求解一类半无限多目标规划的神经网络算法

A Neural Network for Solving a Class of Semi-infinite Multiple Objective Programming

考虑了一类半无限多目标规划问题。根据极大熵方法,先将其转化为半无限单目标可微凸优化问题。再由原问题和对偶问题之间的关系,提出了求解它的能量函数和神经网络模型,并证明该神经网络模型的解是一致渐近稳定的。该模型结构简单,规模小。

Considering a class of semi-infinite multiple objective programming problem.it bases on the maximum entropy method.It translates into semi-infinite differentiable convex optimization problem.Then according to the relationship between original problem and dual problem,energy function and neural network model are proposed to solve it,and proved that the solution of neural network model is uniformly asymptotically stable.The model is simple and small.