Research On Corrosion Analysis And Prediction Method Of Atmospheric And Vacuum Unit Tower Based On Data Mining

With the decline of crude oil quality and the aging of refining equipment and many other factors,leakage problems caused by equipment corrosion and unplanned shutdowns of equipment frequently occur,and the losses caused are incalculable.As the "leading" device of the refining and chemical enterprise,the atmospheric and vacuum unit is more susceptible to corrosion by the corrosive medium in crude oil,especially the core tower equipment of the entire unit.Therefore,the corrosion analysis,prediction and control of the atmospheric and vacuum unit tower can not only reduce the probability of corrosion failure of the atmospheric and vacuum unit,but also greatly reduce the corrosion risk of subsequent devices in the entire process.With the accumulation of a large amount of corrosion monitoring and inspection data,the use of data mining technology to analyze and predict corrosion,and the study of automated and intelligent corrosion supervision methods are the new direction and trend of equipment corrosion management.Based on the big data of atmospheric and vacuum equipment corrosion,this article first uses statistical analysis methods to summarize and organize the corrosion failure cases of the atmospheric and vacuum equipment tower equipment,analyze and summarize the common types of corrosion failures,high corrosion locations and common causes of corrosion.Based on K-Means cluster analysis method,the crude oil properties are classified and the analysis of variance is used to explore-different types of crude oil.Corrosion status of equipment;subsequently,based on Principal Component Analysis(PCA)method and BP artificial neural network,K Nearest Neighbor(KNN),Support Vector Machine(SVM)classification and prediction methods,the corrosion status prediction of the tower top loop of the atmospheric and vacuum unit was established respectively Model and tower equipment corrosion degree prediction model.The results show that the PCA-SVM model has the best effect in predicting the corrosion state of the tower top loop,and the recognition rate can reach 96.552%.PCA-KNN model is used in the prediction of tower equipment corrosion degree.The effect is the best,and the recognition rate can reach 94.737%.Finally,according to the research content and research results of this article,a corrosion supervision method driven by corrosion prediction is proposed,and the result characterization,fortification strategy,corrosion diagnosis,corrosion prediction and optimization decisionmaking are proposed.Comprehensive application method.Combining traditional corrosion diagnosis methods with data mining technology has certain innovative significance and practical industrial application value,which is conducive to promoting the safe operation of atmospheric and vacuum equipment and the development of corrosion protection management in the direction of digitization and intelligence.