Interval prediction model of blast furnace gas utilization rate based on multi-time-scale

Abstract

Gas utilization rate (GUR) is an important indicator that reflects the state of a blast furnace (BF). However, most researches only predict the point values of GUR, it is difficult for the operator to make corresponding operations. This paper presents an interval prediction model based on multi-time-scale to predict the GUR. First, this paper analyzes the impact of the burden distribution and the hot-blast supply on multi-time-scale of the blast furnace. Then, we build a multi-time-scale point prediction model based on support vector regression (SVR). Next, an interval prediction model of multi-objective optimization based on interval prediction indicators and the point prediction model was proposed. Finally, some experiment results base on actual run data shows that the method predicts the GUR more effectively than the point prediction model based on single time scale.