摘要
针对当前多智能体协同优化缺乏碳排动态反馈机制致使施工调度与双碳目标脱节问题构建面向“双碳”目标的多智能体协同优化模型,通过嵌入碳排成本函数的智能体建模、基于Q-learning的动态协同决策机制及PID(Proportional Integral Derivative)闭环反馈引擎,选取三类典型工程进行仿真验证。结果表明,相较于传统关键路径法(Critical Path Method,CPM)和静态碳约束调度,本模型显著提升了碳效控制能力,平均碳效达标率达94.5%,重调度频次较传统CPM下降45%,响应时间稳定在8.0~8.5 s,资源闲置率被有效控制在4.3%以内,展现出优异的调度鲁棒性与资源利用效率,可推动“双碳”目标在施工环节精准落地。
Addressing the current issue of multi-agent collaborative optimization lacking a dynamic carbon emission feedback mechanism,which leads to a disconnect between construction scheduling and the“dual carbon”goals,the study constructs a multi-agent collaborative optimization model oriented toward the“dual carbon”objectives.The model integrates carbon emission cost functions into agent modeling,employs a Q-learning-based dynamic collaborative decision-making mechanism,and incorporates a PID closed-loop feedback engine.Experiments are conducted on three typical projects.Results demonstrate that,compared with traditional CPM and static carbon-constrained scheduling,the proposed model significantly enhances carbon efficiency control:the average carbon efficiency compliance rate reaches 94.5%;rescheduling frequency is reduced by 45%compared to traditional CPM;response time remains stable at 8.0~8.5 s;and resource idleness is effectively controlled within 4.3%.These outcomes showcase the model’s excellent scheduling robustness and resource utilization efficiency,thereby promoting the precise implementation of the“dual carbon”goals at the construction level.
作者
呙伟
Guo Wei(CCCC Harbor Second Engineering Co.,LTD.,Wuhan 430040,China)
出处
《黑龙江科学》
2026年第6期159-161,共3页
Heilongjiang Science
关键词
建筑施工
双碳目标
多智能体系统
碳感知调度
动态反馈控制
Construction
Dual carbon goals
Multi-agent system
Carbon-aware scheduling
Dynamic feedback control
