Research On Algorithm Of Extracting Curve Feature Points In High Speed Railway

The high regularity of high speed railway is the basic guarantee for train running safely and steady. In order to ensure the safety and stability of train, the static geometric linear is detected to accept and evaluat the construction quality of track, which is used as the foundation of maintaining the high regularity. The primary task of evaluating the railway track judges the segments to a nicety in the checking data of track, and extracting the feature points of track curve. The precision of the feature points has a direct impact on the accuracy of evaluation for the static geometry status of track. All of the existing track linear segmentation algorithms are proposed for ballasted track, and have the shortcomings of poor precision and low efficiency. At the same time, due to the different detection methods, the quantity and density of checking data for ballastlness track are dozens times for ballasted track, therefore existing linear segmentation algorithms are often difficult to work at this time. In addition, due to construction errors factors lead to deviation in location between the grade change point of super-elevation and feature points of horizontal curve, which will have disadvantage influence on the irregularity of track.In order to solve the above problems, according to the analysis, which is based on measured data, found that the variation tendency of super-elevation and slope with the mileage are better for reflecting the variation of track geometric linear than other features, such as versine and curvature. According to the variation tendency, which is a continuous and smooth digital broken line, a new algorithm for extracting the turning points of digital broken line, which is named Intersection Method, is proposed in this thesis. According to the super-elevation data and slope data, the precise grade change point of super-elevation is extracted, and the approximate segments for horizontal and vertical linear is also realized. Experiment results show that the calculation model of new algorithm has high accuracy. The linear segments, which are based on super-elevation data and slope data, are with small difference between Intersection Method and the design value. Its result can be used as the initial value of iteration calculation for precisely segmenting the linear of track.Secondly, the robust orthogonal least squares method is used to fit accurately the horizontal and vertical linear. According to the mathematical model of track geometry linear parameters and mileage of segmenting points, the horizontal linear and vertical linear are precisely segmented by iterative calculation. Experiment results show that robust orthogonal least squares linear fitting method can effectively eliminate the adverse effect of the gross errors in observations, and guarantee the accuracy of fitting results. The feature points of track curve can be gotten by iterative calculation with high precision.Finally, the allowable interval between the grade change point of super-elevation and feature point of horizontal curve is approximated. Some corresponding evaluation index which satisfies the requirement of accurate evaluation for track irregularity, are put forward. These work may have a certain reference value to evaluate track construction quality.