【摘要】 针对煤矿突水预测精度低、训练速度慢的问题，提出基于局部线性嵌入(LLE)—果蝇优化算法(FOA)—支持向量回归(SVR)的煤矿突水预测模型。首先，利用LLE在非线性数据特征提取方面的优势，提取煤矿突水影响因素原始数据的本质特征，形成重构因子，减少数据间的冗余信息和噪声。然后，利用FOA对SVR的参数进行迭代优化，并将最优参数代入SVR中，以解决传统SVR参数优化困难的问题。最后，结合实例并将LLE-FOA-SVR模型的预测结果与反向传播(BP)、SVR、LLE-SVR模型的预测结果进行对比。实验结果表明：该模型的预测精度高于其他3种模型，预测精度可达90%,且建模时间和运算时间更短。更多还原

【Abstract】 Aiming at the problems of low precision of coal mine water inrush prediction and slow training speed, a coal mine water inrush prediction model based on locally linear embedding(LLE)-fruit fly optimization algorithm(FOA)-support vector regression(SVR)is proposed.Firstly, using the advantages of LLE in non-linear data feature extraction, the essential features of the original data of coal mine water inrush influence factors are extracted to form reconstruction factors and reduce redundant information and noise between data.Then, the FOA is used to iteratively optimize the parameters of the SVR,and the optimal parameters are substituted into the SVR to solve the difficult problem of traditional SVR parameter optimization.Finally, combine the examples and compare the prediction results of the model with the prediction results of the back propagation(BP),SVR,and LLE-SVR models.Experimental results show that the prediction precision of the LLE-FOA-SVR model is higher than the other three models, the prediction precision can reach 90 %,and the modeling time and computing time are shorter.更多还原