考虑需求波动的单元装配系统构建问题的多目标模型

Multi-Objective Optimization Model for Seru Production System Formation Considering Demand Fluctuation

顾客个性化需求的增强导致企业需要应对动态变化的市场需求,在非确定性需求的环境下提升生产效率和应变能力。日本式单元装配模式以其较高的生产效率和柔性已经逐步成为应对多品种小批量需求的生产方式。由于现有研究多集中于确定性需求,本文针对日本式单元系统构建问题的特点,考虑需求波动的情境下流水线装配向单元装配系统转化问题,构建单元装配系统转化的多目标优化模型。该模型以最小化总加工周期的期望和方差为目标,决策单元装配系统的构建方案,以提升需求波动情境下系统的稳定性和对需求波动的应变能力。根据问题特点设计基于NSGA-II的优化算法求解大规模问题,并结合文献数据说明了模型的应用。数值算例验证了方法的有效性,结果表明考虑需求波动情境下的单元系统构建可以提升生产系统应对波动需求的能力以及稳定性。

With the increasing of demand diversification,manufacturing enterprises encounter challenges of demand uncertainty and fluctuation.Japanese Seru（Japanese style assembly cell）production system（SPS）has received more attention with its efficiency and flexibility especially under uncertain demand.More researches focus on how to implement Seru production system appropriately.However,most of current research focus on the deterministic demand.To enrich the research in this field,a line-Seru conversion problem is investigated in this paper.Considering the fluctuation of product demand,a multi-objective optimization model of line-Seru conversion problem is developed to minimize the expectation and variance of make span（MS）.The minimized expectation of make span offers quick response to a production system with fast delivery.The minimized variance of make span represents the high system stability.The aim of this paper is to offer a method for formulating a robust Seru production system which can ultimately deal with the fluctuation of product demand.In order to obtain optimal solutions of this NP-Hard problem,an NSGA-II（non-dominated sorting genetic algorithm II）based algorithm is developed.Combining the benchmark data from Yu Yang et al[18]with fluctuation factors,numerical experiments are conducted to prove the effectiveness of the model and illustrate how to implement the method.The experiment results show that both delivery time and stability of SPS will be improved with considering the demand fluctuation.At the same time,performance of the system will decline with the increasing of batch size.