The Research Of Sequential Pattern Mining And Rule Extraction On Huge Volume Daily Data In Aluminum Electrolysis

In electrolytic production process, the computer control technology has been popularizaed, and there are plenty of raw data from real-time monitoring and records, these data contain potential valuable information for aluminum electrolysis process. Data mining is to extract meaningful information or patterns from huge volume of data. Therefore, using data mining method in the aluminum production process could be useful for discoverying valuable information, and guiding the production.Firstly, in this paper, based on the analysis of the various clustering method, the clustering method for electrolytic cell is proposed. Based on sequence comparability, this method provides the ability to adjust the final clustering results. Experimental results show that this method can effectively reflect the technical characteristics of the same kind of cells.Secondly, feature extraction method based on electrolytic cell class is proposed. Based on the characteristics of clustering parameters, the concept of electrolytic cell optimal sequence and the mining algorithm are proposed. Experimental results show that the optimal sequence mining can find the key interval of large-scale data.Thirdly, there is further discussion about daily data sequence segmentation method based on the optimal sequence. The segmented sequences can not only show the trend, but also retain the fluctuation characteristics of the original sequence.Finally, on the basis of above studies, Aluminum daily data mining system is realized and applied in real production control, which is not only an effective tool for technical staff to analyze the characteristics of electrolytic parameters, but also provides scientific guidance for the management of production conditions and trends-forecasting.