Prediction of the laws of carbon steel erosion corrosion in sour water system based on decision tree and two kinds of artificial neural network model

To explore the corrosion induced failure problem of carbon steels commonly existed in sour water system, first, erosion failure database is built on the basis of self-built rotary erosion experimental device. Furthermore, a decision tree based erosion level prediction model is proposed. At the same time, by means of two kinds of neural network, erosion rate prediction model is also proposed based on carbon steel erosion experimental samples: first, self-organization mapping (SOM) network is firstly applied to obtain the relevant relationship between variables by the explorative clustering analysis of multivariate samples. Then error back propagation (BP) neural network is adopted to model and predict corrosion rate of carbon steel samples. The test results show that the prediction accuracy of the decision tree model can be 100% and the average error the BP neural network model applied in this paper can be as low as 3.63%, which provides a new method for material selection and real time corrosion prediction and control in petrochemical system.