带自由变量的符号混合整数非线性规划问题的全局优化

Global Optimization of Signomial Mixed-integer Nonlinear Programming Problems with Free Variables

本文针对带自由变量的符号混合整数非线性规划问题(SMINLP)提出一种全局优化算法.该算法首先利用等价转化将(SMINLP)中的自由变量转化为正变量,再利用凸化技术建立其凸松弛规划(RCP).通过对(RCP)可行域的细分及一系列(RCP)的求解过程,使得提出的算法具有全局收敛性.

This study proposes a global optimization algorithm for locating global minimum of a signomial mixed-integer nonlinear programming(SMINLP)problems with free variables.By utilizing equivalent transformation,free variables in SMINLP are first transformed into positive variables,by convexication strategies are relaxation convex programming(RCP)about SMINLP is then established.The proposed branch and bound algorithm is convergent to the global minimum of SMINLP through the successive renement of the feasible region of RCP and the solutions of a series of RCP.