Research On Application Of Association Rule Mining To Blast Furnace Situation Prediction

Long-term, stable and smooth state of the blast furnace is not only the precondition of high yield and low consumption, but also the prerequisite to extend the life of blast furnace. BF iron making is carried out in a closed container, in which the physical, chemical process is extremely complex, the furnace state condition can not be directly observed. To reduce energy consumption in the blast furnace smelting process and achieve the stability and optimization of iron making production demand timely and accurate judgment and forecast to the status of blast furnace, adopt various adjustment measures in a timely and appropriate manner. So data mining is used to scientifically analyze and predict the blast furnace situation, which has become an important research. Because of the complexity of the process, and association rule mining technology based on its so many advantages, the blast furnace receives more and more attention from the scholars at home and abroad.On account of the fitness of time is seldom illustrated by traditional association rules. Temporal association rules are improved and used for the blast furnace situation prediction. Weighted temporal association rule is presented in this paper based on these researches, which can reflect the time value of data and the time tendency of discovered rules. The simulation experiments demonstrate that approach is fairly effective.Other results are as follows:①Firstly problem related to the blast furnace situation prediction is reviewed, then analyzes the lack of correlation of existing algorithm, finally, a multidimensional association rule approach in the blast furnace situation prediction is proposed.②According to the actual operation of the blast furnace, the data source is selected, and done attribute reduction. The relevant property is selected to constitute a prediction-related data sheets. The dimension of association rules are reduced, the number of redundant records is reduced, and then valuable and efficient rules are produced.③The theory of the use of weighted, in the dynamic update of the database when the state of affairs happened close to give greater weight to reflect the value of time, as far as possible to mining the latest and most useful temporal association rules.④According to the previous analysis, a weighted multidimensional temporal association rules algorithm based on attribute reduction is put forward. The main idea is as follows: Firstly, do attribute reduction to form a table, subsequently, introduce the improved algorithm to mine the rules of the blast furnace situation prediction.⑤The experiment conducted by Visual C++ programming, select the accuracy and coverage prediction model to measure the effect, by changing the support and confidence values, enhance the efficiency of mining. Then the proposed algorithm to simulation and assessment shows that new algorithm is more efficient in blast furnace situation prediction compared with the traditional association rules.