Identification of Extreme Temperature Fluctuation in Blast Furnace Based on Fractal Time Series Analysis

Abstract

In this study, we aim to estimate the density distribution for the return intervals of extreme temperature fluctuation in blast furnace during iron making process. We first identified the fractal feature of the data based on R/S analysis and also calculated the Hurst coefficient. Secondly, based on the fractal feature of the data, we estimated a stretched exponential distribution of the return intervals of extreme temperature fluctuation. Finally, in order to test the result, we applied this method to the data of two blast furnaces, and compared with the traditional kernel density estimation method. The comparison was based on 100,000 times K-S test. The comparison results showed that the fractal time series estimation provides a greater fitness than traditional estimation method since it has no rejection in K-S test. With this method, the density of return intervals of unexpected temperature fluctuation can be estimated. This can be applied as a tool of temperature control and also can be applied as a tool to evaluate the efficiency of the control system.