Prediction of blast furnace operation using on-line bayesian learning

The large scale database-based online modeling, called LOM, is a type of just-in-time modeling for blast furnace. In this paper, we propose a new type of LOM using a nonlinear local model to improve the performance of the long-term prediction. To estimate the parameter of the nonlinear local model, we use on-line Bayesian learning scheme with sequential Monte Carlo. The prediction performance of the new LOM is demonstrated by using the real process data of blast furnace.