摘要
采用粗糙集理论约简属性,在保留重要信息的前提下消除冗余信息,简化了神经网络结构,提高了网络训练速度。采用这种方法,用某城市污水处理厂的实际水质参数数据,建立了 SVI 基于粗糙集-神经网络的软测量模型。和未经粗糙集预处理的神经网络模型进行了比较,结果表明有粗糙集预处理后,不仅测量值的误差值更小,而且输入参数量从9个降至4个,大大降低了输入数据的维数,减少了神经网络的训练时间及训练步数,更有利于软测量模型的实用化。
The writer reduces the attributions of madel,and eliminates superfluous data by rough set,this result in simplification of the model structure and speedup of the training speed.By this method,the SVI soft measure model based on rough set using artificial neural network is established,using practical data of water quality parameters in some municipal wastewater treatment plant.The result indicates that the error is smaller when the rough set-artificial neural network model is used than isn't used,and the amount of the input parameters is reduced from 9 to 4.The dimensions of the input data are decreased greatly,and the training time and steps of the artificial neural network are reduced.
出处
《电气自动化》
北大核心
2005年第3期64-66,69,共4页
Electrical Automation
基金
国家"十五"小城镇科技发展重大项目(2003BA808A17)资助
关键词
粗糙集
人工神经网络
软测量
污水参数
rough set
artificial neural network
soft measure
wastewater parameter
