Two-layer fault diagnosis method for blast furnace based on evidence-conflict reduction on multiple time scales

In a blast furnace (BF), faults are difficult to be diagnosed because of the severe environment inside a BF. Most proposed methods for BF fault diagnosis just diagnose states of a BF by short-time-scale data. However, the probability of faults is also related to the long-time-scale running state of a BF. This paper presents a two-layer fault diagnosis method for BFs on multiple time scales. First, the deterioration trend of a BF is analyzed and the deterioration degree is calculated on a long time scale. Then, an improved Dempster-Shafer (D-S) evidence method is designed to diagnose the faults of the BF on a short time scale. The conflicts of evidence are reduced by incorporating the deterioration degree. The paper explains how D-S theory can be extended to reduce the conflicts of evidence by incorporating the deterioration degree of a BF. The improved D-S method accounts for the different time scales compared with the conventional one. Finally, experiment results based on real running data show that the method reduces the conflict of evidence and improves diagnosis accuracy.