一种新的近似目标函数优化方法

本文提出了一种新的优化方法,适用于变量较少、具有p次方形式、无法求导的复杂的目标函数.给出了由误差函数求出近似误差函数和近似目标函数,并以求近似目标函数的最优值来替代求目标函数最优值的详细算法.列出了本算法对于10种试验函数的计算实例.计算结果与其他方法的比较证明了本算法的有效性.最后讨论了本算法的某些局限性.

An Optimization Algorithm by Minimizing Approximate Objective Function

A new optimization method is proposed.It is suited properly to very complicated objective functions with fewer variables, which are in the form of the p th power,and whose derivatives cannot be calculated. A detailed algorithm generating the approximate error function and approximate objective function by the error function, and minimizing the approximate objective function instead of minimizing the object function,is derived. Ten numerical examples of minimizing test functions with this algorithm are listed. The efficiency of this method is verified by comparison of the results with those from other techniques. Finally, some limitations of this method are discussed.