基于内分泌调节机制的行为自组织算法  

Self-Organization Algorithm of Behaviors Based on Endocrine Regulation Mechanism

作　　者：黄国锐[1] 曹先彬[1] 徐敏[1] 王煦法[1] 

机构地区：[1]中国科学技术大学计算机科学技术系,合肥230026

出　　处：《自动化学报》2004年第3期460-465,共6页Acta Automatica Sinica

基　　金：国家自然科学基金项目(69971022)资助~~

摘　　要：借鉴内分泌系统的高层调节机制,提出了一种新的自主体行为自组织算法.此算法用神经系统接受环境信息,通过一种情感学习模型来产生情感因子(即生物激素),再由情感因子来调节神经系统的记忆和行为决策,最后神经系统的记忆与行为模式由遗传系统得以继承.其优点是避免了神经系统复杂的自学习过程,同时保证系统的行为决策具有较高的自组织、自适应能力.为了验证算法的有效性,倒立摆控制的仿真实验表明该算法具有很强的自适应求解能力.Motivated by the high-level regulation mechanism of endocrine system, an a-gent self-organization algorithm of behavior is put forward. This algorithm receives en-vironmental information using a neural system, produces emotional factors (hormones) through an emotion learning model, then uses these factors to regulate the neural system. At last, the memorization and behavior mode of the neural system are exported to the genetic environment. So it avoids the self-study of neural system and gets better self-organization and self-adaptation power. In order to prove the validity of this algorithm, it is used to control an inverted pendulum. The result shows strong self-adaptation performance of the algorithm.

关 键 词：内分泌系统 行为自组织 情感学习 神经网络 遗传算法 倒立摆 

分 类 号：TP273.5[自动化与计算机技术—检测技术与自动化装置]