摘要

This article proposes a novel collaborative optimization (NCO) framework that uses a persuasive multi-agent negotiation method for satellite system design. Satellite system design is a typical multi-disciplinary design optimization problem. The traditional collaborative optimization (CO) is a competitive method but has the disadvantage of inefficient convergence caused by the interdisciplinary compatibility constraints in system-level optimization. By introducing the multi-agent negotiation method, an NCO framework is proposed, in which the system-level and subsystem-level variable negotiations replace the system-level compatibility constraints in traditional CO. The negotiation introduces a modification for candidate values that do not satisfy the compatibility constraints, ensures that an alternative compatible candidate will be obtained and has high convergence efficiency. Two numerical benchmark functions and a multi-disciplinary satellite system design are used to analyse the performance of the proposed NCO. The simulation results show that NCO can efficiently ensure parameter interdisciplinary matching and advanced convergence performance.