MULTI SENSOR DATA FUSION FOR ALUMINIUM CELL HEALTH MONITORING AND CONTROL

The prevailing aluminium electrolysis process demands steady-state conditions within narrow borders, to improve performance with respect to molten metal production per day, energy usage per kg of aluminium, current efficiency, CO2 and flour-gas emissions etc. However, only the current and the cell voltage are obtained by on-line measurements. Many bath parameters are manually measured on a daily or even weekly basis. Innovating measurements of the bath temperature, the bath chemistry, the molten metal height and the height of the electrolyte would all be of substantial importance for the control regime. However, combining new measurements and soft sensors for estimating "unavailable" variables would improve both the monitoring and controlling tasks of the aluminium electrolysis process. This paper gives an overview of many online and off-line measurements and reports some new possible measurement scenarios with increasing potential for extensive, fast, efficient and even real-time data fusion. Finally some interesting examples of data fusion examples based on actual plant measurements covering many months are also included.