不确定性下的固体火箭发动机性能精确代理建模方法

The uncertainty of thrust curve of solid rocket motor （SRM） need be accurately quantified to ensure the reliability and improve the load capacity of solid rocket on multidisciplinary design optimization （MDO） in the scheme design phase. An accurate surrogate modeling method is proposed in this paper to solve the problem of difficulty in accurately quantifying the uncertainty of thrust curves and the low efficiency of uncertainty analysis in the process of MDO considering uncertainty. First the proper orthogonal decomposition is used to realize the dimensional reduction of the engine thrust curve with uncertainty. Then a Kriging surrogate model is built to predict the first four statistical moments of base modal coefficients. Finally， an accurate probability distribution model of the base modal coefficients is established by the maximum entropy method. The precise distribution of the thrust curve is then calculated. The result of uncertainty trust model used on a star grain SRM shows that the predicted confidence level of thrust uncertainty distribution reaches 98%. And the time of single uncertainty analysis is reduced by 99.92% compared with Monte Carlo method.