金属矿采空区危险性判别的PCA-SVM模型研究  

Study on PCA-SVM Model for Evaluation of Gob Hazards in Metal Mine

作　　者：刘志祥[1] 郭虎强 兰明[1] 

机构地区：[1]中南大学资源与安全工程学院,湖南长沙410083

出　　处：《矿冶工程》2014年第4期16-19,共4页Mining and Metallurgical Engineering

基　　金：国家科技支撑计划(2013BAB02B05和2012BAB08B01);中南大学教师基金(2013JSJJ029);国家自然科学基金和上海宝钢集团公司联合资助(51074177)

摘　　要：为了有效合理地判别金属矿采空区的危险性,引入主成分分析法(PCA)及支持向量机(SVM),建立了PCA-SVM判别模型。搜集国内金属矿40组采空区失稳资料,首先利用主成分分析法对采空区失稳因素进行降维处理,消除各因素之间的冗余,获取样本集主要信息。然后利用支持向量机对保留的主成分数据进行建模,并引入遗传算法对SVM模型参数进行优化,改善SVM模型的判别效果。对判别模型进行训练及检验,结果表明,该模型对训练样本和检验样本的判别精度分别达到100%及90%。最后将该模型运用于工程实际中,其判别结果与实际情况相符,表明该模型在工程实际中具有一定的应用价值。A PCA-SVM model was established by introducing principal component analysis （ PCA） and support vector machine （ SVM） for effectively evaluating gob hazards in metal mines. With 40 sets of data about unstable gob in domestic metal mines, factors contributing to the instability of mind-out area were firstly treated with principal component analysis for dimensionality reduction, so as to eliminate redundancies among factors and obtain main information of sample of data set. Then SVM was used to build model for the remaining date after PCA, and parameters of the SVM model was optimized by introducing genetic algorithm, resulting in an improvement in evaluation effect of SVM model. Based on the training and testing of such evaluation model, it is found that the evaluation accuracy of the training samples and testing samples had respectively reached 100% and 90%. Finally, its practical application in projects showed the evaluation result in accordance with practical result, verifying its application value.

关 键 词：采空区 危险性判别 主成分分析 支持向量机 

分 类 号：TD325[矿业工程—矿井建设]