Research On Indoor Positioning Algorithm Based On Fingerprint And PDR

In recent years,with the development of the Internet of Things,the application of related technologies based on location services has become more and more widespread,and people's demand for positioning is also growing.However,in the indoor environment,the transmission signal is susceptible to multipath effects,personnel walking and other factors,resulting in low indoor positioning accuracy.Aiming at the above problems,this paper proposes an indoor positioning algorithm based on location fingerprint and PDR.The main contents are as follows:Aiming at the problem that the position fingerprinting error is relatively large,an indoor positioning algorithm based on kernel principal component analysis and improved gradient boosting regression tree is proposed.In the offline phase,the kernel principal component analysis algorithm is used to extract the principal components of the original location fingerprint vector,and the training set is uniformly sampled into multiple sub-training sets by improving the gradient boosting regression tree algorithm.Each sub-training set constitutes an indoor positioning regression model.Finally,a strong regression model is formed;in the online phase,the kernel principal component analysis transform is performed on the signal strength of the real-time measurement,and the real-time position is predicted based on the mode of the offline regression model calculation result.The simulation results show that the average positioning error of scene 3 is 0.92 m.For the pedestrian dead reckoning,the initial value cannot be judged and there is a cumulative error.The initial value of the position fingerprint location is set as the initial value of the pedestrian dead reckoning,and the acceleration angle and gyro sensor are used to measure the direction angle and acceleration data of the pedestrian.The number of walking steps and step size are calculated by nonlinear model and correlation analysis to determine the position of the pedestrian.The simulation results show that the average positioning error of scene 3 is 0.63 m.The above two algorithms still have the problem of low positioning accuracy.An indoor positioning algorithm based on unscented Kalman filter fusion location fingerprint and pedestrian dead reckoning is proposed.The first position fingerprint positioning result is set as the initial value of the unscented Kalman filter start value and the pedestrian dead reckoning.When the pedestrian moves,the pedestrian dead reckoning and the fingerprint algorithm obtain the pedestrian observation position and the measurement position,and the unscented Kalman filter is used to combine the observation position and the measurement position to estimate the pedestrian's motion trajectory.The experimental results show that the average positioning error of the fusion algorithm reaches 0.43 m,which meets the requirements of indoor positioning accuracy.The paper has 36 diagrams,21 tables,and 63 references.