Study On Prediction Forecast Of Rockburst Based On Matlab Neural Networks

As a great hidden danger of threatening the safety in production of the colliery, rockburst is earthquake that caused by project activity such as mining, tunneling etc. Because geological structure, geology movement and complexity of various kinds of mining activity perturbations, it's more difficult problem to quantitative predicted on rockburst. Theoretical researches indicate, as one mine power phenomenon essential, it is a kind of dynamics calamity brought by the mine exploits.This text proceeds from rockburst mechanism, considering comprehensive influence factor and dynamics characteristic, structured BP neural network based on matlab artificial neural network toolbox, adopting traingdm function (momentum gradient drop the backpropagation algorithm), trainlm function and trainbr function train BP network, neural network can be predicted accurately to set up predict models, made corresponding prediction software, keep non-linear research on predicts of system.Adopt one mine situation to survey contrast , carry on neural network practicability verification of method, the result shows : Use MATLAB neural network toolbox to construct BP neural network fast, high-efficiently, not only a large amount of network struction function that offer , study algorithm and friendly figure interface shorten neural modeling time of network greatly fast, enable network designer's getting rid of the tedious programming work, will study that transfers to how to optimize network configuration , improve the learning efficiency of the network, popularize ability and artificial result to come up especially; And, it predicts performance better, identical with reality basically, can predict rockburst accurately.