Metamodel Assisted Multidisciplinary Design Optimization for Satellite with a Large-Size Payload

To settle the challenge of complex satellite system design, the multidisciplinary design optimization (MDO) problem for a satellite with a large-size payload is investigated in this paper. The satellite MDO problem is defined to minimize the overall mass subject to several practical constraints such as the structural natural frequency and transfer time. Then, considerable efforts are made to establish the analysis models of orbital transfer, space environment, power, geometry, structure, and mass disciplines, considering the inter-coupled relationship between the satellite platform and the payload. Furthermore, a filter-based sequential radial basis function (FSRBF) method is employed to settle the studied satellite MDO problem efficiently and effectively. In this approach, a radial basis function is constructed and adaptively refined to approximate the expensive multidisciplinary analysis (MDA) models for optimization, which notably lowers the computational cost. After optimization, the overall mass of the satellite is successfully reduced by 116.17kg (3.35%) compared with that of the initial design, and all the constraints are met. Moreover, the cost of the metamodel-based optimization method is only 24.1% of that of the differential evolutionary algorithm, which indicates the practicality and effectiveness of this study.