暂态混沌神经网络的算法改进及其在TSP中的应用  

Improved Algorithm of Transient Chaotic Neural Network and Its Application in TSP

作　　者：丛爽[1] 王桢宁[1] 

机构地区：[1]中国科学技术大学自动化系,合肥230027

出　　处：《科技导报》2009年第7期60-63,共4页Science & Technology Review

基　　金：国家自然科学基金项目(60774098)

摘　　要：根据暂态混沌神经网络所具有的通过负反馈系数对神经网络引入退火机制,并通过网络动态运行中系数的不断减少来使网络从初期的混沌状态收敛到稳定收敛状态的特点,将其应用到中国31个省会城市的旅行商问题中。针对应用中所存在的解决较大数目的组合优化问题不易收敛到最优解的情况,将暂态混沌神经网络和传统的启发式算法相结合,提出一种组合算法。给出了改进算法的具体设计以及应用实施的步骤,并对应用结果进行了性能对比分析。A Transiently Chaotic Neural Network （TCNN） is used in the Chinese traveling salesman problem with 31 capital cities by introducing the anneal mechanism through a negative gain, and letting the gain decrease gradually during the dynamic running of the network in order to allow an initial chaotic state converge to a stable state. A new combined algorithm is proposed, which combines the TCNN with traditional heuristic algorithm and can facilitate the combination optimization of problems with a large number. The detailed design and implementation of the algorithm are given, together with application results and performance comparison.

关 键 词：暂态混沌神经网络 参数选择 组合算法 中国旅行商问题 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]