Research On Prediction Model Of Pipeline Corrosion Rate Based On Grey Theory And Neural Network

The pipeline is a very important connection for the operation of the refinery.Therefore,in order to ensure the smooth operation of the device,it's necessary to strengthen the corrosion analysis and management of the pipeline.There are so many factors affecting pipeline corrosion and they are characterized by ambiguity and also characterized by randomness.A complex correlation can been found between these factors.With the continuous development of the refinery industry and the continuous increase of the refining process,the previous pipeline corrosion diagnosis and prediction methods are gradually unable to solve the problems that occur in pipeline corrosion analysis today.The intelligent data analysis algorithm has a good linear mapping function,which can process multiple data in parallel and solve the problem that the correlation between multiple factors cannot be determined in traditional methods.Therefore,using data analysis and excavation methods to predict the corrosion rate of pipelines is a key improvement direction for pipeline corrosion control and corrosion remaining life prediction.Methods of literature review and on-site corrosion inspection have been used in this paper in order to obtain data and basic information and analyzed the corrosion factors and damage mechanism of the obtained pipeline corrosion data.On the basis of pipeline corrosion inspection,the intelligent data analysis algorithm has been used to construct the PSO-MGM(1,1)model and the PCA-GA-BP model to predict the pipeline corrosion rate.Combined with the actual situation of the petrochemical atmospheric device pipeline,the PSO-MGM(1,1)model and PCA-GA-BP model have been analyzed and verified by examples.Through model reliability verification,both models can predict pipeline corrosion rate well.Based on the analysis of model characteristics,the PSO-MGM(1,1)model is more suitable for the prediction of corrosion rate of a single pipeline with a small amount of data and the PCA-GA-BP model is more suitable for the prediction of corrosion rate of pipelines with a complete set or a large amount of data.Based on the PSO-MGM(1,1)model prediction,the corrosion evaluation of the pipeline has been performed,the remaining life prediction of the pipeline corrosion and the RBI quantitative analysis of the pipeline are calculated and combined with the corrosion evaluation results,effective risk inspection and management measures are proposed.In this paper,the intelligent data analysis algorithm has been introduced into the corrosion rate prediction of refinery pipelines and the analysis and verification of a petrochemical atmospheric pressure pipeline have been used as examples,which can provide theoretical methods and technical basis for corrosion damage analysis and corrosion rate prediction of pipelines of similar facilities.