基于管道机器人惯导/多里程计融合的堆石坝内部变形测量方法研究

The internal deformations of rockfill dam are important parameters that reflect its safety status. Traditional internal deformation monitoring methods have the disadvantages of point monitoring and fragile sensor structure. It is urgent to develop accurate, fine and reliable internal deformation monitoring technology. At present, the internal deformation distributed monitoring technology of rockfill dam based on pipeline measurement is difficult to meet the millimeter-level measurement accuracy requirements in related standards. We proposed a method to measures the internal deformation of the rockfill dam by integrating the INS/multi-odometer on axis-centered pipeline robot. This research focuses on the following aspects 1) High-precision INS/odometer calibration method by fixed-axis rotary motion; 2) Three-dimensional curve estimation method based on tight and robust fusion of INS/multi-odometer data under factor graph optimization framework; 3) Estimation method of internal deformation index and errors of rockfill dam based on registered three-dimensional curves. Through the breakthrough of the above key technologies, the pipeline robot system error will be compensated, and the multi-class constraint information in multiple repeated measurements is fully exploited to improve the fusion precision of the INS/multi-odometer and achieve high-precision distribution deformation measurement under the limitation of small size civil INS. This research has theoretical significance for the development of INS/odometer data fusion methods for measurement applications and will provide theoretical and technical support for the internal deformation monitoring of rockfill dams and other large civil structures.

堆石坝内部变形是反应大坝安全状态的重要参数。传统内部变形监测方式存在着点式测量、传感器死亡率高的缺陷，亟需精细、可靠的内部变形监测技术。当前基于管道测量的堆石坝内部变形分布式监测技术在百米以上范围难以满足规范毫米级测量精度要求。本项目通过集成惯导/多里程计的轴对中管道机器人对堆石坝内部变形进行测量，研究1）基于定轴旋转运动的管道机器人惯导／多里程计快速高精度标定方法；2）基于因子图惯导/多里程计紧密鲁棒融合的三维曲线最优估计方法；3）基于配准三维曲线的堆石坝内部变形指标计算模型及误差特点。通过对上述关键技术的突破，补偿管道机器人系统误差，充分挖掘多次重复测量中多类约束信息，提高惯导/多里程计融合精度，实现小型民用惯导性能受限条件下的高精度分布式变形测量。本研究对于发展适用于测量应用的惯导/里程计融合方法具有理论意义，为堆石坝等大型土木结构体内部变形监测提供理论和技术支持。

本项目面向堆石坝内部变形监测中对高精度、连续、长期可靠测量技术的行业应用需求，围绕基于惯导/多里程计的高精度管道测量的堆石坝内部变形监测技术的关键难题开展研究。项目团队设计了高精度管道测量机器人系统，集成高精度激光惯导、多路里程编码器和微小变形补偿装置，实现小口径管道精密测量；发明了利用定轴旋转运动对管道机器人惯导/里程计偏置角误差进行快速高精度标定方法，消除系统误差；提出顾及多类型空间约束的惯导/多里程计数据EKF紧密鲁棒融合方法，提高了基于惯导／多里程计的管道机器人的曲线推算精度和可靠性；在此基础上，构建了基于多期管道线形配准的堆石坝内部变形指标计算模型，消除里程误差影响，实现垂直沉降和水平位移的计算。项目成果在四川两河口心墙堆石坝和贵州夹岩面板堆石坝等大坝中得到成功应用。通过与传统方法监测结果对比，验证了其测量精度，实现了内部变形百米范围毫米级空间连续测量。成果为高堆石坝内部监测提供创新的理论和技术支撑，有效提升了我国堆石坝安全监测的技术水平。核心成果惯性线形测量技术进一步拓展应用到2022年北京冬奥会国家速滑馆冰面基底平整度测量、南沙港铁路大桥轨道线形测量，具有良好的推广应用价值。