摘要
根据人工神经网络的基本优化机理，研究了基于Ｌａｇｒａｎｓｅ函数的适合于求解一般约束问题的神经网络建模方法，探讨了神经无非线性度和拉氏乘子等提高网络优化计算效率的控制策略，测试结果证明了提出的网络和控制策略的可行性和有效性．
According to the basic optimization principle of artificial neural networks,a Lagrange function based neural network modeling approach appropriate for general nonlinear programming is investigated. The control strategy to improve the neural computing efficiency is studied,such as nonlinear degree control strategy of neurons,Lagrange multiplier control strategy,etc.The feasibility and efficiency of the neural network and the control strategy advanced are verifiedwith tests.
出处
《电子学报》
EI
CAS
CSCD
北大核心
1997年第6期85-90,80,共7页
Acta Electronica Sinica
基金
国家自然科学基金
浙江省自然科学基金
关键词
神经网络
约束优化
神经计算
Neural network,Constrained optimization,Neural computing
