Application of projection pursuit model and particle swarm optimization in rock burst prediction

In order to construct the measure of rock burst intensity, the ratio of maximum tangential stress of cave chamber to uniaxial compressive strength of rock, brittleness coefficient and elastic energy index are chosen as the discriminant index, an appropriate analysis model for rock burst prediction was established based on particle swarm optimization and projection pursuit algorithm. Firstly, for the sake of ensuring the accuracy of the model parameters, particle swarm optimization is used to optimize the projection index function, meanwhile the non-linear relationship between projected value and empirical value is obtained by use of logistic curve function. The study shows that the prediction of rock burst intensity with use of the regression model based on particle swarm and projection pursuit has the advantage over traditional forecasting methods in that the deviation caused by subjective reasons can be avoided and its prediction precision is high. Finally, the model was applied to the rock burst prediction of Qinling tunnel and Dongguashan copper ore and the result corresponds with actual situation which shows scientificity, feasibility and effectiveness of the model in rock burst prediction.