Research And Implementation Of Refinery Equipment Corrosion Prediction Technology

This subject belongs to the enterprises and project subject "research and development on the production equipment corrosion of online monitoring system" (2012-0-1-2012).Corrosion projections for refining equipment is the key and core of in-depth analysis of the corrosion trend, the reliable prediction results can be seen in the trend of corrosion and corrosion protection of equipment, which is of great significance to the safety in production run.Equipment corrosion rate is affected by many different factors, the interaction of various factors, adopts the traditional mathematical modeling method for corrosion prediction has certain difficulty and the insufficiency, and due to the influence factors of equipment corrosion rate basically remain unchanged, corrosion rate has the characteristics of basic leveled off in the long term, ARIMA model and the neural network model based on time series is very good to solve the corrosion rate prediction problem; For the remaining life of the equipment, the failure probability statistical analysis can be very good to solve the problem.First, the research status quo of corrosion prediction is summarized in this paper, comparing the advantages and disadvantages of different model algorithm, based on the characteristics of corrosion prediction problem and monitoring data, in view of the refining equipment corrosion rate, using ARIMA model is put forward, and the gradient of RBF neural network based on time series models to forecast; In view of the residual life equipment, adopt the Monte-Carlo method for residual life assessment; The above mentioned problem the problem description and model solving method are given respectively.The focus of this article is on the corrosion rate and residual life prediction, using historical data gathered from the scene, combined with the characteristics of corrosion problems, adopt ARIMA model and the gradient of RBF neural network corrosion rate is forecasted, and comparing with the result of two kinds of models to predict, the results show that the gradient prediction effect of RBF neural network model, which is better than that of ARIMA model prediction effect. On this basis, the Monte-Carlo method is used to evaluate the residual life of equipment in this paper, through to the equipment corrosion failure probability statistics, judge the remaining useful life of equipment, and equipment residual life.Finally, the oil refining equipment corrosion prediction system has carried on the system requirements analysis and design, realized the corrosion rate prediction system based on RBF neural network. Research results can be effective for corrosion prediction, the reliable prediction results can be helpful to the staff timely judge corrosion status of equipment, thereby making good protective measures, reducing or eliminate equipment leakage or failure.