基于复杂模拟的战斗机系统失效敏感性分析

战斗机的系统试验往往成本较高,随着战斗机系统的日益复杂化,基于计算机模拟的表征系统失效信息的功能响应量计算也越来越耗时费力。针对此类基于复杂计算机模拟的敏感性分析问题,提出了一种结合Sobol方法和基于主动学习的Kriging模型的敏感性分析方法,称之为AK-S(Adaptive-Kriging-based Sobol)方法,AK-S方法通过Kriging预测来代替真实响应值计算,因而可以更加高效地计算各输入变量的敏感性指标并得到重要度排序。通过与直接蒙特卡洛法(直接法,MC)和传统Sobol法对数值算例的处理结果进行对比,AK-S方法的计算效率和精度得到了证明。最后,AK-S方法被应用于基于复杂模拟的实际工程案例的失效敏感性分析,并获得了敏感性指标。AK-S算法被证明在同等计算精度的条件下,其效率大大高于MC和传统Sobol法,能很好地解决工程中基于复杂计算机模拟下的失效敏感性分析问题。

Sensitivity Analysis for Fighter System Failure Based on Complex Computer Simulation

The full-scale test of the fighter system is often very expensive,and with the increasing complexity of the fighter system,the computation of system failure information based on computer simulation is becoming more and more time-consuming and laborious.To deal with the sensitivity problem based on complex computer simulation,an algorithm combining Sobol method and active-learning-based Kriging model is proposed,which is referred to as AK-S (Adaptive-Kriging-based Sobol) algorithm.By using adaptive Kriging to predict response value instead of computing the actual response value,AK-S algorithm can more effectively obtain the sensitive indices and the relative importance of each input variable.The accuracy and efficiency of AK-S were demonstrated by comparing the results of the direct Monter Carlo (MC) method and the Sobol method in solving the numerical sensitivity problem.The AK-S algorithm was applied to the sensitivity analysis of actual engineering cases based on complex simulation and the sensitivity indices were obtained.AK-S is demonstrated to be more efficient than MC and Sobol methods with the same accuracy,and it can efficiently solve the failure sensitivity analysis problem based on complex computer simulation in engineering.