基于麻雀搜索算法和长短期记忆神经网络的轨道交通站点客流预测  

Passenger Flow Prediction of Urban Rail Transit StationsBased on ISSA-LSTM Model

作　　者：张开雯 何勇 余家香 陈林 ZHANG Kaiwen;HE Yong;YU Jiaxiang;CHEN Lin(School of Mathematics,Physics and Data Science,Chongqing University of Science&Technology,Chongqing 401331)

机构地区：[1]重庆科技大学数理与大数据学院,重庆401331

出　　处：《四川师范大学学报（自然科学版）》2025年第1期105-113,共9页Journal of Sichuan Normal University（Natural Science）

基　　金：重庆市科技局自然科学基金(CSTB2022NSCQ-MSX0256);中国博士后第71批面上项目(2022M712619);重庆市教委科学技术研究计划重点项目(KJZD-K202201502)。

摘　　要：准确的短时客流预测可以为城市轨道交通的良好运营提供保障,但轨道交通的短时客流具有非线性和高随机性等特点,为了提高对短时客流的预测精度,提出将ISSA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SSA算法收敛速度慢,容易陷入局部最优解的问题,引入黄金莱维飞行策略,通过动态调整探索者移动步长的方法,使得它在未知范围内搜索时,能够覆盖更大的范围,提高SSA算法全局搜索的能力.通过使用ISSA算法对LSTM模型的隐含层、学习率和迭代次数的神经元个数进行优化,构建ISSA-LSTM组合预测模型,用于城市轨道交通短时客流的预测.将该模型与BP、LSTM和SSA-LSTM等3种短时客流预测模型进行对比,结果表明:在针对工作日和非工作日客流的预测中,ISSA-LSTM模型预测误差最小,具有较好的预测效果.Accurate short-term passenger flow prediction can provide a guarantee for the good operation of urban rail transit.However,the short-term passenger flow of rail transit has characteristics such as nonlinearity and high randomness.In order to improve the prediction accuracy of short-term passenger flow,this paper proposes to combine the ISSA algorithm and LSTM model to construct a prediction model of the short-term passenger flow for urban rail transit.In response to the problem of slow convergence speed and easy local optimum solution of the SSA algorithm,the Levy flight strategy is introduced to dynamically adjust the explorer s moving step length,which enables it to cover a larger range when searching in an unknown range and improves the global search capability of the SSA algorithm.By using the relationship between the sine function and the unit circle and introducing the golden section coefficient to narrow the solution space,the explorer can traverse all positions on the circle and quickly search for the area where good results may be obtained,accelerating the search speed of the SSA algorithm.By using the ISSA algorithm to optimize the number of neurons in the hidden layer,learning rate and the iteration times of the LSTM model,we have constructed the ISSA-LSTM combined prediction model for short-term passenger flow of urban rail transit.This model has been compared with three other prediction models for the short-term passenger flow,including BP,LSTM and SSA-LSTM.The results show that the ISSA-LSTM model has the smallest prediction error and better prediction performance in predicting passenger flow on workdays and non-workdays.

关 键 词：短时客流预测 改进麻雀搜索算法 长短时记忆神经网络 组合模型 

分 类 号：U121[交通运输工程]