Research on Residual Life Prediction of Electrolytic Cells Based on Informer

The lifespan of aluminum electrolytic cells directly affects the economic benefits of electrolytic aluminum enterprises. By predicting the remaining lifespan of aluminum electrolytic cells, maintenance plans can be formulated in advance, thereby reducing economic losses. In this paper, a large number of data generated in the production process of aluminum reduction cells are processed with outlier, missing values, normalization, etc., and the data are dimensionally reduced using Kernel principal component analysis. The remaining service life of aluminum reduction cells is predicted using Informer model. The experimental results show that the Informer model has higher prediction accuracy than the Long short-term memory network and time convolution network, and can more accurately predict the remaining life of aluminum reduction cells, which has an important guiding significance for the management of aluminum reduction cells.

I. Introduction

The electrolytic aluminum industry plays an important role in the modern economy, not only providing important raw materials for various fields such as construction, transportation, and electricity, but also promoting national industrialization and economic development. As the core equipment of industrial aluminum smelting, aluminum electrolysis cells are prone to various faults during long-term high load operation, which can affect actual production activities. Therefore, accurately predicting the remaining useful life (RUL) of aluminum electrolysis cells is beneficial for reducing equipment maintenance costs, optimizing production processes, improving production efficiency and safety, ensuring the smooth completion of production goals, and improving economic benefits for the enterprise.