基于IPSO-LSTM组合模型的城市轨道交通短时客流预测  

Short-Term Passenger Flow Prediction of Urban Rail Transit Based on IPSO-LSTM Combined Model

作　　者：赵明伟 张文胜 ZHAO Mingwei;ZHANG Wensheng(No.1 Operation Co.,Ltd.,Jinan Rail Transit Group Co.,Ltd.,Jinan 250300,Shandong,China;School of Traffic and Transportation,Shijiazhuang Tiedao University,Shijiazhuang 050043,Hebei,China)

机构地区：[1]济南轨道交通集团有限公司第一运营有限公司,山东济南250300 [2]石家庄铁道大学交通运输学院,河北石家庄050043

出　　处：《铁道运输与经济》2022年第2期123-130,共8页Railway Transport and Economy

基　　金：中央引导地方科技发展资金项目(206Z0801G);河北省引进国外智力项目(2020);石家庄市科学技术研究与发展计划(211130204A)。

摘　　要：准确的短时客流预测能够为城市轨道交通的良好运营提供保障,为提高预测的精度,提出将IPSO算法和LSTM模型进行组合,构建城市轨道交通短时客流预测模型。针对PSO算法不能很好地区分全局搜索和局部搜索,易陷入局部极值的问题,引入自适应变化的惯性权重和时间因子,动态调整粒子的移动步长,提高PSO算法全局搜索的能力;借鉴遗传算法中的变异机制,引入自适应变异函数,使PSO算法具有跳出局部范围的能力。利用IPSO算法对LSTM模型的迭代次数、学习率和隐含层的神经元个数进行寻优,构建IPSO-LSTM组合预测模型,对城市轨道交通短时客流进行预测,并与BP,LSTM,PSOLSTM共3种短时客流预测模型进行对比,在针对工作日和非工作日客流的预测中,结果显示IPSO-LSTM模型的预测误差最小,具有较好的预测效果。Accurate short-term passenger flow forecasting can provide a guarantee for the good operation of urban rail transit.For higher predictive accuracy,a combination of the IPSO algorithm and the LSTM model was proposed to construct a short-term passenger flow prediction model for urban rail transit.As the PSO algorithm cannot well distinguish between global search and local search and is prone to fall into the local extremum,the adaptive inertia weight and time factor were introduced to adjust the moving step size of particles dynamically and improve the global search ability of the PSO algorithm.In addition,the mutation mechanism in the genetic algorithm was introduced into the adaptive mutation function,and thus the PSO algorithm could jump out of the local scope.The IPSO algorithm was used to optimize the number of iterations,learning rate,and the number of neurons in the hidden layers of the LSTM model,and the IPSO-LSTM combined prediction model was built to predict the short-term passenger flow of urban rail transit.The results show that compared with BP,LSTM,and PSO-LSTM,the IPSO-LSTM model has the smallest prediction error and a better prediction effect of the passenger flow prediction on both working days and non-working days.

关 键 词：城市轨道交通 短时客流预测 改进粒子群算法 长短时记忆神经网络 组合模型 

分 类 号：U121[交通运输工程]