Research On Indoor Positioning Technology Based On Geomagnetic Fingerprint Map And PDR

With the rapid development of wireless communication,internet of things and other technologies,location-based services,especially indoor positioning,have received more and more attention.In indoor location,geomagnetic signals can be used as location fingerprints in the established fingerprint database due to time stability and location difference.However,the location resolution of indoor geomagnetic fingerprint is very low,so it is difficult to locate directly.However,the pedestrian dead reckoning(PDR)based on inertial sensor information is accurate in positioning within a short distance but has accumulated errors.In order to improve the positioning accuracy,this paper will explore and study the geomagnetic fingerprint map construction and PDR algorithm as well as their fusion positioning algorithm.The location method based on geomagnetic fingerprint consists of two steps: offline database building and online matching.In offline database construction,in order to collect points under the condition of the same fingerprint,the construction of higher accuracy,this paper proposes a simulated annealing optimization of traditional BP network model of prediction methods: BP network are optimized by the simulated annealing(SA-BP)in the weights and thresholds,avoid the trapped in local optimal solution in the process of training,improve the performance of the model.The results show that the estimation error of fingerprint map constructed by this method is lower than that of ordinary Kriging interpolation method.The PDR localization algorithm based on inertial sensor is divided into three aspects: step frequency detection,step length estimation and direction estimation.In order to improve the positioning accuracy of PDR,the acceleration data is used and the threshold value and the change of state value are set for the step frequency detection.A nonlinear model wasconstructed by introducing step frequency parameters to estimate step size.A PDR localization algorithm for PSO-IPF is proposed.At the same time,in order to improve the accuracy of pattern recognition of the acquisition equipment and pedestrian motion state,a new classifier was constructed by using the width and penalty factor of Gaussian kernel function in PSO-SVM.The experimental results show that the improved PDR algorithm has higher positioning accuracy than the traditional PDR algorithm.In the fusion location algorithm,a forward sequence matching algorithm is proposed,which serializes the geomagnetic information obtained in continuous time by PDR and merges the location data by using hidden Markov model(HMM).Experimental results show that the proposed algorithm has higher positioning accuracy than the traditional hybrid positioning algorithm(Kalman filter,particle filter).Through the actual indoor environment positioning experiment,the positioning algorithm proposed in this paper has superior positioning performance,and its positioning accuracy reaches 0.921 m,which can meet the needs of indoor positioning accuracy.