Study On Prediction Method And Theorial Model Of Rockburst

Rockburst, a typical geologic hazard, which may kill constructors and destroy equipment, usually occurs in high geostress area. Nowadays, a large number of underground engineering, such as tunnels in railways and highways, underground powerhouse in hydraulic engineering, underground mining and many others, are involved with deep excavations in rock masses with high geostress in China. Therefore, prediction and prevention of rockburst become more urgent than ever and have been a very important research topic in deep rock excavation.Accurate prediction of rockburst is helpful for designers and constructors to take effective counter measures to reduce economic losses and avoid casualties. However, due to the complexity of rockburst mechanism, it is very difficult to perform accurate prediction. Thus, simple classification methods are currently used for rockburst prediction. Due to these methods are too simple to account for the influence of numerous and complicated factors, the results do not agree with engineering practice well.To solve the problem existing in rockburst prediction, following researches were carried out:(1) Rockburst for Cangling tunnel was predicted with several methods based on traditional strength theory and shortage of the strength theory were discussed.(2) General regression neural network (GRNN), which can take into account of the rockburst character and overcome the shortage of strength theory, was employed to build up a prediction model. The particle swarm optimization (PSO) algorithm was employed to reduce the adverse influence of man-induced. Merit and shortage of the PSO-GRNN model were exhibited through predicting the rockburst for Cangling tunnel and the exploratory tunnel of Jinping â…¡ Hydropower Station.(3) Since the influencing factor of rockburst were continuous data and the degree of rockburst were discrete data, a quantitative appraisal about the importance of the influencing factor was made by rough set based on field investigation.(4) Ideal point method is a solution for multi-objective decision, with weight coefficients calculated by rough set, a model combined rough set and ideal point method for rockburst prediction was built. The calculated results agree well with the Cangling tunnel and the exploratory tunnel of Jinping â…¡ Hydropower Station data, and the practicability of this model was tested. (5) Evidence theory, based on the idea of information fusion, can reflect the comprehensive influences of different factors, with achievements from rough set, a model based on rough set and ideal point method was built, which also can give correct result of the two cases above.(6) Three rockburst prediction model, model based on the ideal point method and rough set, model based on the evidence theory and rough set, and model based on the fuzzy mathematics method were compared and discussed.New conclusions obtained from the research above are as follows:(1) Rockburst prediction result for Cangling tunnel show that the PSO-GRNN model has advantage over the common BP model and common GRNN model in stability and accuracy, but result for exploratory tunnel of Jinping II Hydropower Station shows that the PSO-GRNN model is defective.(2) Evaluation with rough set show that, for rockburst, stress concentration is the most important influencing factor, the energy index is the second, and brittleness index is the third.(3) Prediction result show that the model based on ideal point method and rough set is more precise than ideal point method model with weights obtained from analytic hierarchy process method or equal weight.(4) Prediction result show that the model based on the evidence theory and rough set is more precise than other two evidence theory model built with artificial probability distribution.(5) Model based on the ideal point method and rough set, model based on the evidence theory and rough set, and model based on fuzzy mathematics method are comparative, and the first two models have advantage over the fuzzy mathematics method in reflecting the variation trend of rockburst.