Abstract: Rock damage prediction is important for controlling the blasting vibration of high rock slopes. This paper derives a relationship of rock damage versus natural frequency of rock masses using theoretical analysis, and develops a method for predicting rock damage from the natural frequency change of rock masses. We present a case study of Baihetan Dam, predicting the rock damage of its high rock slope of right bank abutment, based on blasting vibration monitoring and acoustic testing. The results show that a prominent linear relationship exists between the change rate of natural frequency and the damage depth of rock masses, and the rock damage prediction model can be adjusted dynamically and becomes more robust with the increasing data samples. For the rock damage depth of this high rock slope, the probabilities of its Bayesian LASSO predictions below the monitoring values are all higher than 0.83. These predictions can cover the observation results in the form of probability and give the uncertainty.

Key words: hydropower engineering, high rock slope, rock damage, natural frequency, blasting vibration, acoustic test