LongShort-Term Memory, LSTM)神经网络相结合,建立一种短时交通流预测模型(VMD-ISSA-LSTM)。首先利用VMD对历史原始交通流数据进行分解;然后采用佳点集、正弦函数扰动和Tent混沌映射等策略对标准的SSA算法加以改进,增强ISSA算法的寻优能力;最后,将每个分量送入ISSA-LSTM中进行预测,同时将预测结果线性叠加,得到交通流量预测值。以上海市中山北路-曹杨路口2018年11月1日—30日的历史交通数据对模型进行验证。结果表明,与LSTM、VMD-LSTM、VMD-SSA-LSTM等传统预测模型相比,VMD-ISSA-LSTM模型的预测结果的平均绝对百分比误差为1.278 4%,能够更好地应用于短时交通流预测中。
In allusion to the problems of strong random fluctuation,low reliability and poor prediction accuracy of urban short-term traffic flow,a short-term traffic flow prediction model(VMD-ISSA-LSTM)is established by coupling variational mode decomposition(VMD)and improved sparrow search algorithm(ISSA)with long short-term memory(LSTM).VMD is used to decompose the historical original traffic flow data.Then,the standard SSA algorithm is improved by means of the good-point set,sine function perturbation and Tent chaotic mapping strategy to enhance the optimization ability of ISSA algorithm.Each component is sent to ISSA-LSTM for prediction,and the prediction results are linearly superimposed to obtain the traffic flow prediction value.The model is verified by the historical traffic data from November 1,2018 to November 30,2018 at the intersection of Zhongshan North Road and Caoyang Road in Shanghai.The results show that in comparison with the traditional prediction models such as VMD-SSA-LSTM,LSTM and VMD-LSTM,the average absolute percentage error of the prediction results of the VMD-ISSA-LSTM model is 1.2784%,which can be better applied to short-term traffic flow prediction.
作者
庞学丽
宋坤
姚红云
李一博
曹志富
PANG Xuei;SONG Kun;YAO Hongyun;LI Yibo;CAO Zhifu(College of Traffic&Transportation,Chongqing Jiaotong University,Chongqing 400074,China;College of Electronic and Information Engineering,Guangdong Ocean University,Zhanjiang 524088,China)
出处
《现代电子技术》
北大核心
2024年第8期31-36,共6页
Modern Electronics Technique
基金
国家自然科学基金青年科学基金项目(51008321)
重庆市教育委员会-青年项目(KJQN202100715)
重庆交通大学研究生科研创新项目(2022S0035)。
关键词
短时交通流预测
变分模态分解
改进麻雀搜索算法
长短期记忆神经网络
佳点集
正弦函数扰动
Tent混沌映射
short-term traffic flow forecasting
variational mode decomposition
improved sparrow search algorithm
long short-term memory neural network
good-point set
sine function perturbation
Tent chaotic map
