在轨空间智能制造:分布式调度建模与优化

In-space intelligent manufacturing:Distributed scheduling and optimization

在轨空间制造系统是在行星大气层外的需要地面工厂、在轨空间工厂、天地运载工具协同的以进行空间设施建造为目标的一类分布式制造系统.分布式调度建模和高效优化求解技术是实现在轨空间智能制造的关键技术之一.本文针对一类具有组件地面分布式制造及运输、地空分批次运输、组件在轨装配等典型特点的在轨空间智能制造系统,将其分解为分布式同质流水线调度,考虑运输时间的同速并行机调度,考虑工件释放时间、机器可用时间、机器处理能力的单机批调度以及考虑组件释放时间、优先约束的单机调度等问题,并基于模型协调思想建立以最小化组件生产到产品装配总时长为目标的分布式多阶段调度模型.进而,将用于求解连续优化问题的易理优化算法扩展到离散调度问题,提出求解该分布式调度问题的基于易理优化的模因算法.基于中规模、大规模算例的仿真结果和算法分析比较表明:相较于粒子群算法、教学算法、水波算法等智能优化算法,所提算法是一种求解分布式多阶段调度问题的可行、有效算法.值得一提的是,这是第一篇关于在轨空间智能制造系统调度优化的研究.

In space manufacturing system is a kind of exoatmospheric, distributed manufacturing system which coordinates the ground-based factories, in-orbit space factories, and space transportation vehicles and aims at space facilities construction. Key issues to realize intelligent in-orbit space manufacturing are the modeling of the distributed scheduling systems as well as the designing of the efficient and effective optimization methods. In this paper, an in-orbit space intelligent manufacturing system with distributed ground manufacturing and transportation, ground-air batch transportation and in-orbit assembly process is addressed. The scheduling of the system is modeled as multi-stage distributed scheduling problem, which is decomposed into four subproblems, i.e., distributed homogeneous flow shop scheduling, parallel machines scheduling with transportation time, single machine batch scheduling problem with release time, machine available time and machine capacity constraints, and single machine scheduling problem with release time and precedence constraints, with respect to the criterion of minimizing of the maximum completion time. In addition, the recently proposed I Ching philosophy inspired optimization (ICO) which originally focused on continuous optimization is extended to solve combinatorial optimization, and ICO based memetic algorithm (ICO-MA) is proposed to solve the aforementioned scheduling problem. Experimental results on middle scale and large scale instances show the proposed algorithm is effective and efficient compared with the state-of-the-art algorithms, e.g., particle swarm optimization, teaching learning based optimization and water wave optimization. The proposed ICO-MA could be a feasible and effective algorithm to solve distributed multi-stage scheduling problems. To the best of our knowledge, it is the first study on distributed scheduling of in-orbit space intelligent manufacturing system.