Research On Distributed Shared Bicycle Location Algorithms

Shared bicycle systems have become increasingly popular in recent years,and shared bicycles are evolving into solutions for most Chinese urban,providing urban residents with faster,more environmentally friendly urban travel services.At present,the bicycle positioning is based on Global Position System,and the data is transmitted internally through the Global System for Mobile communication sensor to communicate with the data cloud,and the shared bicycle position and the current usage status of the shared bicycle are transmitted to the service control center in a period of time.The mainstream bicycle distribution method is to realize the redistribution of shared bicycles by using the Bike sharing Rebalancing Problem and using additional vehicles.However,currently every bicycle requires a built-in power supply,GPS module,GSM module and smart lock,which is expensive.And it increases the difficulty of riding in complex environments,such as high-rise buildings and overpasses.The centralized placement of shared bicycles equipped with GPS positioning chips will result in different degrees of positioning drift of 60%-70%.Along with the increasing number of shared bicycles,the shared bicycle system usually has an imbalance of “supply in short supply” and “supply oversupply”,which has caused a serious “tidal phenomenon”.In this paper,a comprehensive analysis is made on the problems arising from the above-mentioned shared bicycle system,and a new solution is proposed in combination with machine learning in the traditional technology.This article mainly discusses the following aspects:(1)Analyzed the GPS-based positioning technology and principle,as well as the positioning principle of wireless sensor networks,and summarized the functions,structures,technical indicators and positioning methods of various components of the system and wireless sensor networks.The two positioning modes of GPS positioning and wireless sensor positioning are compared and summarized.The typical algorithm for solving the imbalance of shared bicycle distribution is reviewed.(2)A new method for sharing bicycles is proposed.In order to better improve the positioning accuracy of shared bicycles,reduce costs and reduce operations,combined with the idea of machine learning development,a shared bicycle positioning algorithm LDS based on distributed subspace is proposed.The traditional positioning method is transformed into the classification learning problem of different sensor nodes.The designed feature vector is the Euclidean distance between the unknown node and the anchor node.The nodes are classified according to the similarity between the two points of the unknown node and the anchor node,and then further positioned in the subspace.Experimental results show that the designed algorithm effectively improves the positioning accuracy.(3)For the problem of shared bicycle distribution,on the basis of the existing algorithms,deep learning is used to extract the characteristics of the number of shared bicycles in different distribution areas,and the automatic optimization learning is carried out in different regions at different times according to the characteristics of different regions.By interacting with the environment to know the next action,setting the reward value of the reinforcement learning,shifting the number of bicycles between the regions into different actions,and achieving the adaptive distribution between the shared bicycle regions by obtaining the maximum reward value.The experimental results show that in the context of the known environment,the different time ends of multiple regions are considered comprehensively.The proposed new algorithm achieves the balance of bicycle distribution in the case of long time and multiple regions,and reduces the allocation between regions.