具有非负困难度的符号几何规划的一种分解方法

A Decomposition Method for Signomial Geometric Programming with Nonnegative Degree of Difficulty

本文针对具有非负困难度的符号几何规划问题提出了一种新的分解方法。该方法首先利用指数变换及矩阵理论,将原问题等价地转化为一个非线性程度较低的可分离规划,然后,将所得等价问题分解成一系列易于求解的子问题,并且当困难度为零时,文中给出了求解子问题精确解的方法。最后,通过数值实例验证了新方法的有效性和可行性。

In this paper, a decomposition algorithm for signomial geometric programming with nonnegative degree of difficulty is proposed. Through the exponential transformation and matrix theory, a problem equivalent to the original problem can be obtained which is a separable programming with lower nonlinear degree. This equivalent problem is decomposed into several suhproblems, and the explicit solution to each subproblem can be acquired when the degree of difficulty is zero. At last, some examples are given to verify the feasibility and efficiency of our algorithm.