轨道沉降的连续运动测量理论与方法研究

Railway track settlement monitoring urgently needs fast measurement theory and technology, however, fast continuous moving measurement method of track settlement is unavailable utilizing existing measurement theory because the measurement of absolute track settlement by moving measurement method requires the measurement value should be based on the static geodetic coordinate system (GCS). If the analytical relationship of differences and uncertainties could be established between the dynamic coordinate based-on the measurement system and the static GCS, a breakthrough would be made in moving measurement theory and technology of track settlement.. .This project, therefore, aims at proposing a new moving measurement method to answer the question how the geodetic measurement value would be obtained from the moving measurement value. Hopefully, the project would create a model for innovations in moving measurement theory, method and technology of track settlement. The concrete method is to introduce computer vision parameters into inertial measurement signals by means of a motion sensing calibration, to let computer vision obtain the datum-point information and shifting the multi-sensor measurement values below the CVS coordinate, to establish the analytical relationship between the dynamic coordinate and the GCS, and thus to have the motion trajectory of the vehicle in the GCS. ..The project involves the following researches: (1) to construct an moving measurement model for track settlement; (2) to model the high precision calibration of the multi-sensor unit in the measurement system and the mapping relationship between coordinates; (3) to obtain the motion trajectory of vehicles and revise the theory of errors in motion; and (4) to design multi-variable spatial variable filters to overcome distorted signals caused by non-constant moving speed of the measurement system and by signal mapping from time domain to spatial domain. If this research project is supported by NSFC, new academic achivements would be obtained, proving a new solution of measuring railway track settlement parameters in continuous moving condition.

轨道沉降的监测急需高精度快速测量理论和技术，现有理论方法难以实现快速连续运动测量，原因是要实现运动条件下轨道沉降的绝对测量，其测量值必须是以建立在大地坐标系下为前提。如果能将测量系统自身的动态坐标系与基准的大地坐标系之间存在的差异和不确定性形成明确的映射关系，轨道沉降的运动测量理论和技术将获得突破。为此，本项目提出在惯性信息中引入机器视觉参数，建立运动感知联合定标模型，将多传感测量值映射到视觉坐标系下，由视觉单元获取轨道测量基准点信息，建立动态坐标系与大地坐标系的解析关系，获得建立在大地坐标系下的载体运动轨迹，解决从运动测量值恢复轨道大地测量值的关键理论问题。研究包括：轨道空间曲线运动测量模型的构建；多传感单元的高精度定标及其坐标映射建模；载体运动轨迹获取及误差修正理论；克服运动速度变化及信号时空变换影响的空间移变滤波方法。研究成果有望解决现有轨道沉降无法连续测量的瓶颈问题。

轨道沉降的监测急需高精度快速测量理论和技术，现有理论方法难以实现快速连续运动测量。本项目提出了一种多传感融合的轨道空间线形运动测量方法，从而得以解决轨道沉降参数的测量。项目研究建立了集机器视觉、载体姿态信息融合的运动测量模型，并具体研究了以下关键理论问题： . （1）建立了基于机器视觉和惯性单元的轨道沉降运动测量模型。在算法上，研究基于扩展卡尔曼滤波的高精度载体空间动态运动曲线解算算法。实验结果表明：该方法测量的位移曲线与运动平台控制设计的曲线之间保持一致，误差在0.2mm以内，测量计算的空间曲线准确反映了实际平台的运动轨迹。（2）研究了运动过程中的动态定标方法。提出两步法计算惯性单元及视觉传感器之间的旋转和平移关系，通过建立空间垂直向量、平移向量和扩展卡尔曼滤波法动态计算和优化它们之间的关系。研究表明：随着测量次数的增加惯性坐标系相对于视觉传感器坐标系各轴欧拉角的误差逐渐下降，该方法可实现惯性及视觉传感器之间的空间相对位姿精确定标。（3）基于谱采样的多变量移变滤波器法实现惯性单元信号的空间幅频特性补偿。具体研究基于时间、空间、频响特性多变量控制的移变滤波方法，实现惯性信号的高精度动态补偿，使得测量系统具有强鲁棒性。经以上理论和技术的研究，现场试验表明：测量的轨道空间线形与现场线形相一致，具有良好的可重复性，误差在0.5mm以内。该基于多传感信息融合的轨道空间运动测量模型具有良好可工程应用性。. 通过与理论研究结果对比分析，本项目在实验室和现场条件下实现了轨道沉降中轨道空间线形基础参数的精确、可靠运动测量，基本解决了运动状态下轨道沉降的测量问题。为工程应用提供了科学的方法与手段，具有重要理论意义与实用价值。