Algorithm Research Based On The Positioning Of The Beidou Satellite

With the rapid development of railway in China,more and more attention has been paid to the stable operation of trains,among which satellite navigation combined with inertial navigation(INS)has been a hot topic in the international research because of its unique advantages.However,although the positioning accuracy of BDS satellite navigation system is very high and covers a wide area,there will be a blind area in the dense urban buildings or in the dense suburban jungle.In order to better enhance the accuracy and stability of the train positioning of the Beidou satellite positioning/inertial navigation integrated positioning system(BDS/INS),this paper combines the wavelet neural network to study the combined positioning of the train in the Beidou signal,effectiveness and the Beidou signal failure(blind area).The characteristics of Kalman filter(KF),extended Kalman filter(EKF),Untraced Kalman filter(UKF)and particle filter algorithm(PF)are analyzed in the case of the Beidou Positioning and the inertial navigation signal.The problem of constructing the train is nonlinear and unaccurate,and the wavelet neural network is combined with the wavelet neural network.The Kalman wavelet filter is designed based on PSO wavelet neural network.The neural network is combined with the wavelet neural network and the particle swarm optimization,in which the input of the neural network is the signal before the Kalman filter,and the output is the filtered signal.Meanwhile,considering the factors that affect the accuracy of wavelet neural network,we optimize the weights,thresholds and learning rates.On this basis,by adding logarithmic function to improve the learning rate,genetic algorithm(GA)and particle swarm optimization(PSO)combination are used to optimize the weights and thresholds.In order to verify the effectiveness of the PSO wavelet neural network aided kalman filter,the mathematical model is established and several algorithms are compared and simulated.The simulation results show that the optimized wavelet neural network not only improves the convergence speed.but also improves the positioning accuracy effectively,while maintaining the overall trend of the original velocity waveform.Under the premise of the potential.we reduce the signal error caused by filtering and achieve the purpose of experiment.When the Beidou positioning signal is invalid and the inertial navigation is effective,this section first constructs a small wave neural network model by using the position velocity information of the inertial navigation system when the Beidou positioning signal is norrmal,and the position velocity information of the Beidou Positioning and inertial navigation signal fusion is used as the output,and the positioning letter in the Beidou is located.When the number is invalid,the position and velocity information of inertial navigation and the output of wavelet neural network are selected as the input of optimized UKF.Finally,the validity of the train blind location algorithm is verified by the data simulation.Even during the interruption of the Beidou positioning signal,the Beidou inertial navigation combination positioning can still effectively meet the 10m position error precision of the two generation of the Beidou navigation and positioning system.