The prediction analysis of failure pressure of pipelines with axial double corrosion defects in cold regions based on the BP neural network

Abstract

The conventional corrosion assessment codes show a certain conservatism in predicting the failure pressure of pipelines with double corrosion defects, which will lead to the failure of reasonable maintenance of pipelines. In addition, when the numerical simulation method is used to predict the failure pressure of the pipeline, the time cost is relatively large. To predict the failure pressure accurately and quickly, ABAQUS and MATLAB were used to analyze the failure pressure of pipelines with axial double corrosion defects in cold regions in this paper, and developed the BP (Back Propagation) neural network for predicting the failure pressure. The results show that the corrosion depth has the most obvious effect on the failure pressure, and the maximum failure pressure loss rate is 42.40%; corrosion length is second, 17.56%; and corrosion spacing is the smallest at 7.16%. Compared with the codes, the failure pressure value predicted by the BP neural network is the closest to the experimental data. The maximum error is 10% and the minimum error is 1%. It proves that it is reasonable to use the BP neural network to predict the failure pressure.