摘要
绿色物流领域新兴的电动汽车车辆路径问题,由于需要对车辆路径和充电决策同时优化,搜索空间急剧增大,且需要同时满足容量和电量双重约束,现有方法难以快速找到质量较优的可行解。为此,提出一种基于双种群的协同进化算法,通过忽略电量约束构造简单带容量约束的车辆路径问题,辅助原始复杂问题的快速求解。为实现其间信息交互,设计一种基于改进距离邻接矩阵的解序列特征表示方法,旨在同时获取客户访问顺序和车辆指派信息;利用降噪自编码器构建2个问题解之间转换关系,以实现问题域间知识迁移。将该算法与目前常用的3种启发式算法和2种进化算法在不同规模测试集上进行对比,试验结果表明所提算法具有更快收敛速度且所获解集具有更好收敛性。
The emerging field of green logistics presents a challenge in the form of electric vehicle routing.This issue requires simultaneous optimization of routing and charging decisions,significantly expanding the search space.Moreover,solutions must comply with capacity and power constraints,making it difficult to quickly find feasible solutions using existing methods.To address these challenges,we propose a dual population-based co-evolutionary algorithm.This approach involves constructing a simpler problem to expedite the solution process for the original,more complicated problem.To facilitate information exchange between these two heterogeneous problems,we designed a solution representation method.This method,which is based on an improved distance adjacency matrix,allows to obtain information on customer visits and vehicle assignments.Subsequently,we employed a commonly used denoising autoencoder to establish the transformation relationship between solutions from these two problems.This step enables knowledge transfer between the two problem domains.Our proposed algorithm was tested against three heuristic methods and two evolutionary algorithms on test sets of different sizes.The experimental results show that the proposed algorithm not only converges faster but also yields solutions with superior convergence.
作者
王朝
秦芳
刘蓉蓉
江浩
WANG Chao;QIN Fang;LIU Rongrong;JIANG Hao(School of Artificial Intelligence,Anhui University,Hefei 230601,China)
出处
《智能系统学报》
CSCD
北大核心
2024年第2期438-445,共8页
CAAI Transactions on Intelligent Systems
基金
国家自然科学基金项目(62106002,62372001)。
关键词
绿色物流
电动汽车车辆路径问题
电量约束
双种群
进化算法
距离邻接矩阵
降噪自编码器
知识迁移
green logistics
electric vehicle routing problem
electricity constraint
two-population
evolutionary algorithm
distance adjacency matrix
denoising autoencoder
knowledge transfer
