Advancing drilling parameter reliability: A data-driven comparison of MWD and DPM for stratigraphic identification

Stratigraphic interface characterization and strength parameter assessment of geomaterials constitute fundamental research priorities in geological and geotechnical engineering. While measurement while drilling (MWD) and drilling process monitoring (DPM) have emerged as critical techniques for acquiring real-time drilling parameters, inherent limitations in data interpretation persist. The critical challenge of random fluctuations in MWD-derived penetration rate measurements exhibits poor correlation with the stratified homogeneity characteristics of geological formations. Such discrepancies undermine the reliability of stratigraphic classification and mechanical property analysis. Through systematic comparison of MWD and DPM datasets combined with quantitative parameter evaluation, this investigation reveals significant methodological distinctions in data acquisition accuracy. Machine learning-enhanced analysis employing Support Vector Machine (SVM) algorithms demonstrates that DPM-derived parameters provide superior stratigraphic identification capabilities. Our findings indicate that DPM implementations achieve 20.57% and 38.01% higher resolution in interface detection along two drillholes compared to the conventional MWD approaches. This improvement allows for better prediction of stratigraphic profiles and more precise guidance in subsequent geological and geotechnical engineering practices.