本文描述了一种逐步缩小搜索域的全局优化法,利用此方法在全域随机投点进行判断,可以确定全局最小点所在的邻域,并给出该域中近似最小点值及其函数估值。在此邻域中再用局部优化梯度法,即可得到全局最小点。此法收敛迅速。使用方便,已用于微波电路最优化通用设计程序。最后给出验算结果和应用实例。 This paper describes a globally narrowing search domain optimization method, the use of this method in the global random vote to determine the neighborhood where the global minimum point, and gives the approximate value of the domain and its function valuation . In this neighborhood and then use the local optimization gradient method, you can get the global minimum. This method converges rapidly. Easy to use, microwave circuit has been used to optimize the general design process. Finally, the checking result and application examples are given.