Research On Test Method Of Train Positioning Unit With Multi-sensors

In new train control systems,there has been a trend to use multi-source fusion train positioning units that do not rely on traditional ground-based positioning equipment.Multi-source fusion positioning units use a combination of satellite positioning,inertial navigation and wheel speed sensors to achieve positioning with all-weather and multigeographic capability.To ensure that the developed train positioning unit meets the design requirements in a real operating environment,it needs to be fully tested to verify the system functionality and performance.This thesis study a test method of train positioning unit based on field environment data,starting from the positioning function of the unit,analyzing the causal events leading to the positioning failure of the unit,designing corresponding test cases,analyzing the environmental data along the railroad line,constructing test sequences according to the distribution of typical scenes along the line,and evaluating the test results in terms of positioning accuracy and positioning failure probability to test the train positioning unit.The main work of this thesis is as follows:(1)The UML model and the fault tree model are used to analyze the failure of the positioning function of the train locating unit.Based on the results of the fault tree analysis and the equivalence class division method,the causal events leading to the failure of the positioning function of the train locating unit are extracted,and test cases are constructed based on them.(2)Combined with the satellite positioning principle,the typical scenes along the railroad line and satellite visibility are analyzed,the railroad line environment is parameterized,the SVM-based segmentation method of typical scenes along the railroad line is studied,and the line test segments are reasonably divided according to the distribution of the typical scene segments.Combined with the testing requirements of the positioning function of the train positioning unit,the test condition vector and the test task vector are designed,and a neural network-based test sequence generation method is studied to predict the test cases that can be executed according to the test conditions available in the zones,and the test sequences that satisfy the line environment conditions are obtained in series to achieve a close match between the test sequences and the test conditions.(3)Design the evaluation method of train positioning unit,adopt the error ellipse and reference system-based positioning accuracy evaluation method to evaluate the static positioning accuracy and dynamic positioning accuracy respectively;for the situation that the failure probability of some events of train positioning unit is unknown and the failure state is polymorphic,a positioning failure probability analysis method based on fuzzy Bayesian network is given to realize the positioning unit failure probability calculation.(4)The proposed method is validated by field testing using actual test data of the Hamu line as an example.The experiment proves that the proposed method of segmentation of typical scenes along the railroad line can effectively distinguish the typical scenes between Chaidar and Huanqiang;the test sequences generated according to the environmental scenes meet the test conditions of each segment,and the test sequences are arranged reasonably by manual verification;the positioning accuracy of the train positioning unit meets the actual requirements,and the probability of failure of the positioning is lower than that of the train positioning unit using only satellite positioning technology.Meet the design requirementsFigure 67,Table 29,Reference 58...