Corrosion Analysis And Corrosion Prediction Technology Research Of Typical Crude&Vacuum Distillation Unit

With the development trend of crude oil inferiority and large-scale refining equipment,tThe safety problems caused by corrosion failure of pipelines and equipment are increasing,and the consequences are becoming more and more serious.Therefore,corrosion protection is more important in preventing production accidents and improving equipment safety and reliability.With the advent of the "big data" era,corrosion prediction based on a large amount of production data to guide the current anticorrosion work is becoming an important development direction of corrosion research.This article is based on the corrosion inspection results of the atmospheric and vacuum equipment of a petroleum refinery in China and combined with actual production Research on corrosion analysis and corrosion prediction technology was carried out.Firstly,the process flow,material circuit and main corrosion mechanism of the atmospheric and decompression device were studied.The equipment and pipeline were divided into 5 corrosion circuits,and on the basis of this,corrosion analysis based on the corrosion circuit was carried out and determined based on the corrosion inspection results 2 key corrosion sites.Secondly,the 2 key corrosion sites were screened out from the corrosion inspection results for further corrosion mechanism research,corrosion morphology analysis and corrosion product scale sample analysis,what's more,Improvement measures and anti-corrosion suggestions were put forward,and online monitoring layout plans were designed for key corrosion sites.Then,the principle and training process of artificial intelligence algorithms such as BP neural network,Elman neural network and genetic algorithm are studied,and the applicability,advantages and disadvantages of the three algorithms as well as the principles and problems to be followed when designing the algorithm are summarized.The data collected through the corrosion online monitoring system is preprocessed to provide a theoretical basis and data support for the establishment of corrosion prediction models.Finally,three corrosion rate prediction models including the BP neural network model,the Elman neural network model and the BP neural network model based on genetic algorithm optimization have been used to achieve the effect of corrosion prediction.Through model training simulation verification and model comparison analysis,the BP neural network model based on genetic algorithm optimization has the best fitting effect and stability and performs best in corrosion prediction for key corrosion sites.