基于人工蜂群算法优化支持向量机的采场底板破坏深度预测

发布时间：2018-03-07 22:36

本文选题：蜂群算法(ABCA)　切入点：支持向量机(SVM)　出处：《重庆大学学报》2015年06期 　论文类型：期刊论文

【摘要】：为确定合理的底板防水煤岩柱尺寸,减少底板突水安全事故的发生,利用支持向量机(SVM)与人工蜂群算法(ABCA)综合研究底板破坏深度问题。由于SVM训练参数惩罚因子C和核函数宽度g的选择对预测精度的影响显著,采用ABCA优化该训练参数的选择过程,建立基于SVM的底板破坏深度预测模型。选取采深、煤层倾角、采厚、工作面斜长、底板抗破坏能力和是否有切穿断层或破碎带作为影响底板破坏深度的主要影响指标,利用现场实测的30组数据作为样本对该模型进行训练和预测。结果表明:该预测模型的平均相对误差为12.5%,平均绝对误差为0.986m,均方误差为0.005,平方相关系数为0.980,较其他预测模型具有更强的泛化能力和更高的预测精度。
[Abstract]:In order to determine the reasonable size of waterproof coal and rock pillar of floor slab and reduce the occurrence of water inrush safety accident, Support vector machine (SVM) and artificial bee colony algorithm (ABCA) are used to study the depth of floor damage. Because the choice of penalty factor C and kernel width g of SVM training parameter has a significant effect on prediction accuracy. Using ABCA to optimize the selection process of the training parameters, the prediction model of floor failure depth based on SVM is established. The mining depth, coal seam inclination angle, mining thickness, working face slope length are selected. The ability of the floor to resist failure and whether there are shear faults or broken zones are the main factors affecting the depth of failure of the floor, The results show that the average relative error, absolute error, mean square error and square correlation coefficient of the model are 12.5, 0.986m, 0.005 and 0.980, respectively. His prediction model has stronger generalization ability and higher prediction accuracy.
【作者单位】：辽宁工程技术大学矿业学院;辽宁久安安全技术咨询有限公司;
【基金】：国家自然科学基金资助项目(51274117)~~
【分类号】：TD745.2;TP18

【相似文献】