A neural network model from RUSAL ITC, specially trained on a large dataset of RUSAL's aluminum division's historical data, calculates the chemical composition of the electrolyte in the electrolyzer and adjusts the technological process. It has already undergone industrial testing.
The AI calculates the chemical composition in the electrolyzer based on dozens of available technological parameters: electrolyte temperature, current, voltage, initial chemical composition. This significantly reduces the time required to determine the chemical composition of the electrolyte in the electrolyzer, which, before connecting to the neural network, was carried out using samples over tens of hours.
[...] Measurement for each electrolyzer, considering their number, is possible no more than once every two to three days. Based on the measurement, a conclusion is made about the progress of the technological process and the need to supply additional alumina or fluoride salts to the electrolyzer; timely supply, in turn, allows increasing the speed of aluminum smelting and making the process more efficient. With the introduction of the new technology, the chemical composition will be determined by the neural network, which will supplement rare laboratory measurements with predictive values, will be able to make calculations as often as necessary, and will adjust the electrolysis process much more quickly.
The technology underwent testing and pilot operation at the Sayanogorsk Aluminum Plant for about a year on the latest generation RA-550 electrolyzers, the most advanced model of high-amperage electrolyzers developed by RUSAL. It proved to be effective, and now this AI model is planned to be integrated at all RUSAL aluminum plants.
Read related materials:
New Russian gas turbine engine AL-41ST-25 with a capacity of 25 MW tested