摘要
提出一种基于聚类和主成分分析的神经网络模型,用于高炉运行指标的实时预测.首先采用谱系聚类将特性分散的样本划分成不同的子类,然后采用主成分分析方法对影响目标数据的众多变量进行降维处理,在此基础上,构建了高炉运行指标的神经网络预测模型,大大改善了预报的精度和效率.通过对采集的高炉数据进行测试,表明本文提出方法的有效性.
Presents a neural network prediction model based on clustering and principle components analysis（PCA）,which is applied to real-time prediction of economic-technical indexes in blast furnace. The proposed method divides samples which have characteristic of decentralization into different sub-classes with the aid of pedigree clustering ,and uses principle components analysis method to reduce the dimensionality of the feature space,and then builds neural network forecasting model. The experiments show that the proposed method is efficient.
出处
《小型微型计算机系统》
CSCD
北大核心
2005年第12期2160-2163,共4页
Journal of Chinese Computer Systems
基金
沈阳市科委科技项目(1041036-1-06-07)资助
关键词
聚类
主成分分析
神经网络模型
clustering
principle components analysis
neural network model
