Multi-objective optimization of aluminum electrolysis based on improved SHAP value

Aiming at the problem that many parameters affecting each other in the process of aluminum electrolysis make it difficult to improve the current efficiency and unable to reach the expected production index, the multi-objective optimization idea is proposed to optimize the current efficiency as the primary objective, and optimize the other important parameters that have a large impact on the current efficiency and affect each other as the secondary objective. Firstly, the variance threshold and the improved SHAP value method are used to extract the important features such as aluminum level and electrolyte level. Then the prediction model and MOTROA multi-objective optimization algorithm are used to establish a multi-objective optimization model for the aluminum electrolysis production process to obtain the optimal ranges of parameters such as aluminum level, electrolyte temperature, and so on, and to achieve the goal of improving the current efficiency. The experimental results show that these important parameters are located in the optimal range while the current efficiency is increased.