2011 Volume 51 Issue 4 Pages 588-592
With the data of silicon content in the hot metal collected from the three blast furnaces as a sample space, the multiscale trend decomposition was employed to analyze the local trend of time series. The singularity of fluctuation at different scales was quantified by the Hurst index. The Horton-Strahler topological classification of ramified pattern was constructed to identify the difference between Hurst index computed on different time series. Simulation results demonstrate that time series of silicon content in the hot metal from the three different blast furnaces exhibits significant local singularity. The duration of increasing and decreasing trends as well as the fluctuation of silicon content series are the main cause of differences among the Hurst index computed.