Research on Intelligent Prediction of Rock Strength with Different Crack Geometric Characteristics
Based on the experimental results of rock mechanical properties with different crack geometric characteristics,and taking into account the interrelationships between crack characteristics,confining pressure,physical and mechanical param-eters,a database of characteristic parameters affecting rock strength was constructed.A rock strength prediction model under the influence of multiple characteristic parameters was established using the random forest algorithm,and the mapping relation-ship between rock strength and parameters was established.The results show that the accuracy of the experiment and training sets of the prediction model is 76%and 85%,respectively.Class Ⅲ and Class Ⅳ with rock strength ranging from 30～45 MPa and 46～60 MPa have the best prediction performance,with a sample prediction accuracy of 100%;Class Ⅱ prediction at 16～30 MPa has good performance,with a prediction accuracy of 80%;The Class Ⅰ prediction effect at 0～15 MPa takes second place,with a prediction accuracy of 71%.By using the joint distribution function,it can be concluded that the selected feature parameters have a certain degree of correlation,which can be obtained through simple physical experiments and non-destructive testing,and exhibit a strong linear relationship with rock strength;The model classification training calculation shows that the weight values of confining pressure and crack length are 0.294 and 0.263,respectively,which are key factors affecting rock strength.The importance ranking of other parameters is longitudinal wave velocity>Poisson's ratio>saturated mass>crack dip angle>dry mass>crack penetration>crack number.
万方数据
维普
