【作者】 闫志刚； 杜培军； 汪云甲；

【Author】 YAN Zhi-gang,DU Pei-jun,WANG Yun-jia(School of Environment and Spatial Informatics,Chinese University of Mine and Technology,Xuzhou 221008,China)

【机构】 中国矿业大学环境与测绘学院；

【摘要】 为了有效处理矿井突水信息,提出了一种新的数据挖掘方法,即SVM-RS方法。首先使用SVM对原始信息进行预处理,以约简去除无用、无效属性,确定突水预测的有效信息维,然后再利用SVM对连续的属性值进行离散化处理,最后利用RS分析数据,提取矿井突水预测规则。该方法集成了SVM和RS的优势,克服了它们的不足,在实际应用中表现良好。更多还原

【Abstract】 To process the information of water gushing in coal mine effectually,a novel method was put forward,namely SVM-RS.Firstly,SVM was used to preprocess the raw data,the useless or invalid attributes were reduced,and a more compatible feature space was built for water gushing prediction;then the continuous-valued attributes were discretized by SVM;finally,the preprocessed data were analyzed by RS,and the rules were extracted for prediction.The method integrates the advantages of SVM and RS,and offsets their deficiencies.The experimental results show that the method is proved practicable and useful.更多还原