摘要
提出一种基于主成分分析法(PCA)和改进型多步E lm an网络的实时预报方法.该方法能够在保留大量原始数据信息的前提下,消除样本数据间相关性,简化网络结构,通过动态递归算法实现复杂非线性系统实时预报.将该网络应用于宝钢某高炉铁水含硅量的预报,以±0.05作为预报误差,预报命中率达到88.17%.
A real-time prediction method based on principal component analysis （PCA） and improved multi-step Elman net is presented. With most original data information, this method eliminates the relativities among data and simplifies the net structure by processing the sample data with PCA. It can predict complex and nonlinear system with dynamic recurrent algorithm. The hit rate reached 88. 17% to forecast the silicon content of a blast furnace on Bao Steel with errors ranging ±0. 05.
出处
《控制与决策》
EI
CSCD
北大核心
2006年第11期1312-1315,1320,共5页
Control and Decision
基金
辽宁省自然科学基金项目(20042020)
