基于深度学习的城市轨道客流预测实训基地培训系统设计与实现
    点此下载全文
引用本文:李 运.基于深度学习的城市轨道客流预测实训基地培训系统设计与实现[J].计算技术与自动化,2022,(2):173-177
摘要点击次数: 55
全文下载次数: 0
作者单位
李 运 (无锡地铁培训学院有限公司江苏 无锡 214023) 
中文摘要:为进行城市轨道客流量的科学预测,设计实现了实训基地培训系统,在系统中分别构建BP(Back Propagation)神经网络模型与DBN(Deep Belief Network)深度置信网络模型来进行城市轨道客流量数据的收集与整理。针对城市轨道交通车流量短时间快速增长的问题,在系统中建立实时数据分析处理模块,结合深度置信网络,使用算法对轨道客流进行实时准确的预测。将采集到的大量数据通过系统的数据处理模块进行分析验证,数据分析结果显示:在深度置信网络模型中,轨道交通各节点平均均方根误差相比BP神经网络模型大约减少了0.01475,每批次任务实时计算时间平均为3.5s,系统预测实时性较好,准确率较高,有很好的预测效果,可以合理地规划城市轨道交通线路,有效提高了城市轨道的交通利用率。
中文关键词:BP神经网络  深度置信网络  城市轨道  客流量  培训系统
 
Design and Implementation of Training System for Urban Rail Passenger Flow Prediction Training Base Based on Deep Learning
Abstract:The training system for training base is designed and implemented to scientifically forecast the flow of urban rail passenger, in which BPNN (back propagation neural network) model and DBN (deep belief network) model are constructed to collect and sort out the flow data of urban rail passenger. Aiming at the problem of rapid growth of urban rail flow in a short time, a real-time data analysis and processing module is established in the system. Combined with DBN, the algorithm is used to predict the rail passenger flow in real time and accurately. A large number of collected data are analyzed and verified by the data processing module of the system,and the results of data analysis show that: compared with BPNN prediction model, the average root mean square error of rail transit nodes is reduced by about 0.01475, and the average real-time calculation time of each batch of tasks is 3.5s. The system has excellent real-time prediction performance, high accuracy, good prediction effect, and can reasonably plan urban rail transit lines, thus effectively improving the utilization rate of urban rail.
keywords:BP neural network  deep belief network  urban rail  passenger flow  training system
查看全文   查看/发表评论   下载pdf阅读器