基于模型预测控制的多钢筋并行排布智能深化设计方法  

Intelligent Detailed Design Method of Parallel Layouts of Multiple Rebars Based on Model Predictive Control

作　　者：齐宏拓 伍洲 周绪红[3] 李盛 许成然 郑晓磊 刘界鹏[3] 廖耘竹 QI Hongtuo;WU Zhou;ZHOU Xuhong;LI Sheng;XU Chengran;ZHENG Xiaolei;LIU Jiepeng;LIAO Yunzhu(Institute for Smart City of Chongqing University in Liyang,Liyang 213300,Jiangsu,China;School of Automation,Chongqing University,Chongqing 400044,China;School of Civil Engineering,Chongqing University,Chongqing 400044,China;School of Civil Engineering,Zhejiang University of Technology,Hangzhou 310014,Zhejiang,China)

机构地区：[1]重庆大学溧阳智慧城市研究院,江苏溧阳213300 [2]重庆大学自动化学院,重庆400044 [3]重庆大学土木工程学院,重庆400044 [4]浙江工业大学土木工程学院,浙江杭州310014

出　　处：《计算机工程》2024年第9期333-343,共11页Computer Engineering

基　　金：国家重点研发计划(2021YFF0500903);国家自然科学基金(52178271,52378128)。

摘　　要：传统钢筋深化设计的大量排布工作是基于手动或半自动化的方式完成的,时间及人工成本较高。目前基于单根钢筋的智能设计方法存在3个问题:同向钢筋没有共享机制,无法利用相似信息,造成大量重复性计算;钢筋之间不具备协同机制,无法保证钢筋前后左右间距的一致性;优化策略基于局部信息进行智能体轨迹规划,难以应对复杂和特殊的钢筋排布场景。为解决上述问题,提出基于模型预测控制(MPC)的多钢筋并行排布智能设计方法,以提高钢筋的排布效率及质量。MPC是一种可以兼顾全局信息和局部信息的寻优策略,通过滚动优化和最优解首元素的应用,提高钢筋应对复杂环境的能力,实现钢筋智能避障和长度优化。多钢筋并行排布策略采用领航跟随者模式实现信息共享和行为协同,利用同向钢筋的相似信息,可减少重复性工作并保证钢筋间距一致。为提高优化效率和质量,结合群体智能的优点进一步改进差分进化算法,并将其作为MPC的求解器。实验结果表明,通过与单钢筋排布方法在不同障碍物情形下的对比,提出的多钢筋并行排布方法能够有效提高钢筋排布的质量、优化效率和应对复杂障碍物场景,并以预制混凝土楼梯构件设计为示例,验证了所提方法的可行性和有效性。A significant portion of the layout work in traditional rebar deepening relies on manual or semi-automated methods,incurring high time and labor costs.The current intelligent design method based on a single rebar faces three primary issues:first,there is no sharing mechanism for same-direction rebars,preventing the use of similar information and resulting in numerous repetitive calculations;second,there is no synergistic mechanism between rebars,failing to ensure the consistency of front-back and left-right spacing;and third,the optimization strategy relies on local information for planning the intelligent body trajectory,making it difficult to handle complex and special rebar placement scenarios.To address these problems,an intelligent design method based on Model Predictive Control(MPC)is proposed for a parallel layout of multiple rebars to enhance the efficiency and quality of the rebar layout.MPC is an optimization strategy that considers both global and local information.Hence,it improves the ability of rebars to navigate complex environments through rolling optimization and optimal solution head elements to realize intelligent obstacle avoidance and length optimization of steel bars.The multiple rebar parallel layout strategy employs the leader-follower model for information sharing and behavioral collaboration,leveraging similar information about rebars in the same direction to reduce repetitive work and ensure consistent rebar spacing.To further improve the optimization efficiency and quality,the differential evolutionary algorithm is enhanced by incorporating the benefits of group intelligence and is used as a solver for MPC.By comparing this method with the single rebar layout method in various obstacle scenarios,experimental results demonstrate that the proposed multiple rebar parallel layout method effectively improves the rebar layout quality,optimizes efficiency,and handles complex obstacle scenarios.The feasibility and effectiveness of this method are validated using the design of prefabricated con

关 键 词：智能建造 深化设计 智能优化设计 钢筋排布 模型预测控制 进化算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]