Font Size: a A A

Research On Evaluation And Analysis Of Temperature Heating Curve Of Anode Baking

Posted on:2019-01-04Degree:MasterType:Thesis
Country:ChinaCandidate:Y N WangFull Text:PDF
GTID:2321330545990057Subject:Computer technology
Abstract/Summary:PDF Full Text Request
Baking is an important step in the production process of anodes used for aluminum electrolysis.The baking heating curve has an important influence on the physicochemical properties of prebaked anodes.Therefore,scientific and effective baking heating curve evaluation methods are urgently needed in the production process of anode blocks.However,the application of the algorithm model in this type of problem is rarely.In this paper,by studying the relationship between the anode baking temperature curve and the quality grade of baked anode block,an optimized RCNN-K model is proposed to realize the evaluation of the temperature curve.The specific research contents are as follows:(1)On the basis of studying the quality standard of baking block,the evaluation standard of baking block was formulated.The quality of the anode carbon block for aluminum electrolysis is mainly measured by seven parameters.Each parameter has its own physical and chemical performance indicators.This paper provided a conversion method of anode carbon block evaluation indicators according to the user,a number of anode carbon block quality parameter evaluation index is converted into a quality grade evaluation.According to the structure of the roaster,the flue temperature on both sides of the baking block is used as the temperature data features.Based on the characteristics of the anode baking temperature data which is related to time series,the differencing process in the time series prediction model ARIMA is applied to expand the temperature data features.(2)The RCNN model is attempted to be applied to the anode baking field and optimized.For the problem that the sub-sampling of the RCNN model caused a large amount of information loss,the original structure is improved,and an optimization scheme RCNN-K is proposed.Based on the number of heating stages,max pooling is used on the time dimension features,and the new convolution layer and pooled layer are added.The features of the same time step is convoluted,features are further extracted,and compared with the original algorithm,the validity of the model is verified based on the experimental results.(3)Design and implementation of anode baking temperature curve evaluation and analysis system.The system mainly includes parameter processing,evaluation model building and evaluation result analysis module.The above classification model is applied to the task of evaluation of anode baking temperature curve.The experiment is carried out through the data of the baking temperature and the quality parameters of baking block.And the experimental results are displayed in the form of charts.
Keywords/Search Tags:prebaked anodes, heating curve evaluation, recurrent neural network, convolution neural network, evaluation index transformation
PDF Full Text Request
Related items