【摘要】 文章把实验设计中的均匀设计思想引入优化设计 ,提出了一种将确定性和随机性相结合的优化方法。该方法根据均匀设计原理在优化模型的设计变量空间内均匀分布一系列点 ,然后将可行域内的上述系列布点作为优化计算的系列初始点 ,并选用常规模拟优化算法 ,分别进行优化计算 ,得到优化模型的一系列局部最优点。如果有必要再在局部最优区加密网格 ,然后再模拟数据 ,就能得到这些数据的改进最优解 ,并且重复插值过程 ,该算法可求取非线性多峰函数的全局最优解。文章编制了计算程序 ,给出了计算实例 ,计算结果表明该设计方法是可行的更多还原

【Abstract】 Some ideas of uniform design in the test design are introduced into optimization design,and a new optimization method combining the deterministic method with the random method is proposed. The method generates a series of uniformly distributed points in the variable design space of the optimization model by the principle of uniform design,then regards the points in the feasible domain as the series start points for optimal calculation. After that,the optimal calculation is conducted separately with the optimization algorithms of conventional simulation,and a series of local optimal points are gained for the optimization model. Refining the grids in local optimal space and simulating the extra data,if necessary,the improved value of the optimal solution can be obtained. With interpolating repeatedly,the global optimal solutions of non-linear multi-peak function can be derived by the method. The article compiles the calculating program and provides a real example of calculation. The results show this method is feasible.更多还原