An attention-based long short-term memory prediction model for working conditions of copper electrolytic plates

Copper is an important source of non-ferrous metals, with electrolytic refining being one of the main methods to produce fine copper. In the electrolytic process, plate states seriously affect the output and quality of the copper. Therefore, timely and accurate prediction of the working states of the plates is of great significance to the copper electrolytic refining process. Aiming at the issues associated with traditional plate state detection algorithms of large lag, poor anti-interference ability and low accuracy, a plate state prediction model based on a long short-term memory (LSTM) neural network with an attention mechanism is here proposed in this paper. The average gray values of the plates in infrared imagery are used to characterize the plates' working states. To address the problems of large fluctuation and the large amount of time series data required in such a study, a double-layer LSTM neural network structure is used to improve the efficiency and accuracy of model training. Meanwhile, in view of the periodicity of the time series data and the possible correlation between adjacent data, a unique attention mechanism is proposed to enable the model to learn this correlation between the adjacent data so as to improve the accuracy of the model prediction. The experimental results show that the accuracy of the proposed model for plate state prediction reaches 95.11%. Compared with commonly used prediction algorithms, the plate state prediction model proposed in this paper demonstrates stronger prediction ability.