Abstract: Value-decreasing resources are widely present in retail service systems and intelligent scheduling scenarios, with the core challenge being how to achieve efficient resource allocation and circulation under conditions of continuously declining resource value, member autonomy, and decentralized information. To address the inefficiencies in scheduling and revenue loss experienced by convenience store groups during the operation of value-decreasing goods—caused by resource depreciation, member autonomy, and limited information disclosure—this paper proposes a multi-party win-win mechanism for retail service systems. The mechanism is modeled using a Multi-Agent System (MAS), allowing each service member to make autonomous decisions under limited information. Through dynamic demand reporting and optimized transaction matching, the mechanism enables internal resource reallocation within the group. It fully considers the heterogeneous strategic behaviors of members under local knowledge, indirectly reflecting their true intentions and goals through actions such as bidding and auctions, and promotes continuous participation from all parties through a profit redistribution mechanism. To validate the effectiveness of the mechanism, a simulation environment for the service system is constructed based on the JaCaMo platform, featuring various types of agents (stores, customers, headquarters, etc.) and market demand disturbance scenarios. A systematic comparison is conducted among the proposed multi-party win-win mechanism, the VCG auction mechanism, and the traditional decentralized operation mechanism. Experimental results show that, over a 30-day operational cycle, the multi-party win-win mechanism increases the group's overall service revenue by approximately 15.4%, and outperforms both the VCG mechanism and the decentralized operation model in key metrics such as customer satisfaction rate and resource usage rate. This research provides a theoretical foundation and empirical support for the practical application of distributed scheduling of value-decreasing resources and multi-party win-win mechanisms in service systems.

摘要： 价值单调递减资源广泛存在于零售服务系统与智能调度场景中，其核心挑战在于如何在资源价值持续递减、成员自主经营和信息分散的背景下，实现高效的资源分配与流转。针对便利店集团在价值单调递减商品经营过程中因资源价值递减及成员自治、信息有限披露导致的调度低效与收益损失，本文提出了一种面向零售服务系统的多方共赢协同机制。该机制基于多智能体系统（Multi-Agent System, MAS）建模，允许集团各服务成员在有限信息条件下自主决策，通过动态需求上报与交易撮合优化，实现集团内部资源的重新配置。机制充分考虑成员在局部知识下的策略异质性行为，通过出价与拍卖等事件间接体现成员的真实意图与目标，并通过利润再分配机制促进各方持续参与。为验证机制有效性，基于JaCaMo平台构建服务系统仿真环境，设定多类智能体（门店、顾客、总部等）及市场需求扰动场景，系统性对比多方共赢机制、VCG拍卖机制与传统分散经营机制。实验结果显示，在30天经营周期内，多方共赢机制将集团整体服务收益提升约15.4%，并在顾客满足率与资源使用率等关键指标上优于VCG机制与分散经营模式。研究为价值递减资源的分布式调度与服务系统多方协同机制的实际应用提供了理论基础与数据支撑。