Residual Strength Of Multi-point Corroded Pipeline Based On Finite Element And PSO-BP

Oil and natural gas are the main components of fuel in the world,and pipelines are an important structure for transporting oil and gas over long distances.Although the pipeline is a very safe way of energy transportation,it also faces many threats,especially corrosion.If the pipeline fails,it will bring huge environmental harm and economic loss.In engineering practice,the corrosion shape mostly appears in the form of irregular group corrosion,and most of the existing methods for calculating the remaining strength,the object is a single corrosion defect pipeline,so the evaluation results are often too conservative.In order to determine whether the pipeline can be in normal service and avoid accidents,it is necessary to study and analyze the failure pressure and the interaction mechanism between the defects with multi-point corrosion defects,and provide a reference for the safety evaluation of the pipeline.The main research contents of this article are as follows:(1)According to the general process of finite element analysis and the actual research problems,the specific parameters of finite element analysis are set up,the finite element analysis model is established,and the failure pressure of multi-point corroded pipeline is analyzed and predicted.The prediction results of the finite element model in this paper are compared with the results of blasting experiment and calculation in literature,and the error analysis shows that the finite element calculation is basically consistent with the real results.(2)A prediction model for failure pressure of multi-point corroded pipeline based on PSO-BP was established.Firstly,based on the collected experimental data of multi-point pipeline blasting with corrosion,the influencing factors of failure pressure were screened out by Lasso regression,and the input variables of BP neural network were determined.Then,PSO is used to optimize the initial weight threshold of BP neural network.Finally,the optimized BP neural network is used to predict the pipeline failure pressure.Through error analysis,it is concluded that the calculation result of PSO-BP neural network has little deviation from the real result,which indicates that the prediction model of failure pressure of multi-point corrosion pipeline established by PSO-BP has high accuracy.(3)The prediction effect of two models of finite element and PSO-BP was compared with that of the pipeline under the same working condition,and the applicability and accuracy of the two methods were compared.The two methods are compared with RSTRENG and DNV-RP-F101.The results show that the predicted values of the two methods are better than the common evaluation criteria.The finite element is more accurate than PSO-BP prediction,but the workload is large,and the neural network can be predicted only by training a certain amount of sample size.(4)Combining the advantages of finite element and PSO-BP model,the multi-point corrosion defect interaction analysis based on finite element and BP neural network was proposed,and the database was built and the neural network was trained by using finite element.The trained network was used to study the interaction between axial and circular double point corrosion.The results show that when the axial double-point corrosion spacing is greater than 2.0 and the annular double-point corrosion spacing is greater than 3.0,the interaction between the corrosion can be ignored.