基于冲突合作关系的生鲜选址-路径多主体优化模型与算法

Multi-agent optimization model and algorithm for perishable food location-routing problem with conflict and coordination

针对城市生鲜农产品配送成本高,产品损耗大等特点,考虑生鲜销售商和外包冷链配送公司共同参与又独立决策的实际运营情况,建立了基于冲突合作关系的生鲜配送选址-路径多主体优化模型.模型中,考虑客户模糊时间窗,主导层生鲜销售商以系统总成本最低为目标,而从属层冷链配送公司仅考虑运输相关成本最小化.设计了GAPSO混合算法求解该问题,结合精英选择策略和自适应权重粒子更新策略,采用田口分析法获取参数合理取值.与CPLEX求取小规模算例精确解对比,GAPSO算法在保持准确性的同时,求解时间降低了96.17%;对基于Barreto和Prins经典LRP基准案例集的中大规模算例,与HybridGA算法和已知最优解BKR对比分析,结果表明GAPSO对于中小规模问题,能很好的收敛于全局最优解,对于大规模问题能求得近似最优解,能为实际生鲜选址和配送问题提供有效决策支持.

Considering high distribution cost and high product loss,and complicated relations between a fresh food seller and a hired transportation company,a new model is formulated for perishable food locationrouting problem with conflict and coordination(PFLRP-CC).In this model,customers’ time windows are taken as fuzzy variables,the seller as the leader aims at minimizing total costs,while the transportation company is the follower only caring about transportation costs.A GAPSO hybrid algorithm is developed to solve the PFLRP-CC,where an elite selection and an adaptive weighted particle optimization are adopted.Taguchi analysis is used to obtain reasonable values for GAPSO parameters.Small-size problems are solved by GAPSO and then compared with the exact method using CPLEX,the results of which show that GAPSO reduced the computing time by 96.17%;As for medium and large-size cases from Barreto and Prins benchmarks,compared with HybridGA and the best known results,the proposed GAPSO can effectively converge to optimal solutions for medium cases,and find approximate optimal solutions for large cases,which indicate that the GAPSO is efficient and effective for solving the real PFLRP-CC.