基于改进能量函数的混沌神经网络JSP解法  

Chaotic neural network method for job-shop scheduling problem based on improved energy function

作　　者：李桂秋[1] 陈志旺[2] 

机构地区：[1]常州机电职业技术学院,江苏常州213164 [2]燕山大学工业计算机控制工程河北省重点实验室,河北秦皇岛066004

出　　处：《计算机应用研究》2010年第5期1726-1728,共3页Application Research of Computers

摘　　要：针对传统Hopfield神经网络(HNN)在求NP类问题的解时易陷入局部最优点的不足,提出基于改进能量函数的模拟退火混沌神经网络算法。通过在Hopfield神经网络中引入混沌机制,并结合退火策略控制混沌动态,有效避免了陷入局部极小的缺陷,因此将其用于求解JSP(作业车间调度)。算法改进了表示JSP的换位矩阵,给出了包含目标函数的能量函数,保证了网络的稳态输出为全局可行解。This paper presented chaotic neural network of simulated annealing（ACNN）method based on improved energy function for the shortage that the conventional Hopfield neural networks（HNN） tended to be trapped into local minima.Controled HNN with chaos mechanism and chaos dynamics by annealing strategy,therefore ACNN would not fall into local minima and was used to resolve the job-shop scheduling problem（JSP）.This method improved permutation matrix of JSP and gave an energy function with objective function,which made the network steady-state output had the overall situation feasible solution.

关 键 词：组合优化 作业车间调度 混沌神经网络 模拟退火 能量函数 

分 类 号：TG273[金属学及工艺—铸造]