论文部分内容阅读
以设计变量存在不确定的火星探测轨道设计模型为研究对象,采用最小最大鲁棒优化方法获得该模型的鲁棒优化解.在传统的最小最大方法的基础上,考虑了决策变量的扰动,提出了一种嵌套的差分演化算法.在该算法中,内层差分演化算法计算不确定域的最差目标函数值,外层差分演化算法获得全局的鲁棒优化解.通过试验对该方法的有效性进行了验证,结果表明:该方法适合于目标函数没有解析表达式、高度非线性的问题,实际工程问题中的不确定性不可忽略.
Taking the uncertain orbit of the Mars exploration orbit design model as the design variable as the research object, a robust and optimal solution of this model is obtained by using the minimum and maximum robust optimization methods. On the basis of the traditional minimum and maximum methods, the perturbations of the decision variables are considered A nested differential evolution algorithm is proposed in which the inner differential evolution algorithm calculates the worst objective function value in the uncertain region and the outer differential evolution algorithm obtains the global robust optimization solution. The results show that this method is suitable for the problem that the objective function has no analytical expression and is highly non-linear. The uncertainty in the actual engineering problem can not be neglected.