摘要
应用小波分析方法对高炉铁水硅含量进行预测。通过小波变换将铁水硅含量的时间序列依三重尺度分解成不同的层次,并对不同层次上的序列分别运用合适的自回归模型进行预测,然后通过序列重构得到原始时间序列的预测结果。利用山东莱钢1号高炉在线采集的数据作为实际预测案例,与原始时间序列的自回归模型预测结果比较,小波预测方法显著提高了预测命中率。
A new approach to predict the silicon content in hot metal based on wavelet analysis is proposed. By wavelet transform, the original time series for silicon content in hot metal can be decomposed into several series according to the scale. Then the decomposed time series are reconstructed and forecasted with appropriate auto-regression models to obtain the forecast results of original time series. Using the real time data on No. 1 BF in Laiwu Iron and Steel Group Co. , the practical results showed that the proposed method has increased significantly the hitting rate of silicon content in hot metal comparing with traditional auto-regression forecasting model.
出处
《钢铁》
CAS
CSCD
北大核心
2005年第8期15-17,37,共4页
Iron and Steel
基金
国家级科技成果重点推广计划项目(99040422A)
关键词
铁水硅含量
预测
小波分析
AR模型
silicon content in hot metal
prediction
wavelet analysis
auto-regression model
