Research On Train Location Algorithm Based On GNSS And Image Feature Recognition

The continuous development of high speed railway makes the train positioning system face more challenges.In recent years,with the global navigation satellite system(Global Navigation Satellite System,GNSS)and the continuous development of the train location using the GNSS technology to the traditional,such as track circuit and odometer are optimized,which can improve the positioning accuracy of the train,to change the traditional interval length unit block mode,reduce the use of ground equipment.Due to the complex running environment of the train,some special characters is contained after GNSS is introduced into the field of train positioning.For example,the distance of train running is a relative position distance from 0 kilometers in per line,not the latitude and longitude information in the traditional earth coordinate system,the position information is updated more frequently because of the higher speed and the running line is more fixed.In order to realize the independent positioning of the train,this paper proposes a train combination location method based on GNSS and image feature recognition.This paper first introduces the research status of train positioning system based on satellite navigation,compares the advantages and disadvantages of various train location.On the base of the analysis and summarization,a method for multi-sensor data fusion based on GNSS/SINS is proposed.In order to ensure the reliability of data fusion,a new fault-tolerant design for train positioning system based on wavelet and singular value decomposition.The train operation environment is complex,and lead the satellite signal is not available when train travels across such as tunnel,cutting and forest areas,which results in rapidly divergence problems of error of SINS.Combined with the characteristics of train operation and the constraint to the train by the running rails,a UKF algorithm based on the constraint of train movement is put forward to restrain the divergence.Long time running of the positioning system will appear cumulative error inevitably.In order to eliminate the accumulated error,the point correction algorithm combined with the characteristics of train laying trackside equipment based on image feature recognition is put forward,which uses the absolute kilometer information of wayside equipment to correct positioning results.Finally,a combined navigation system based on GNSS/SINS platform is designed,and the applicability of the proposed algorithm is confirmed by vehicle experiments.The results of simulation and vehicle experiment show that the framework of the multi-sensor data fusion system proposed in this paper can effectively improve the precision of the positioning system,reduce the decrease of the train positioning performance caused by the multi source noise in the positioning environment,and have the application value for the improvement of the positioning accuracy of the train,and improve the security of the positioning system.On the other hand,the suppression algorithm for inertial navigation error combined with the train running characteristics can effectively improve the accuracy of the train calculation and solve the problem of positioning by the rapid divergence of the inertial navigation error.Finally,the fixed point correction work of the train position is realized by using the image recognition method.It can provide accurate and absolute position information for the train positioning system.The results show that the combined location algorithm proposed in this paper can provide accurate,continuous and real-time location services for trains and can be applied to actual requirements.