基于线性回归近似替代模型的分布式卫星系统不确定性分析方法

分布式卫星系统在空间科学领域具有广泛的应用前景，同时也对概念设计阶段的分析工作提出了更高的要求.在分布式卫星系统概念设计阶段，评价不确定性参数对于最终探测效能的影响有着重要的工程应用价值.传统的不确定性分析方法存在解析困难、数值模拟计算效率低且耦合关系表征不明确的缺点.本文结合概念设计阶段不确定性分析的特点和需求，提出一种线性回归近似替代模型，将不确定性参数对于探测效能影响的结果计算降维为一个线性组合系数的求解问题.利用一个空间多点探测物理场分界面任务作为典型事例进行仿真验证，结果表明，相比传统数值模拟方法，本文方法在计算效率上具有明显优势且对关键不确定性参数有一定识别作用. 

Linear Surrogate Uncertainty Analysis Method for Distributed Satellite System

Distributed satellite system is widely used in space science mission. The complexity of system brings challenges in conceptual design phase. To evaluate the impact of uncertainty parameters on the system detection result is significant. Traditional analysis method such as differential method and Monte Carlo method have disadvantages in solving this problem. Differential method needs system can be expressed explicit while Monte Carlo method is time consuming which cannot fit fast iteration demand. In this paper, a surrogated linear model method is proposed. The problem is solved by reducing the dimensionality of the results of the uncertainty parameters into a linear combination of coefficients. The coefficients are obtained by weighted regression method using data generated by Monte Carlo process. Simulated with a typical distributed satellite mission, the proposed method is valid compared to Monte Carlo simulation method.