| With the development of urban modernization,the demand for high-precision positioning services in logistics,catering,management and other industries is becoming more and more urgent.Nowadays,outdoor positioning technology based on GNSS signals has become more and more mature.How to achieve low cost while obtaining high precision indoor positioning results has become a research focus.Due to the influence of building materials,the distribution of geomagnetic field in indoor environment is different,which can realize indoor positioning.Pedestrian dead reckoning only needs the motion parameters of pedestrians to realize indoor positioning.The above methods do not need additional physical facilities to support,and have a wide coverage,strong continuity,and considerable development prospects.Therefore,this paper mainly focuses on the optimization of single geomagnetic positioning,the realization of PDR positioning and multi-source data fusion positioning,and verified by different experiments.Its main research contents are as follows:(1)According to the low resolution for building a traditional magnetic modulus value,matching algorithm is simple,the issues that led to the positioning error is bigger offline phase by using the method of coordinate transformation processing three axial magnetic data,will replace the traditional magnetic modulus value as the characteristic values of the different spatial interpolation methods and experimental comparison,select the radial basis kernel function to fingerprint interpolation processing;In the online stage,support vector machine was used for regression prediction,and multi-population genetic algorithm was introduced on the basis of traditional genetic algorithm to optimize the weight and threshold of support vector machine,and the MPGA-SVM positioning model was constructed.Experimental results show that the proposed algorithm improves the positioning accuracy by 32.3% compared with GA-SVM,and has higher positioning accuracy.(2)There are still some points in a single magnetic positioning error matching probability is bigger,the indoor light intensity as one of the characteristics and the coordinate transformation of geomagnetic three axis as the input data,spatial interpolation method with different resolution interpolation processing,build the double resolution fusion fingerprint of geomagnetic/light intensity,and USES the BP neural network as a fusion model,The Tent-ASO-BP geo-magnetic/light intensity fusion localization model is established based on the improved atom search algorithm based on Tent chaotic map,and the weight and threshold parameters of BP neural network are optimized by its optimality.Different experiments are set up to verify the localization performance of the fusion model.Compared with the single geomagnetic positioning model,the geomagnetic/light intensity fusion positioning model constructed in this paper improves the positioning accuracy by 34.5% and 32.9%,respectively,and the positioning standard deviation is significantly smaller than the single geomagnetic positioning model,which has better positioning performance.(3)There will be a large error in the location of geomagnetic/light intensity at a certain point.If the track is located in a long and narrow place such as corridor or roadway,this factor will lead to the track of the line of the location point "passing through the wall".To solve this problem,this paper introduced PDR orientation and put forward a kind of indoor positioning method,double information fusion will have obtained the geomagnetic/light intensity positioning coordinates,as a result the first layer data fusion,then in the same experiment on the path for PDR positioning results,will be two ways to get the positioning results of common input to the second Tent-ASO-BP fusion positioning model,Finally,the geomagnetic/light intensity /PDR fusion positioning coordinates are obtained.In the experiments of the two scenarios,the accuracy of the geomagnetic/light intensity /PDR fusion positioning method proposed in this paper is within 1m,which can effectively suppress the occurrence of "positioning back jump" and "trajectory through the wall" phenomenon in trajectory positioning.trajectory positioning.Figure 48 Table 8 Reference 94... |
PDF Full Text Request