基于粒子群优化与宽度学习系统的地铁客流预测模型
付建广1尤斌2,3林毅4陈德旺2,3
Metro Passenger Volume Prediction Model Based on Particle Swarm Optimization and Broad Learning System
FU JianguangYOU BinLIN YiCHEN Dewang
-
作者信息:1.1.福州轨道交通设计院有限公司，350003，福州；
2.2.福州大学数学与计算机科学学院，350116，福州；
3.3.福州大学智慧地铁福建省高校重点实验室，350116，福州；
4.4.福州地铁集团有限公司，350004，福州
-
Affiliation:Fuzhou Rail Transit Design Institute Co., Ltd., 350003, Fuzhou, China
-
关键词:
-
Key words:
-
DOI:10.16037/j.1007-869x.2023.05.005
-
中图分类号/CLCN:U293.1+3
-
栏目/Col:学术专论
摘要:
为了提高地铁客流量预测的准确性，基于传统的PSO（粒子群优化）算法与BLS (宽度学习系统)，提出一种新的地铁客流预测模型，即PSO\|BLS算法。首先，对地铁站点的繁华程度、前一时段进站量、前一时段出站量及前一时段断面客流量等参数进行分析，并根据分析结果提出需要根据工作日和双休日分别对地铁客流量进行预测。其次，利用PSO算法对BLS的特征层偏置进行优化。最后，以福州地铁1号线AFC（自动售检票）系统中记录的大量乘客出行数据为例，对所提PSO-BLS算法进行验证。验证结果表明：与传统的地铁客流量预测算法BP（反向传播）神经网络和ELM（极限学习机）相比，PSO-BLS算法获得的计算结果在多项性能指标中均取得了较优异的表现；对BLS的特征层偏置进行优化可以提高BLS的计算精度，为地铁客流量预测提供更精确的计算结果。
Abstracts:
To improve the accuracy of metro passenger volume prediction, a new metro passenger volume prediction model based on the conventional PSO (particle swarm optimization) algorithm and BLS (broad learning system) is proposed, known as PSO-BLS algorithm. Firstly, parameters including the prosperity level of metro station, the inbound passenger volume, the outbound passenger volume and the sectional passenger volume in the previous time period are analyzed, the metro passenger volume of weekdays and weekends are predicted separately according to the analysis results. Secondly, PSO is applied to optimize the characteristic layer bias of BLS. Finally, based on the large amount of passenger travel data recorded in AFC (automatic fare collection) system of Fuzhou Metro Line 1, the proposed PSO-BLS algorithm is verified. The verification results show that: compared with the conventional metro passenger volume prediction algorithm BP (backpropagation) neural network and ELM (extreme learning machine), the calculation results of PSO\|BLS algorithm have achieved better performance in a number of performance indicators; optimizing the characteristic layer bias of BLS can improve the accuracy of BLS and provide more accurate calculation results for metro passenger volume prediction.
过刊浏览
资料下载
友情链接