The Research And Application Of High Utility Pattern Mining On Quality Analysis Of Alumina Clinker

Alumina clinker quality analysis occupies an important position in the production of alumina. As the temperature of alumina clinker kiln depends on totally, the quality analysis of alumina clinker has a great impact in its yield and quality. Now the way of alumina clinker quality analysis not only has a serious threat to the safety and health of the staff, but also has the problems of justify reliance on the experience and the lack of scientific. Based on the relationship between the sound of clinker falling and clinker quality, we designed and implemented the clinker quality intelligence analysis system, used to improve the accuracy of clinker quality judgment. And with the help of analysis on sound and clinker quality rules, alumina clinker quality analysis can be carried out safely and accurately, while improving the yield and quality of alumina clinker.Through many researches on the advanced mining algorithms, we designed a new high utility itemsets mining algorithm, called CHUI-Mine, in this paper. The performance research has done on both synthetic datasets and real datasets. The correctness of the algorithm justifies with theoretical analysis. CHUI-Mine algorithm is implemented with a new type of tree data structure CHUI-Tree. We proposed the concept of dynamic pruning in the algorithm, which implement the method of mining while constructing. When read the transactions in database and construct the CHUI-Tree, prune it and mine the conditional pattern once necessary. The method can greatly improve the efficiency of the algorithm and reduce memory usage.This paper designed and implemented a set of intelligent analysis system for the analysis of alumina clinker quality. The system collects the sound data and then preprocesses it according to the characteristics of the sound. We find the relationship between the sound data and the quality of clinker, as the basis to judge the quality of the clinker. The processing results of real-time feedback to staff. Then high utility pattern mining method is used to analysis the rules. The system not only improves the accuracy of clinker quality judgment, but also provides a scientific basis for quality analysis.