| In recent years,with the development of the global navigation satellite system(GNSS),location-based services(LBS)gradually become an indispensable part of people's life or work.As "the last kilometer" navigation system indoor positioning is becoming a hot spot of the mobile Internet era,people demand for indoor location service is more old more urgent.But indoor environment is variable and complex,patterns vary,there are many interference,it is difficult to application in indoor.The key scientific problems of indoor positioning error theory and complex indoor environment mechanism has not been fundamentally solved,such as indoor positioning stability,continuity,reliability and high accuracy is still a technical bottleneck,which limits the application of indoor location services.According to the above problem,this paper carried out indoor positioning technology based on the intelligent terminal and in this paper,intelligent algorithm of the multiple information fusion are studied,the complete content and contributions are as follows.This article put the indoor positioning system is divided into two types of absolute positioning system and relative positioning system,and context design scheme of two different sets of indoor respectively.1)The relative positioning system,the use of intelligent terminals in carrying a variety of sensors and special module is implemented pedestrian navigation calculation(PDR)algorithm and do some improvement,based on the improved PDR algorithm,according to the algorithm due to long-term work produces the cumulative error problem,designed four kinds of different ways of correction calibration location,the design for indoor positioning system based on intelligent terminal,and evaluate the system performance in the different experimental scene,average positioning accuracy is within one meter,meet the requirements of indoor location.2)The absolute positioning system,Based on position fingerprint,completed two parts respectively.(1)Positioning system based on the wireless fingerprint matching mode,designed a new fingerprint structured,we put forward improved positioning matching algorithm,and design of four different localization algorithm for comparison,finally on the intelligent terminal to realize the system,and evaluate the system performance through experiments,the average localization accuracy is 1.66 m.(2)Aiming at the problem that locatemode based on fingerprint susceptible to environmental change and the influence of the multipath effect,the paper design positioning system that completed based on the deep learning of multi-source heterogeneous information fusion including Wi-Fi,Beacons,geomagnetic information,increase the input feature dimensions,using greed type training methods,we obtain network parameters to construct fingerprint database and realized the high precision,strong robustness indoor absolute positioning system,and assess its performance through experiments,average positioning accuracy is 0.52 m,the maximum position error is1.53 m,92.3% of the error within 1m. |
PDF Full Text Request