Research on indoor positioning based on deep learning and multi-sensor fusion
In view of the large positioning error of traditional neural network algorithm and the shortcomings of single geomagnetic technology,as well as the shortcomings of other indoor positioning methods,this paper proposes a method based on long short-term memory recurrent neural network and geomagnetic,light intensity and wavelet noise reduction technology.In the off-line stage,the three-axis geomagnetic data converted by coordinate axes and the corresponding light intensity data are combined to improve the feature dimension of the positioning point.The wavelet denoising method is adopted to denoise the data and establish a fingerprint database,which is input into the LSTM neural network model to establish the geomagnetic positioning model.In the online stage,the established geomagnetic positioning model is used to output the positioning results.The experimental results show that the average error of the fusion positioning method is reduced by 17％compared with the average error before noise reduction,and the average error of the fusion positioning method is increased by 34.5％compared with the single geomagnetic positioning technology.The positioning model has good positioning performance and can be effectively applied to indoor positioning.Moreover,the indoor positioning technology of geomagnetic/light intensity/wavelet noise reduction can significantly improve the positioning accuracy and stability,and solve the positioning accuracy problem in complex environments.
NETL
NSTL
万方数据
