| Summary: | Welded defects in gas polyethylene pipelines exhibit low occurrence frequency and high variability, and the traditional attribute measurement analysis method fails to assess their performance comprehensively. A two-tiered progressive attribute measurement analysis model is proposed. Initially, key statistical measures for assessment agreement (Pa, Cohen’s Kappa, and Fleiss’ Kappa) were systematically evaluated to identify those suitable for various analysis tiers. Subsequently, the application of Cohen’s Kappa and Fleiss’ Kappa across varying levels of agreement was examined, and thresholds for sample size selection were established. Lastly, an attribute measurement analysis model for gas polyethylene pipeline welded defects is developed: the first tier is based on preliminary assessment and data filtering regarding the presence of welded defects; the second tier employs weighted kappa analysis to quantitatively assess agreement on specific defect types, This results in a comprehensive evaluation of the performance of the measurement system.
|
|---|
