Reliability Analysis of Multiple Machine Learning Methods for Rapid Discrimination of Rock Types
In response to the high cost and long time consumption of traditional drilling and testing methods for obtai-ning rock mass strength in complex geological conditions in water conservancy and hydropower engineering,machine learning methods have been introduced into rock type discrimination to achieve rapid identification of composite directional drilling rock types,which provides support for the rapid classification of surrounding rock types in underground engineer-ing.Based on the key technology research and application project of ultra-deep composite directional drilling in water con-servancy and hydropower engineering,combined with the existing geological data,10 machine learning classification algo-rithms are used to distinguish the rock type of composite directional drilling parameters,and the discrimination effect of the model is compared and analyzed in detail.The results show that the RF,AdaBoost,CatBoost,KNN,SVM,and Ex-traTree perform well,with AdaBoost performing the best;The feasibility and reliability of machine learning methods in rock type discrimination have been verified,providing guidance for subsequent algorithm selection and optimization.
万方数据
维普
