首页 | 本学科首页   官方微博 | 高级检索  
     

基于CA-RL的堆石坝填筑仓面碾压作业动态路径规划
引用本文:崔博,钟航,王佳俊,谭添文,林威伟,佟大威. 基于CA-RL的堆石坝填筑仓面碾压作业动态路径规划[J]. 水利学报, 2024, 55(3): 253-265
作者姓名:崔博  钟航  王佳俊  谭添文  林威伟  佟大威
作者单位:天津大学 水利工程智能建设与运维全国重点实验室, 天津 300350
摘    要:碾压作业是堆石坝填筑仓面施工的关键工序,科学合理规划碾压路径能够在保障碾压质量的前提下提高碾压效率。在当前路径规划研究基础上进一步考虑碾压机数量变化、压实质量感知等动态要素可以提高路径规划模型面对复杂动态的仓面施工环境的适应性。本文通过建立基于元胞自动机的填筑仓面信息模型、提出条带整体压实质量评价方法,解决压实质量等仓面信息的储存更新的问题;通过建立基于强化学习的碾压作业路径规划模型、构建碾压机状态集和动作集、设计奖励函数和探索利用策略,解决碾压机数量变化的路径分配的问题;耦合上述两种模型,实现堆石坝填筑仓面碾压作业动态路径规划。结合国内某堆石坝工程开展了工程应用,结果表明:本方法可动态考虑碾压机数量变化、压实质量感知等要素,路径规划结果在保障碾压质量的前提下路径长度相较现场施工实际路径平均缩短22.3%,能够有效提升碾压效率。

关 键 词:堆石坝填筑  仓面碾压作业  动态路径规划  元胞自动机  强化学习
收稿时间:2023-05-31

Dynamic path planning of rockfill dam warehouse surface rolling operation based on CA-RL
CUI Bo,ZHONG Hang,WANG Jiajun,TAN Tianwen,LIN Weiwei,TONG Dawei. Dynamic path planning of rockfill dam warehouse surface rolling operation based on CA-RL[J]. Journal of Hydraulic Engineering, 2024, 55(3): 253-265
Authors:CUI Bo  ZHONG Hang  WANG Jiajun  TAN Tianwen  LIN Weiwei  TONG Dawei
Affiliation:State Key Laboratory of Hydraulic Engineering Intelligent Construction and Operation, Tianjin University, Tianjin 300350, China
Abstract:Rolling operation is essential to the construction of rockfill dam.Scientific and reasonable planning of the rolling path of warehouse surface can improve the rolling efficiency and maintain the rolling quality.Current research on roller path planning during storehouse surfaces compaction lacks in-depth consideration of dynamic factors such as changes in the number of rolling machines and real-time analysis of compaction quality.Regarding this situation,this paper proposes a dynamic path planning method for rolling machine groups based on reinforcement learning-instructed cellular automata model instructed.First,a cellular automata-based rolling surface information model is established,and a method for evaluating the overall compaction quality of strips is proposed to store and update the compaction quality and other warehouse surface information.Then,a path planning model based on reinforcement learning is established,the state set and action set are constructed,the reward function is designed and the utilization strategy is explored to solve the path assignment problem as the number of rolling machines changes.Coupled with the above two models,the dynamic path planning of roller groups in rockfill dam construction is realized.The engineering application shows that the proposed method can dynamically consider changing factors such as number of rollers and perception of compaction quality.The planned path reduces in length by 22.3% on average compared with on-site construction while maintaining high compaction quality.The proposed method can significantly improve the rolling efficiency.
Keywords:rockfill dam construction  warehouse surface rolling operation  dynamic path planning  cellular automata  reinforcement learning
点击此处可从《水利学报》浏览原始摘要信息
点击此处可从《水利学报》下载免费的PDF全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号