Research on Prediction of Remaining Life of Internal Corrosion of Oilfield Water Injection Pipelines
The life of buried pipelines directly affects the economic benefits of oil and gas pipeline companies,and accurate prediction of the remaining life of buried pipeline corrosion can make maintenance plans in advance and reduce economic losses.In order to estimate the remaining safe service life of the pipeline,a Least Squares Support Vector Machine(LSSVM)prediction model based on Principal Component Analysis(PCA)and Particle Swarm(PSO)combined with Ant Colony(ACO)Hybrid Continuous Optimization Algorithm(HCACO)was created.Firstly,the main influencing factors of pipeline corrosion were extracted by PCA dimensionality reduction to optimize the input variables of the prediction model.Secondly,HCACO was used to optimize the penalty factor and kernel function pa-rameters in LSSVM,and the optimized parameters were substituted into the LSSVM prediction model.Finally,the remaining life prediction model of corroded pipelines based on PCA-HCACO-LSSVM was constructed.Taking the water injection pipeline of an oilfield as an example,and comparing it with the other three models BP,SVM and the current popular GRA-XGBoost,it is concluded that the root mean square error(RMSE),mean absolute error(MAE)and goodness-of-fit(R2)of the PCA-HCACO-LSSVM model were 0.303,0.175 and 0.993.The evaluation index values of the model are better than those of the other three models,the prediction ac-curacy and generalization performance have also been significantly improved,and the predicted value of remaining life is closer to the actual value,which opens up a new way for the anti-corrosion research of oil and gas pipelines.
oilfield water injection pipelinesinternal corrosionremaining life predictionprincipal component analysis(PCA)hybrid continuous ant colony optimization(HCACO)least squares support vector machine(LSSVM)
万方数据
维普
