To enhance the quantitative accuracy of pipeline magnetic flux leakage (MFL) detection for pipeline defect depth and precisely control the external corrosion state of long-distance pipelines,an experimental platform of pipeline MFL signal acquisition was constructed. The pipeline MFL internal detection pulling experiments were conducted,and 120 groups of triax-ial MFL signals of internal and external pipeline defects were extracted. A prediction model on the defect depth of pipeline MFL detection based on AOA-XGBOOST was established,and BPNN,SVR,and XGBOOST models were used as the control group for verification calculation. The results show that the AOA-XGBOOST model exhibits better accuracy and superiority for the quantification accuracy of the internal MFL detection signal,which can resolve the quantification issue of internal MFL de-tection signal forthe pipeline defect depth,and effectively enhance the detection accuracy of pipeline state. The research re-sults can provide technical reference for the intelligent analysis of pipeline MFL detection signals.
关键词
油气管道/漏磁检测/缺陷深度量化/机器学习
Key words
oil and gas pipeline/magnetic flux leakage detection/defect depth quantification/machine learning
维普
万方数据