论文部分内容阅读
以包钢6号高炉、邯钢7号高炉和莱钢1号高炉在线采集的铁水含硅量([Si])的时间序列为样本,利用多分辨分析剔除样本的长期趋势,对样本保留的波动趋势进行多重分形特征辨识.通过计算广义Hurst指数、尺度函数、多重分形谱,全面、细致量化了序列的局部及不同层次的波动奇异性.计算结果表明:去除长期趋势后,三座高炉[Si]序列的波动呈现显著多重分形特征,这样的波动过程仅用单一的Hurst指数或box维数来描述是不够的.
The time series of silicon content in molten iron ([Si]) collected by Baosteel No.6 blast furnace, No.7 blast furnace of Handan Iron and the No.1 BF of Laiwu Steel were taken as samples. The long-term trend of samples was eliminated by using multi-resolution analysis. The trend of fluctuating trend was identified by multifractal feature.The volatility singularity of local and different levels of the sequence was quantified comprehensively and quantitatively by calculating generalized Hurst exponent, scaling function and multifractal spectrum.The results show that after removing the long-term trend, the three blast furnaces [ Si] sequence exhibits significant multifractal characteristics. Such a fluctuation process is not sufficient to describe with a single Hurst exponent or box dimension.