| 袁 颖.基于网格搜索法优化支持向量机的围岩稳定性分类模型[J].地质与勘探,2019,55(2):608-613 |
| 基于网格搜索法优化支持向量机的围岩稳定性分类模型 |
| Evaluation model for surrounding rock stability based on support vector machine optimized by grid search method |
| 投稿时间：2016-12-02 修订日期：2017-04-01 |
| DOI：10.12134/j.dzykt.2019.02.015 |
| 中文关键词: 围岩稳定性 支持向量机 网格搜索法 分类模型 BP神经网络 |
| 英文关键词: surrounding rock stability, support vector machine, grid search method, classification model, BP Neural Network |
| 基金项目:国家自然科学基金项目（编号 ：41301015）、河北省教育厅重点项目（编号： ZD2015073和ZD2016038）、石家庄经济学院国家自然科学基金预研基金（编号： syy201308）和河北地质大学第十三届学生科研基金项目（编号：KAG201607）联合资助 |
|
| 摘要点击次数: 1636 |
| 全文下载次数: 655 |
| 中文摘要: |
| 为了科学评价围岩稳定性，借助支持向量机(SVM)处理小样本、非线性问题能力强的特性，对围岩的稳定性进行了分类。选取16组围岩数据作为学习样本，以岩石质量指标、岩石单轴饱和抗压强度、完整性系数、结构面强度系数和地下水渗水量5个指标作为模型输入，围岩稳定程度为模型输出，建立了基于支持向量机的围岩稳定性分类模型。为增强模型的推广性能，提高其预测准确率，运用改进的网格搜索方法(GSM)寻找最优的支持向量机参数，并对8组围岩数据进行预测，并同BP神经网络模型的预测结果做出对比。结果表明，建立的GSM-SVM模型对预测样本的评判结果与实际结果一致，其预测精度较BP神经网络有很大的提升。 |
| 英文摘要: |
| In order to make a scientific evaluation of surrounding rock stability, this work used the characteristics of support vector machine (SVM) for dealing with small samples and nonlinear problems to classify surrounding rock stability. Sixteen groups of surrounding rock data were chosen as learning samples, five indexes including rock quality, rock uniaxial saturated compressive strength, structural surface intensity coefficient and amount of groundwater seepage as the model input, and surrounding rock stability as a model output, thus the classification model for surrounding rock stability based on support vector machine was established. In order to improve generalization and prediction accuracy of SVM model, the SVM parameters were optimized by improved grid search method (GSM). Then eight groups of roadway surrounding rock data were predicted and results were compared with those of the BP neural network model. The results show that the stability classification results by GSM-SVM model for surrounding rock accord with the actual results, and the prediction accuracy of GSM-SVM model is greatly improved than that of BP neural network. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
