Research On Robust Estimation Theory Based Train Integrated Positioning Method

ABSTRACT: Train positioning is crucial for train control system to generate effective train control strategies and guarantee the safety and efficiency of train operation. Traditional train positioning system with the single-sensor configuration could not meet the developing requirements of modern train control system, while the implementation of multi-sensor based train integrated positioning system has become an inevitable trend. As multi sensors and more complicated architecture are adopted in train integrated positioning system, multi-sensor fusion, which is the core function of the integrated system, should be designed with more consideration to robustness of the filtering and estimation, so that a desired performance level could be realized as well as safety.In this thesis, according to the practical characteristics of train positioning and structure of the integrated system, based on the H∞robust filtering method, the robust estimation theory based train integrated positioning method is studied, and a cubature H∞robust filtering algorithm is proposed. A federation structure based decentralized filtering strategy is presented for cubature H∞filter, and a variable restraint coefficient based adaptive filtering approach is developed. Furthermore, the present study expands the perspective to cover the complete procedure of train integrated positioning, and explores integrity assurance method for the train integrated positioning system.The innovations of the thesis are as follows:(1) The study summarizes the model of filtering estimation based train integrated positioning, proposes a cubature H∞nonlinear robust filtering algorithm, and presents an information distribution coefficient adaptive decentralized filtering solution for the cubature H∞filter, on the basis of a federation structure.(2) The study, through the exploration of the restraint coefficient mechanism and solution for optimal restraint coefficient, presents a variable restraint coefficient based cubature H∞adaptive filtering approach, and determines a feasible dynamic adjustment strategy for the restraint coefficient.(3) The study proposes a novel integrity assurance method for the train integrated positioning system, where an Autonomous Integrity Monitoring and Assurance (AIMA) scheme is established, and specific integrity assurance solutions are formed for the sensor collection stage, sensor fusion stage and map matching stage respectively.(4) The study establishes a train integrated positioning simulation system, where the issues of sensor error models and train trajectory generation are solved, and realizes the validation of theoretical results in this thesis by sensor collaborative simulation under different scenarios.In this thesis, simulations and field experiments are both employed for validation, which demonstrate that the proposed robust estimation theory based train integrated positioning method could reach a favorable balance between different performance indices of train positioning, and is applicable for modern train control systems.