| Most areas of people’s life rely on Global Navigation Satelite Systems(GNSS)to provide position services.In the case that the satellite position signal is blocked or shielded,indoor location feature information such as WIFI and Bluetooth can be used to solve position services.Indoor pedestrian position technology based on smart phones is becoming increasingly important in smart cities,smart car research and medical products development.Aiming at limitations of algorithm redundancy,difficulty in generalization and accuracy of various indoor location,a seamless navigation algorithm integrating indoor pedestrians method was proposed.The following research contents are described in detail:Firstly,an adaptive wavelet denoising algorithm based on optimal threshold function is proposed to deal with weak non-stationary characteristics of inertial sensor signal.In this algorithm,the adaptive threshold calculation method is adopted to calculate the adjusting factors of the control threshold through the mean and standard deviation,and sample covariance and sample variance of original data and the data decomposed by wavelet transform are used to adjust the high and low threshold,so as to realize preprocess of subsequent micro-inertial device data.The threshold function which is one of the most important parameters for affecting signal denoising effect of micro-inertial sensor,is improved effectively,and power series and exponential function are added to increase the smoothness and denois of the threshold function.Secondly,aiming at the impact of random jitter and false walk state of indoor pedestrian navigation of step counting algorithm,a pedestrian step counting algorithm based on adaptive dynamic variance threshold constraint is proposed.The algorithm estimates the weighted step size by adding weight factors and compensates the step size by adjusting model parameters.According to the pedestrian motion state has state continuity,a number of adaptive dynamic threshold judgment conditions are added to step counting wave peaks.Based on the amplitude variation of acceleration variance value,the dynamic variance threshold comparison of acceleration variance value at N+1 point was conducted,and the pedestrian motion state and step size were effectively corrected.Finally,aiming at the uncertainty of measurement noise in multi-source fusion indoor pedestrian heading estimation,a pedestrian indoor navigation algorithm based on optimized step size and head estimation is proposed.In this algorithm,fad memory factor and limited memory weighting factor are added to weaken the noise covariance estimation,and the noise parameters measured by the covariance matrix are estimated and corrected to improve the weight of the latest data.The course angle is corrected by exponential weighting adjustment of fixed length historical data.In addition,the fuzzy adaptive algorithm is used to adjust and control the noise parameters,and the covariance matrix and adjusting factor are defined to adjust the covariance matrix parameters,so as to improve the accuracy of indoor pedestrian navigation algorithm.Simulation results show that the optimal threshold function proposed in this thesis has good continuity,and the useful signal deviation is small after denoising,and the signal becomes smooth after the adaptive threshold calculation.At the same time,the pedestrian step counting algorithm proposed in this thesis can effectively identify the spurious wave peaks caused by slight jitter and false walk.The accuracy of step size fitting can reach95.78%,and the accuracy of step number detection can reach 92.62%.Compared with CKF and ACKF,it is proved that the fusing fuzzy inference can increase the adaptability and robustness of ACKF algorithm,and can effectively correct model noise parameters.In addition,the indoor navigation algorithm for pedestrians in this thesis is compared with the traditional algorithm,and the navigation algorithm proposed in this thesis has less error,and the indoor navigation algorithm is more optimized. |
PDF Full Text Request