Research on primary aluminum production scheduling optimization based on expert system and long-short-term memory network (LSTM)

This article studies the influence of the material and thermal balance of the electrolytic cells in the aluminum production. Specifically, these two metrics reflect the current efficiency, material utilization rate and the lifespan of the electrolytic cells. Therefore, reasonable planning of the amount of siphon aluminum(yield) and production time are two key factors in the aluminum electrolysis process. However, due to the occurrence of numerous emergencies in the actual production process, manual scheduling aluminum production is still a major challenge, which motivate us to develop an LSTM of the rule-based expert system aluminum production decision-making model. First, the input dataset are processed and selected according to the expert system. Then, a deep LSTM model is built. Finally, the output results are planned and maintained based on the expert system. The performance in the experiments also proves that the method is feasible in terms of both accuracy and robustness aspects.