Research On Visualization Analysis Technology Of Aluminum Electrolysis Big Dat

The production process of aluminium electrolysis generates a large amount of data,which contains a lot of valuable information as the raw material for creating value in Industry 4.0.Effective analysis of this data and information can improve production efficiency.However,aluminium electrolysis data is difficult to analyse because of its large scale,variety,fast flow and low value density.Therefore,this paper takes aluminium electrolysis production data as the research object,visualisation technology design as the research objective,visualisation and analysis technology of abnormal and correlation of basic parameters,decision-making method of key parameters,and overall visualisation and analysis technology of key parameters as the research content,aiming to intuitively present the value behind aluminium electrolysis big data and help aluminium electrolysis enterprises to make scientific and effective decisions.The main work of this paper includes:(1)To address the problem that traditional methods cannot effectively analyse the basic parameters of aluminium electrolysis,various visualisation techniques are designed and used to analyse the anomalies and correlations of the parameters.Firstly,seven visualisation techniques are designed to analyse the characteristics of single and multiple parameters from the static analysis level.Then,from the level of dynamic analysis,the monitoring screen is designed to realise multiangle and multi-view collaborative dynamic analysis.Finally,a variety of algorithms and visualisation techniques are combined to initially deal with the abnormalities reflected in the analysis results and to explore the correlation between parameters.(2)To address the problem that human beings cannot make scientific decisions on key parameters of aluminium electrolysis in a timely manner,two machine learning algorithms,ID3 and C4.5,are improved and used to build a decision-making method that combines expert experience and machine inference.Firstly,the original ID3 and C4.5 algorithms are improved by introducing AF correlation mechanism and Gini index to solve the problem of low computing efficiency and poor decision making when the original algorithms analyse large data of aluminium electrolysis.Then,the improved algorithm was trained to make decisions on the values of aluminium output,electrolysis temperature and mole ratio using the data after correlation analysis.Compared with the original algorithm,the highest correct decision rates of the improved algorithm were improved by2.12%,1.43% and 1.39% respectively.Finally,a tree diagram was designed to visualise and analyse the decision patterns formed.The experimental results show that business managers are able to make scientific and effective decisions on the key parameters of aluminium electrolysis production by this method.(3)To address the problem that existing visualisation techniques cannot effectively analyse multiple different key parameters of aluminium electrolysis,the parallel coordinate visualisation technique is improved to analyse the overall patterns of multiple processed key parameters in one visualisation technique.Firstly,the KPCA dimensionality reduction technique and spectral clustering technique are purposefully introduced to solve the problems of low visibility and inconspicuous patterns when visualising large data of aluminium electrolysis by the original technique.Then,the integration of axis sorting technology and Bessel curvilinearization technology enhances the aggregation and visual continuity of the folds.Finally,interactive techniques are integrated to analyse the overall patterns of several key parameters.The experimental results show that the proposed method improves visualisation,increases the efficiency of managers in analysing and capturing the overall patterns of data,and enhances the ability of companies to explore the value behind aluminium electrolysis big data.