高心墙堆石坝原生集成式智能无人碾压系统研发及应用  

Development and application of native-integrated intelligent unmanned compaction system for high core rockfill dam

作　　者：王佳俊[1] 祁宁春 钟登华[1] 佟大威[1] 岳攀 崔博[1] WANG Jiajun;QI Ningchun;ZHONG Denghua;TONG Dawei;YUE Pan;CUI Bo(State key laboratory of civil engineering simulation and safety,Tianjin University,Tianjin 300072,China;Yalong River Hydropower Development Co.,LTD,Chengdu 610051,China)

机构地区：[1]天津大学水利工程仿真与安全国家重点实验室,天津300350 [2]雅砻江流域水电开发有限公司,四川成都610051

出　　处：《水利学报》2022年第12期1421-1432,共12页Journal of Hydraulic Engineering

基　　金：国家自然科学基金雅砻江联合基金项目(U1965207);国家自然科学基金青年科学基金项目(52009089)。

摘　　要：无人碾压技术是大坝填筑施工研究的前沿,有助于提高堆石坝施工质量和效率。然而,目前缺乏对多料种复杂施工环境条件下无人碾压系统的研究,且现有的无人碾压多基于RTK-GNSS和毫米波雷达对位姿与障碍物进行感知,缺乏对压实质量、环境障碍物和位姿的多模态感知;现有的无人碾压多采用外挂执行机构的方式进行改装,设备耐久性和稳定性难以得到保证。针对上述不足,采用感知设备、执行设备和控制设备统一设计和集成装配的方式,自主研发了集成多模态智能感知、智能规划决策和智能控制等功能的无驾驶舱原生集成式智能无人碾压系统。其中,智能感知采用RTK-GNSS、毫米波雷达和机器视觉等多种模态方法实现对车体位姿和环境障碍物的精准感知;基于能量守恒原则,采用加速度计等实现对有效压实功率的精准感知;智能规划决策集成了事前的全覆盖路径规划与事中的碾压参数动态优化等算法;智能控制中通过蝴蝶优化算法动态调控比例-积分-微分算法的控制参数,实现了复杂条件下循迹的精准控制;此外,基于B S架构研发了原生集成式智能无人碾压数字孪生平台,实现了碾压施工过程中物理空间与孪生空间的双向交互。研发的原生集成式智能无人碾压系统在两河口水电站堆石区、过渡区和砾石土心墙区等复杂施工条件下开展了应用研究,结果表明,相较于传统碾压施工方法,该系统一次碾压质量达标率提升2.42%,效率提升15.20%。Unmanned compaction is the forefront of dam filling construction research,which helps to improve the construction quality and efficiency of rockfill dams.However,there is currently a lack of research on unmanned rolling systems under complex construction conditions with multi-type materials,and the existing unmanned rolling systems are mostly based on RTK-GNSS and millimeter-wave radar to perceive poses and obstacles,and lack the multi-modal perception of quality,environmental obstacles and posture.The existing unmanned rolling is mostly modified by external actuators,and the durability and stability cannot be guaranteed.In view of the above deficiencies,a unified design and assembly of sensing equipment,executive equipment and control equipment in the factory was adopted to develop a cockpit-free native integrated intelligent unmanned rolling system independently,that integrates functions such as multi-modal intelligent perception,intelligent planning decision-making and intelligent control.Among them,various modal methods such as RTK-GNSS,millimeter-wave radar and machine vision are used to realize the accurate perception of pose and environmental obstacles.While the effective compaction power is obtained by accelerometer based on conservation of energy principle,intelligent planning and decision-making integrate the coverage path planning and dynamic optimization of rolling parameters.In intelligent control,the control parameters of proportional-integral-derivative algorithm are dynamically adjusted by butterfly optimization algorithm,which realizes precise control of tracking under complex conditions.In addition,based on B S architecture,a native integrated intelligent unmanned rolling digital twin platform was developed to realize the two-way interaction between physical space and twin space during rolling construction.The native integrated intelligent unmanned compaction system has been applied in complex construction conditions such as the rockfill area,transition area and gravel-soil core wall area of Lia

关 键 词：高心墙堆石坝 原生集成式智能无人碾压 多模态智能感知 多料种 有效压实功率 智能规划决策 

分 类 号：TV641[水利工程—水利水电工程]