Intelligent Diagnosis Method for Girth Weld Quality of Natural Gas Pipeline Based on Multi-source Heterogeneous Data
The quality diagnosis of natural gas pipeline girth weld mainly relies on the segment-by-segment excavation verification method guided by the manual detailed analysis of girth weld data,leading to high operating costs and low accuracy & efficiency and can't meet the needs of safe and efficient operation and maintenance of modern pipelines.It is of great engineering significance to ensure the service safety of welded pipeline segments using efficient intelligent quality diagnosis methods to predict the quality of weld in buried natural gas pipelines and carrying out targeted excavation review work.The paper constructs an artificial intelligence method to quickly diagnose and predict the quality of natural gas pipeline girth welds as well as solve these above difficulties,uses the variance selection method and the correlation coefficient map to select 15 main features of the ring weld,including the length of the pipe section,the unit,the form of the counterpart,and the inter-station section to improve the computational efficiency of the model,establishes an intelligent diagnostic model for the quality of gas pipeline girth welds based on the random forest method,optimizes the parameters using a Bayesian optimizer,validates the model using untrained engineering data,and compares the performance with support vector machine,decision tree,and K-nearest neighbor algorithms.The results show that the quality diagnosis of girth weld engineering data using the random forest model has a high accuracy rate,and can provide theoretical guidance for the excavation and exclusion work of girth welds of in-service natural gas pipelines.
万方数据
维普
