A Deep Learning-Based Optimization Method for Electrolyte Regional Feeding Strategy

Abstract

This invention relates to the field of aluminum electrolysis technology and provides a method for optimizing the feeding strategy of an electrolytic cell based on deep learning. The method includes the following steps: acquiring current data of the aluminum electrolytic cell; acquiring energy consumption data and alumina feeding data of the aluminum electrolytic cell; dividing the aluminum electrolytic cell into feeding areas according to the feeding mechanism; extracting features from the current data, energy consumption data, and alumina feeding data using a deep learning model; generating a first feeding strategy for the aluminum electrolytic cell based on the current data features output from the current data path; and evaluating the reliability of the first feeding strategy and optimizing the feeding strategy based on the energy data features output from the energy data path and the feeding data features output from the feeding data path. This invention achieves the perception and multi-objective collaborative optimization of the feeding area by acquiring the current, energy, and feeding data of the aluminum electrolytic cell and constructing a deep learning model, thereby optimizing the feeding strategy for large aluminum electrolytic cells.

Images