Abstract:

Aiming at the disadvantages of single BP net work, an ensemble BP(Back-Propagation) net work is presented by using AdaBoostR2 for improving the prediction accuracy of single BP net work. A new updating method is proposed for the characters of ensemble BP net work based on AdaBoostR2. The new method can update the model effectively and overcome the disadvantage of traditional updating methods. The new methods are used to predict the rock burst in Fushun Laohutai mine. The main influence factors for rock burst are analyzing detailed and used as the inputs of prediction model. The ensemble BP net work based on AdaBoostR2 is used as intelligent algorithm. The prediction model for rock burst predicting has a good prediction accuracy.

Key words: rock burst, neural network, AdaBoost;, prediction model, model updating