地下峒室岩爆预报的自适应模式识别方法

Adaptive Pattern Recognition to Predict Rockbursts in Underground Openings

应用神经网络系统理论，提出了地下峒室岩爆预报的新方法──自适应模式识别方法．该方法结合专家的经验，从积累的工程实例中抽取岩爆模式特征，建立输入模式输出模式对，采用自学习的方法建立从输入模式到输出模式的非线性映射．并进行推广，由网络推理出待识别岩石岩爆发生的可能性和烈度．实际应用表明，本文的方法科学、可靠．

With the neural network system theory applied to the prediction for probable rockbursts in underground opening,an adaptive pattern recognition is developed the way the rockburst pattern features are collected from previous cases in combination with expertises so as to set up I/O pattern pairs. Nonlinear mapping between various input patterns and their output patterns is thus established by self-learning of neural network and extrapolated. It is proved that the network and values of weights after learning are available to the identification of rockburst intensity for a new type of rock. The results show that this method is scientific and reliable.