Aluminum Electrolytic Cell Health Evaluation Method And System Based On Combinatorial Weighted Naive Bayes

In the production process of aluminum electrolysis,there will be differences in the operating state of aluminum electrolytic cells,and the health status of the electrolytic cells will also be inconsistent.Unhealthy aluminum electrolytic cells have high energy consumption and low aluminum productivity,seriously effecting the overall production efficiency.At present,technicians mainly judge the operating state of electrolytic cells by knowledge,which is subjective and one-sided.In view of the above deficiencies,from the perspective of knowledge and data-driven,a research on the health evaluation of aluminum electrolysis cells is carried out in this paper,So as to assist technicians to make rational decisions to improve production efficiency.First,considering that the voltage vibration and swing are two important basis for health evaluation,in order to extract state features of voltage vibrations and swings,a classification method based on SMOTEMVS-CNN-BiLSTM for voltage vibration and swing is proposed in this paper.On the one hand,to solve the issue of data imbalance,according to the features of vibration and swing data,the SMOTEMVS is proposed to balance the dataset;On the other hand,to solve the issue of classification based on time series data,a CNN-BiLSTM network model is established to extract features of vibration and swing data and classify,and the effectiveness of the method is proved by experiments.Then,in order to evaluate the health status of aluminum electrolytic cells comprehensively and accurately,combined with the state features of vibrations and swings,other attribute features,mechanism knowledge and expert knowledge,a hierarchical health evaluation index system is established,and a health evaluation method of aluminum electrolytic cell based on combinatorial weighted Naive Bayes is proposed in this paper.The model is modified by introducing the value of the data obtained from knowledge,and the problem that the traditional method of only using the information of the data is prone to deviate from the actual situation due to missing data or other situations is solved,which enhances the interpretability of the method.Experiments show that health judgment of a large number of electrolytic cells can be realized by this method in a relatively short time,and this method is suitable for industrial applications.Finally,combined with the research work of this paper,a data analysis and intelligent decision-making system for aluminum electrolysis industry is developed.In this system,the health status of a large number of aluminum electrolytic cells can be judged quickly and accurately,and maintenance plans for aluminum electrolytic cells based on the current economic situation and the health of the aluminum electrolytic cells can be formulated.Figures(43),Tables(10),References(74)...