Research On Integrity Monitoring Method For Satellite Based Train Positioning Under Non-Gaussian Noise Conditions

Global satellite navigation system has become a key part of people's travel life,and is also an important link to ensure the safety of railway operation.At present,the precision of satellite navigation and positioning has reached the requirement of providing train running information more accurately most of the time.However,if the navigation system fails during train running,there will be serious potential safety hazards if the warning information cannot be provided to the train in time.Therefore,it is of great significance to study the integrity of the ability to provide warnings to users in case of failure.Since the existing receiver autonomous integrity monitoring algorithms all process the integrity of linear Gaussian systems,it is difficult to accurately describe the distribution of observed noise in real operating environment with a single deterministic distribution.In this paper,a Gaussian mixture particle filter RAIM algorithm is proposed.Gaussian mixture model is introduced.The expected maximum algorithm is used to approximate the observation noise of the satellite from unknown noise to Gaussian sum.Particle filter method is used to carry out method research and verification on the algorithm of autonomous integrity of satellite positioning receiver under non-Gaussian conditions.Test statistics are established to identify the time of satellite failure and isolate the failed satellites,thus providing an important guarantee for the safety of train operation in complex environment.The main research contents of the thesis include:(1)For complex non-Gaussian noise,combined with Gaussian mixture model,Gaussian model decomposition of unknown noise is realized by using expected maximum algorithm,probability density decomposition results under different Gaussian model decomposition numbers are evaluated,and the best decomposition number and Gaussian decomposition model are selected.(2)A Gaussian mixture particle filter RAIM algorithm is proposed.The particle filter is suitable for non-Gaussian noise,and the improved Markov Monte Carlo method is used to identify and isolate the failed satellites.(3)Design and develop the satellite positioning observation statistical characteristic modeling tool and Gaussian/non-Gaussian noise integrity monitoring software,select the real train operation scene,complete the satellite fault detection and diagnosis under Gaussian/non-Gaussian noise conditions,and verify the particle filter RAIM method of Gaussian mixture model proposed in the paper.The Gaussian mixture particle filter RAIM algorithm proposed in this paper is used as a non-Gaussian noise-oriented integrity monitoring method to carry out simulation experiments in different scenes.Compared with the conventional RAIM algorithm based on the least square method,the results obtained in fault scenes injected with different types and pseudorange deviations prove that the proposed Gaussian mixture particle filter RAIM algorithm has the capability of satellite fault identification and fault isolation under non-Gaussian noise conditions.The results obtained in this paper can effectively improve the adaptability of the positioning system to complex and uncertain operation and observation environment.Even in the face of non-Gaussian noise conditions,it has the ability of fault identification and fault isolation,thus providing further guarantee for the satellite positioning performance of trains and even the safety of train operation.Figure 80,Table 6,references 35.