Multiscale models on time series of silicon content in blast furnace hot metal based on Hilbert-Huang transform

The processes in blast furnace ironmaking are often extremely complex and measurements from their interior are scarce due to hostile conditions, while the high standard product requires that the silicon content be strictly controlled. Mathematical models can play a central role in achieving the goal. However, the traditional analytic models failed to take the multiscale or the nonlinear and non-stationary feature. In this paper, the Hilbert-Huang transform (HHT) method and two simple basic models are applied to carry out some multiscale models on the time series of silicon content in hot metal collected from a pint-sized and medium-sized blast furnace. According to the Simulation results, the proposed improved multiscale model exhibits a better behavior in its modeling and prediction.