| In the information age,people’s demand for location-based services is also increasing.In outdoor positioning,GPS(Global Positioning System,GPS)technology is widely used and gradually becomes a necessity in people’s life.But in the indoor environment,the effect of GPS positioning is not ideal,so many indoor positioning technologies have been developed.UWB(Ultra Wide Band,UWB)technology has high precision and stable service in the indoor environment,but the indoor environment is complex,and in the case of multipath interference and non-line-of-sight,there are still errors in the use of UWB positioning technology.In order to reduce the error To the minimum,researchers have been conducting optimization research on related algorithms.Aiming at the problem that UWB indoor positioning will be affected by a large amount of noise interference and non-line-of-sight errors,which will affect the positioning accuracy,this thesis uses an incremental extended Kalman filter(Incremental Extended Kalman Filter,IEKF)algorithm,a firefly algorithm(Firefly Algorithm),FA),RBF(Radial Basis Function,RBF)neural network algorithm combined with a new algorithm to reduce the positioning error.The main work contents of this thesis are:1.The traditional Kalman filter algorithm is studied,and its application method in TDOA(Time Difference Of Arrival,TDOA)positioning Chan algorithm is introduced.Introduce and introduce the incremental extended Kalman filter algorithm,use it to denoise the TDOA measurement value to reduce the error caused by the environment,equipment and other factors,and then use the Chan algorithm to solve the coordinate value.Through simulation,the incremental extended Kalman filter is compared with the traditional Kalman filter for positioning effect.2.The traditional RBF neural network is improved by introducing a firefly algorithm to find the optimal parameters and inter layer connection weights of the RBF network during its training process,which speeds up the convergence speed of the RBF neural network training.In the process of network training,the influence of the number of hidden layer neurons of RBF neural networks on their network performance was explored and tested.When FA-RBF neural network is used for TDOA positioning,its positioning accuracy is also improved compared to traditional RBF neural network.3.Use the sample data processed by the incremental extended Kalman filter to train the FA-RBF neural network.4.The indoor positioning simulation of the proposed IEKF-FA-RBF neural network algorithm is carried out,and the effects are compared and analyzed with the Chan-Taylor algorithm,Chan-IEKF algorithm,and IEKF-RBF neural network algorithm.With the assistance of the extended incremental Kalman filter and the firefly algorithm,the RBF neural network has achieved good positioning results,and can maintain stable accuracy in non-line-of-sight scenarios.It can be seen from the simulation results that the proposed joint algorithm can effectively suppress the error and achieve good results,and the positioning accuracy can be controlled within 20 cm. |
