引入多Agent协商的协同优化在卫星设计中的应用

A multi-Agent negotiation based collaborative optimization application in satellite design

卫星设计是复杂的多学科优化问题,协同优化算法具有模块化特点适合于处理多学科优化问题。在对卫星进行多学科优化时发现,现有的协同优化算法为了保证共享变量在系统级和分系统级间的一致性,收敛困难,计算量大,不对分系统自身进行优化不能保证卫星整体性能最优。本文通过引入Agent概念和基于Agent的协商算法,提出了一种基于劝说式多Agent协商的协同优化设计算法。针对对地观测卫星,以总质量最小为目标,建立了包含结构、轨道控制、姿态控制和电源4个分系统Agent和1个系统级Agent的卫星优化设计模型并进行了协同优化设计。研究结果表明:引入Agent协商算法的协同优化算法,可以保证共享变量的一致性;在协同优化算法中增加分系统自身优化模块,可以保证在系统级优化过程中各分系统一直都是最优的。

Satellite design is a complex problem of multidisciplinary optimization. Collaborative optimization algorithm,with its feature of modularization,is suitable for being applied to the multidisciplinary optimization of satellite design. To guarantee the consistency of shared variables between the system level and the subsystem level,the traditional collaborative optimization has the limitation such as convergence dilemma and excessive computational cost. And the traditional collaborative optimization which cannot optimize the subsystems cannot ensure the optimization of satellite overall performance. This paper proposes a multi-Agent negotiation based collaborative optimization algorithm by introducing the concept of Agent-based model and multiAgent negotiation. We take earth observation satellite as an example,and with the object that the total weight is the minimum,a system-level Agent model and four subsystem-level Agent models including structure subsystem,orbit subsystem,attitude subsystem and power subsystem are built. The results indicate that the multiAgent negotiation based collaborative optimization algorithm can ensure the consistency of shared variables,and the subsystem optimization model ensures that the subsystems are optimal when system-level optimization is carried out.