Research On Locating Method Based On IMM-EKF Combined Train

Railway transport production is the major artery of the national economy and it is the link connecting production,distribution,exchange,and consumption of the society.It undertakes the turnover of passengers and the transportation of goods throughout the country,the railway is an important means of transportation to enrich the people's material life,raising the people's cultural level,meetting people's travel needs,and strengthen national defense construction.Accurate train positioning algorithm is a key factor to ensure the safety of railway transportation,and it protects the growing demand of the national economy and people's lives.With the continuous development of railway transportation,the variability of train movement,and the change of environment,we need to constantly improve the target train movement model,improving the self-adaptability of train positioning algorithm has a certain significance in practical applications.When using the fixed-structure model alone to estimate the state of train positioning,it fails to match the actual train movement model,due to the limitations of the algorithm itself,the accuracy of the filtering is not high,this thesis presents an improved IMM-EKF train positioning method based on Beidou satellite navigation system.This thesis proposes a system based on the Beidou satellite navigation system,IMM-EKF improves the train positioning method.This thesis studies the train positioning method in three aspects of train motion state recognition,IMM-EKF Algorithm,IMM-EKF positioning data processing based on curve fitting,verified and analyzed the improved algorithm.Firstly,through the study of the dynamic model of the train passing through the track ballast,the relationship between the angular velocity of the high speed train passing through the track ramp and the motion transition(train straight,right turn and left turn)is analyzed and summarized;According to the characteristics of the three basic curve characteristics of the train motion state recognition results,combined with the fuzzy logic if-then rules to determine the noise parameters in the IMM algorithm,it lays a theoretical foundation for improving the parameter adaptation in the IMM-EKF combined positioning algorithm model;Then,for the nonlinear estimation of the state equation,this thesis using the EKF(Extended Kalman Filter)algorithm to improve the IMM algorithm,using the improved IMM-EKF algorithm model to estimate the train's operating position.Finally,using the machine learning curve fitting method to process the IMM-EKF positioning data.The route of the train is closer to the railway line.After the field measured data preprocessing,first to validate the improved IMM-EKF algorithm,because the real-time train state recognition results were introduced to adjust the model parameters,the experimental results show that the overall trajectory and turning local error accuracy are smaller than the uncorrected algorithm,it shows that the improved model of this algorithm is reasonable and effective,guarantees the positioning accuracy,and has good self-adaptive positioning in the actual environment.Secondly,using the method of improved machine learning curve fitting to fit the train route,finally compared with the IMM-EKF improved filter train positioning method and actual railway line,experimental results show that improved machine learning curve fitting method can reduce positioning data error.Accurately reflects the detailed characteristics of the train's trajectory and closer to the actual railway line.Provide support and guarantee for efficient and high-performance train positioning and train operation control.