岩爆等级预测的PCA-OPF模型

为了高效准确预测岩爆烈度,将主成分分析(PCA)和最优路径森林(OPF)算法相结合,选取岩石单轴抗压强度、应力系数、脆性系数、弹性能量指数以及完整性系数这5个指标建立了岩爆预测的PCA-OPF分析模型。通过SPSS软件对国内外50组岩爆工程实例数据做主成分分析,依据方差累计贡献率得出3个主要影响因素,作为输入因子对OPF模型进行训练、评估、测试。试验结果的平均预测准确率可以达到91.25%,对比于其它数学模型,PCA-OPF模型预测准确率更高且更稳定,表明PCA-OPF模型在岩爆等级预测中有较好的实用性,可作为一种新的岩爆等级预测方法。

PCA-OPF Model for Rock Burst Prediction

For effectively predicting rock burst magnitude, a PCA-OPF model, a combination of principal component analysis(PCA) and optimum-path forest(OPF) algorithm, was established for rock burst prediction with uniaxial compressive strength, the stress coefficient, the brittleness coefficient, the elastic energy index and integrality coefficient of rock as the predictors. 50 groups of rockburst practical data at home and abroad were taken for the principal component analysis by using SPSS. Based on the cumulative variance contribution rate, three principal influencing factors were determined, which were taken as the input factors for training, evaluating and testing of OPF model. It is shown that the average prediction accuracy can reach 91.25%, which is higher and more stable than the prediction results of other models. It is concluded that the PCA-OPF model is more practicable for predicting rock burst classification and can be used as a new method in underground engineering.