A method for forecast and auto pre-eliminate the anode effect based on spectrum analysis in the aluminium electrolytic cell

In general, the timely diagnosis of the anomalies occurring in the aluminum electrolysis process and prompt action are important issues in increasing the productivity of the electrolysis series and extending the life of the electrolysis furnace. The methods of diagnosing anomalies used in electrolytic processes include expert knowledge of electrolytic processes, electrochemical and thermodynamic modeling, expert diagnostics using artificial intelligence, regression analysis using mathematical statistical techniques, diagnosis using correlation analysis models, and spectral analysis of time series data. However, when diagnosing using these methods, it is difficult to construct the model as the same as the real object, and thus there is some error in the accuracy of the diagnosis results. Therefore, a method for diagnosing anomalous states that can be used to improve accuracy by using parameters such as the electrolytic voltage directly affected by the anomalous state occurring in the aluminum production process is required. One of the effective signal analysis methods, the spectral analysis method, is the most effective method because it can analyze the changes in the frequency domain, which do not occur in the time domain of the signal, by converting the signal from the time domain to the frequency domain by calculating the spectra of the time series of the electrolytic process. The introduction of spectrum-based diagnostic methods in aluminum control systems is important for increasing diagnostic accuracy and ensuring the reliability of control systems. In this paper, we present a method for predicting the anodic effect by performing spectral analysis of the cell voltage and series current signals, which are signals that can be detected directly in a self-anodic aluminum electrolyzer. In addition, we propose a control system configuration to auto-rectify the anodic effect previously using compressed air and verify the effectiveness of the proposed method through simulation.