Research On The Prediction Model Of Residual Strength Of Corrosion In Pipe Girth Weld Based On Hybrid Intelligent Algorithm

The transportation of oil and gas resources is often carried out with the help of largediameter pipelines,and in practice,long-distance pipelines are connected together by welding technology,which will cause changes in the material properties of the pipe weld position.Affected by the internal medium of the pipeline and the external environment,there are different degrees of corrosion defects,and the girth weld position is more prone to corrosion than the base metal due to its uniqueness,thus becoming a weak link in the oil and gas transportation process.Therefore,the prediction of the residual corrosion strength of the girth weld of the defective pipeline can clarify the operation status of the pipeline,which is of great significance to the safety of pipeline operation.Based on the combing and analysis of the corrosion principle and type of girth weld,using Spearman correlation analysis,combined with WOA algorithm,FA algorithm,SMA algorithm and GRNN neural network,the IWOA-GRNN model and IFASMA-GRNN model were constructed to explore the prediction efficiency of the improvement direction of the two commonly used algorithms.In the first part,the existing evaluation criteria such as ASME B31G-2012,DNV RP-F101,PCORRC,GB/T 19624-2019 and other influencing factors of pipe girth weld corrosion are screened,and the correlation between different influencing factors and residual strength is clarified by Spearman correlation analysis.In the second part,taking the parameter improvement method as the idea,the population is initialized by Cubic chaotic mapping,the change mode of convergence factor is adjusted to nonlinearity,the adaptive weight factor is increased to improve the WOA(whale optimization algorithm),and then the improved IWOA is used for smoothing factor optimization,so as to construct the residual strength prediction model of corrosion of the girth weld of IWOA-GRNN pipeline.Divide the experimental data into training set and test set to train and test the model;In the third part,taking the algorithm fusion method as the idea,keeping the Cubic chaotic mapping initialized population unchanged,the FA(firefly optimization algorithm)and SMA(slime mold optimization algorithm)are fused to form the IFASMA optimization algorithm,and the remaining strength prediction model of IFASMA-GRNN pipe girth weld corrosion is constructed.The same data set is used for model training and prediction,the prediction effect of the two improvement ideas is compared,and the actual situation applicable to different improvement methods is analyzed.The feasibility and accuracy of the constructed model are verified by actual cases,and the corrosion maintenance strategy of girth weld is proposed based on the actual working conditions and prediction results to improve the work efficiency.Compared with the conservative evaluation criterion of prediction results,the two sets of pipe girth weld corrosion residual strength prediction models constructed in this paper have good performance and have certain accuracy and feasibility.Due to the fact that the detection of corrosion defects in the girth weld and the blasting test are less frequent and the time is limited,the influence of more comprehensive coupling factors on the residual strength of the girth weld is the direction of further research in the future.