A step towards implementation of state observers in industrial aluminium smelters

The control algorithm of contemporary industrial aluminium smelters is primarily based on the pseudo-resistance measurement signal and its time derivative. The lack of thorough physical interpretation of this signal leads to frequent instabilities in the form of anode effects. These may increase energy consumption or require operator intervention. To prevent this, observing the smelter’s internal state using the available pseudo-resistance measurement is crucial. In this paper, we establish a robust foundation for designing observers for industrial smelters. Namely, we introduce a modified model incorporating a simple yet precise pseudo-resistance expression that additionally accounts for the pseudo-resistance hysteresis. Then, we explore the model’s observability properties, offering theoretical support for the under/overfeeding control strategy. Additionally, we develop an Extended Kalman Filter and validate its effectiveness through initial testing on a simulator and subsequent application to industrial data.