【摘要】 提出一种新的求解无约束全局优化问题的方法,此方法把修正的Broyden-Davidon-Fletcher-Powell(BFGS)方法与填充函数方法相结合,可以从目标函数f(x)的当前极小点x1*出发找到另一个局部极小点x2*,且f(x1*)≥f(x2*),然后再以x2*为初始点用同样的方法来求f(x)的更小的局部极小点,反复以上过程,最终可以找到f(x)的全局最小点xg*.经过数值检验,表明方法是可行有效的.更多还原

【Abstract】 This paper presents a modified filled function method for finding a global solution of the unconstrained optimization.The algorithm combines modified Broyden-Davidon-Fletcher-Powell(BFGS) with filled function,and its key idea is to leave from a current local minimizer x*1 to another lower minimizer x*2 of the original objective function f(x).With x*2 replacing x*1,a much lower minimizer of f(x) can be found in the same way.Repeating the above process,the global minimizer x*g of f(x) can be obtained finally.An algorithm is developed from this modified filled function.The computational results show that this algorithm is efficient and reliable.更多还原