Font Size: a A A

Predictions Of Indoor Passenger Flow Based On Deep Learning

Posted on:2018-12-17Degree:MasterType:Thesis
Country:ChinaCandidate:B P LiFull Text:PDF
GTID:2348330512999473Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
There are more and more large buildings during rapid urbanization,such as shopping mall,airport and convention center.People spent more time indoors.The WiFi infrastructure not only provide WLAN networks but also provide an approach to monitor indoor passenger flow.It is important for the successful of material and human resources management to predict passenger flow accurate and timely.In this paper,we review the progress of the indoor passenger flow prediction firstly.Then we summarize some problems in practice:ineffectively discriminate indoor movements from outdoor movements base on original WiFi signal strength records.To address the problem,this paper presents a novel approach based on random forest.we also present a method to build training data which inspired by crowdsourcing,and we show that our techniques get high accuracy on our platform.Besides,existing passenger flow prediction methods,mostly time series methods,mainly focus on outdoors and are still based on the context relation with the past records.However,indoor passenger flow is not only related to the past records separately,but also to the past records of different regions.According to the hypothesis,we present a method integrated convolution neural network and long-short term memory and extract spatial and temporal feature tensor to predict passenger flow of different regions in large building.As a result,we verify the reliability of the method with the dataset from Guangzhou Baiyun International Airport,acquire good performance.The mean square error is about 5.Our method is much cleaner and shows better generalization ability in comparison with other methods.
Keywords/Search Tags:Indoor position, Deep learning, Convolutional Neural Networks, Data mining, Feature learning
PDF Full Text Request
Related items