Research Of Relevance Quantification Of Process Parameters In Aluminum Electrolytic Based On Data Mining

In the traditional control of the aluminum electrolysis, there were many problems, for example, it relied on people's experiences to set quantify of the process parameters, which be with a strong subjective. It does not have full use of the existing the a large number of historical data from the aluminum during the production process, no data be found in the mass inherent in the enterprise production and management has an important role in guiding the rules. In order to promote the production of aluminum, lower production energy consumption, extend equipment life, improve the efficiency of production, data mining technology has been introduced to quantify electrolytic process parameters, and conduct in-depth theoretical and experimental research to identify the law of it and the best quantitative program. The main contents of this paper and innovation are as follows:1. Detailed study the principles, methods, algorithms and applications of data mining technology and regression analysis model.2. Though analyzing the parameters and it's quantify in current situation in the electrolytic production, to draw the best electrolytic process parameters to quantify the program: At first, extracting the aluminum dataset on local. then doing data preprocessing for these dataset. at last, analyzing and mining the dataset to find the best quantization.3. In the process of data preprocessing of missing data, in accordance with the old an new information or it impact, through the interpolation method to fill vacancies ,which in the value of better data on changes in the process of preservation. For the noise data, using sub-boxes, the scope of amendments and clustering algorithm for processing.4. In accordance with the characteristics of aluminum production process, established model of regression analysis. By researching in regression analysis, genetic algorithm, neural network algorithm, proposed to establish a regression analysis model, and used genetic algorithm to carry out mining the historical data of to quantify the program.5. Because of genetic algorithms are good at global search, and neural networks are good at local search, in the paper, it proposed the way of improving the genetic algorithm - integration of genetic algorithm and neural network, namely genetic algorithm neural network algorithms, which is applied to analysis and mining the historical data in electrolytic process parameters. Experimental results show that: genetic algorithm neural network algorithms is better than the least squares method and genetic algorithm.On that basis, designed and developed the data mining system of quantifying the aluminum processing parameters. Using the system in the field of aluminum electrolysis, can effectively guide the stability of cell production, extend the equipment's life, increase productivity and provide a basis for scientific management.