基于进化算法的动态环境特征探测  

Detect changes in the characteristics of dynamic environment

作　　者：姚誉[1] 杨小芹[1] 黎明[1] 

机构地区：[1]南昌航空大学自动化学院无损检测重点实验室,江西南昌330063

出　　处：《微计算机信息》2010年第21期172-174,共3页Control & Automation

摘　　要：从动态环境下进化算法性能变化特征的角度,定量分析动态环境的变化特征,利用人工神经网络预测动态环境的特征变化趋势。引入算法性能确认模块和环境探测模块,性能确认模块由种群适应值均值和种群多样性组成,反映环境变化特征;环境探测模块对环境探测分为两个阶段,首先判断环境是否为动态环境,其次研究动态环境变化的剧烈程度,分析动态环境的变化特征,模块包括周期探测和强度探测。结合已探测到的数据用人工神经网络进行分析和处理,建立一个变化趋势的模型。From the point of changes in the characteristics of the performance of evolutionary algorithms in dynamic environment, quantitative analysis changes in the characteristics of dynamic environment, forecast trends of changes in the characteristics of dynamic environment in BP Network. The introduction of performance of evolutionary algorithms module and environmental detection module. Performance of evolutionary algorithms module composed by average population fitness and population diversity reflect the changes in the characteristics of dynamic environment. Environmental detection module composed by cycle and intensity divided into two parts： first, to determine whether it＇s a dynamic environment; Second, analysis changes in the intensity and cycle of the Dynamic environment; Final, The establishment of a trend model in combination with the data has been detected.

关 键 词：动态环境 进化算法 探测模块 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]