一种非线性优化的神经网络  

A METHOD OF NONLINEAR OPTIMIZATION BASED ON NEURAL NETWORK

作　　者：任宏萍[1] 陆建东[1] 尹申明 

机构地区：[1]华中理工大学计算机系

出　　处：《计算机研究与发展》1995年第6期50-54,共5页Journal of Computer Research and Development

摘　　要：本文提出了用人工神经网络求解具有约束条件的非线性优化问题的具体方法，分析了神经网络能量函数的构成形式，并在常规的Ｈｏｐｆｉｅｌｄ网络模型的基础上构造了一个非全局连接的神经网络动力学模型。这种修改的Ｈｏｐｆｉｅｌｄ网络克服了常规的Ｈｏｐｆｉｅｌｄ网络在求解非线性优化问题时权值不好映射的困难，具有结构清晰．易于软件模拟和硬件实现的优点。This paper describes a specific method of applying neural network approach to solve nonlinear optimization problem. The forming of the energy function is analyzed, and then a neural network dynamic model of nonglobal link on the basis of the regular Hopfield network model is structured. The network improved overcomes the difficulty of the weight map when using regular Hopfield network to solve nonlinear optimization problem. The network structure is distinct,and the software analog and the hardware implementation are easy.

关 键 词：HOPFIELD模型 非线性优化 神经网络 目标函数 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]