Addressing the current inadequacies of a single criterion to predict the severity of damage caused by rockburst due to its complexity of the variables, this paper proposes a methodolgy with quantitative and qualitative analysis based on the engineering information from the deep and long diversion tunnels of Jinping II Hydropower Station. This method incorporates in its model of damage severity estimation by using the Genetic-BP neural network algorithm which microseismic monitoring information and engineering conditions are taken into account, and as a result provides a comprehensive estimation on the severity of damage caused by rockburst with the elimination of influence of randomness to some extent. A good agreement between the model's estimation and the statistical data collected from measurements made in the field was observed, indicating the efficacy of the model in estimation the severity of damage caused by rockburst.
Along with the exploitation of underground rock engineering approaching to the deep constantly, rockburst hazard under the condition of high stress has become increasingly acute. The study of estimation on rockburst has turn into a challenging problem to be solved in the forefront of research accordingly.
Due to the influence of numerous conditions on rockburst, no single criterion (strain energy index, stress-strength rate, brittleness index and so on) can satisfy the need of accurate estimation of rockburst. However, the method of neural network can provide advantages in this respect.
The advantage of this method is that it could use previous engineering data to research current engineering problems, and reduce the interference of human factors due to not needing of establishing an analytical criterion (Feng, 2000). Thus it is more objective and has strong anti-jamming (Fig. 1).
The paper choses the deep and long diversion tunnels of Jinping. Hydropower Station as the research subject. The station is located on the Yalong River in West China, is a long diversion-type hydropower station. The average length of four diversion tunnels is around 16.7 km. These diversion tunnels are generally with an overburden depth of 1500–2000 m, and the maximum depth is 2525 m. The maximum in-situ stress measured in field is 46.1MPa, and the maximum principal stress after regression at the largest overburden depth of the diversion tunnels could reach 72MPa (Shan and Yan, 2010). Due to the complex geological conditions, the large overburden depth and high in-situ stresses, rockbursts were freguently occurred during the excavation which seriously threated builders' lives and equipment seriously and influenced the progress of construction.
