Nash-equilibrium algorithm and incentive protocol for a decentralized decision and scheduling problem in sustainable electroplating plants
Nash-equilibrium algorithm and incentive protocol for a decentralized decision and scheduling problem in sustainable electroplating plants
复制标题
可持续电镀厂分散决策和调度问题的纳什均衡算法和激励协议
DOI：
10.1016/j.cor.2020.105130
复制
发表时间：
2021
影响因子：
4.6
通讯作者：
Kaize Yu
中科院分区：
文献类型：
--
作者：
Pengyu Yan;Hongru Miao;Ada Che;Kaize Yu
本文解决了考虑环境可持续性的智能电镀厂的分散生产决策和调度问题。由于电镀工艺的特点，零件加工时间受到时间窗口的限制，并在多智能体和分布式制造技术的支持下由独立自主的智能体控制。结果表明，化学/物理罐中的零件加工时间不仅决定生产率，还影响环境成本。因此，应该解决两个基本问题：（a）如何在这种分散的制造环境中找到代理的均衡策略（即零件加工时间）； (b) 如何设计一个激励协议，同时将均衡策略引入到生产率和环境成本方面的最优策略。针对第一个问题，本文建立了非合作序贯博弈模型，模拟策略从不稳定状态到均衡状态的演化过程。然后，开发了一种迭代算法来寻找基于循环双值图的纳什均衡策略。为了解决第二个问题，通过探索时间窗口约束和环境成本函数的特征，为激励协议提出了一种新的启发式规则。数值实验结果验证了纳什均衡算法的性能和启发式规则的有效性。
This paper addresses a decentralized production decision and scheduling problem in smart electroplating plants considering environmental sustainability. Due to the characteristics of the electroplating process, part processing times are confined to time-window constraints and controlled by independent autonomous agents with the support of multi-agent and distribution manufacturing technologies. It is shown that part processing times in chemical/physical tanks not only determine the productivity but also affect the environmental cost. Hence, two essential issues should be addressed: (a) how to find equilibrium strategies of agents (i.e., part processing times) in such a decentralized manufacturing setting; and (b) how to design an incentive protocol inducing the equilibrium strategies towards the optimal ones concerning the productivity and the environmental cost simultaneously. To deal with the first issue, this paper establishes a non-cooperative sequential game model, which simulates the evolution process of the strategies from an unstable state to an equilibrium one. Then, an iterative algorithm is developed to find Nash equilibrium strategies based on a cyclic bi-value graph. To address the second issue, a novel heuristic rule is proposed for the incentive protocol by exploring the characteristics of time-window constraints and environmental cost functions. The results of numerical experiments validate the performance of the Nash-equilibrium algorithm and the effectiveness of the heuristic rule.