Research On Bluetooth Location Technology In Underground Space Of Urban Rail Transit

With the continuous development of urban underground space,location information has become an important foundation of big data in the new era of the Internet of everything.However,the acquisition of location information in the unexposed space is still in an early exploration stage.The underground space of urban rail transit is a typical unexposed space.If the location service is realized by using appropriate positioning technology in this environment,it is of great social and economic value to both ordinary passengers and operators.This paper studies the positioning technology of urban rail transit underground space,through the comparison of various positioning technologies,focuses on the application of Bluetooth fingerprint positioning technology in subway stations.The main research and achievements are as follows:(1)Aiming at the problem that there are different influencing factors in the construction of bluetooth fingerprint in subway station.First of all,this paper carries on the environmental survey of all kinds of subway stations,and formulates the corresponding AP layout scheme for different types of subway stations according to all kinds of spatial characteristics.Secondly,the influence factors of fingerprint location environment are analyzed.In view of the impact of human flow on RSS,Kalman filter is used to smooth the signal samples in the dense area of human flow.In view of the lack of consideration of application scenarios and AP stability in the traditional AP selection algorithm,through the analysis of the location environment AP region and volatility on the basis of the traditional AP selection algorithm,this paper combines the AP territorial characteristics,the maximum average value and the stability ranking method to realize the optimization of AP combination.Finally,the positioning accuracy of the proposed AP selection method is tested through seven test points in subway stations.The test results show that the average positioning error is 1.09 m,and the average positioning error is more than 0.45 m lower than that of the traditional AP selection method,which verifies the effectiveness of the comprehensive AP selection strategy proposed in this paper.(2)Aiming at the problem that the applicability of traditional Bluetooth fingerprint database in subway station is not high.This paper optimizes the construction of fingerprint database in the off-line stage of traditional fingerprint location,and puts forward a Metro-Btloc location method which is suitable for subway environment.First of all,the method integrates the original fingerprint data to obtain the initial fingerprint database,then carries on the clustering division,and finally simplifies the fingerprint database with the comprehensive AP selection strategy to improve the fingerprint quality.Among them,in order to solve the problem that the initial fingerprint database is too complex,and the clustering results are easy to fall into local optimization because of the randomness of the initial value selection in the traditional K-means clustering algorithm,this paper proposes a GAWK-means clustering algorithm,which optimizes the weight of the Euclidean distance in the objective function according to the discreteness of the fingerprint data itself,so that it can better reflect the intra-class similarity.The improved K-means is combined with genetic algorithm.Through three experimental tests,the DBI index of GAWK-means reaches 0.5591,0.4153 and 0.4091 respectively,which is better than the traditional clustering algorithm.(3)In this paper,the proposed Metro-Btloc positioning method is tested and verified in the subway station environment.The average location error of the subway station Bluetooth fingerprint location method proposed in this paper is 1.41 m,and the on-line matching time is 0.0153 s.Compared with the unimproved method,the average location error is reduced by more than 0.6m and has better matching efficiency.In terms of positioning stability,a total of 1500 groups of data were tested at each test point,and the average recognition accuracy was 80.52%.The overall experimental results show that the location method proposed in this paper is feasible in the subway station environment.