Application of Ultrasonic-Rebound Method in Fast Prediction of Rock Strength

Abstract

Fast and accurate prediction for rock strength of intact rock in underground engineerings is the key to ensure the construction safety. Hence, a test method combing ultrasonic wave and rebound for rock strength is established through the test of rock surface hardness and internal defect. Meanwhile, the ultrasonic-rebound-strength model is built by combining traditional mechanical rebound value and new energy rebound value with rock longitudinal wave velocity test value with the methods of bivariate regression analysis, BP neural network and support vector machine (SVM). It applies the models established in the prediction of gneiss strength prediction and determines the optimal model for ultrasonic-rebound strength prediction after analyzing correlation coefficient and relative standard strength deviation. From the prediction results of five test samples, the rock strength prediction model of ultrasonic-rebound method obtained from the SVM method reveals the highest accuracy. At the same time, according to the strength test results of the test samples, the use of new energy resiliometer can greatly improve the prediction accuracy of rock strength.