Data-Driven Abnormal Condition Identification and Self-Healing Control System for Fused Magnesium Furnace

Abstract

In the smelting process of fused magnesium furnaces (FMFs), frequent changes in the raw material granule size and impurity constituent will cause the arc resistance between the lower end of the electrode and the surface of the molten pool to vary, and thus, the smelting currents fluctuate. Consequently, abnormal conditions, which can arise if the setpoints of electrode currents are not properly adjusted on time, will causethe performance to deteriorate or even the overall operation to stall. Through analysis of the characteristics of different operating conditions, this paper presents a data-driven abnormal condition identification and self-healing control system. The proposed system extracts the identification rules according to the current tracking error, as well as the rate and duration of the current fluctuations, and identifies the abnormal conditions based on rule-based reasoning. The self-healing control is developed using case-based reasoning to correct the current setpoints based on the identification results. The outputs of the control loop track the corrected setpoints, thereby forcing the process to recover from the abnormal conditions. The proposed method and the developed control system have been applied to a real FMF, and substantial improvement is achieved with many benefits provided to the factory. The implementation results show that occurrence of abnormal conditions has been reduced by more than 50%, and the product quality has been increased by more than 2%.