矿井突水水源识别的RS-LSSVM模型

发布时间：2018-04-17 18:03

本文选题：安全工程 + 矿井突水　；参考：《安全与环境学报》2017年05期

【摘要】：为了对矿井突水水源进行准确、高效的判别,综合考虑水化学特征,选取Ca~(2+),Mg~(2+),K~++Na~+,HCO-3,SO2-4,Cl~-和总硬度7个指标的质量浓度(mg/L)作为矿井突水水源的最初判别指标。利用粗糙集(RS)理论的属性约简来筛选水化学特征指标,用以作为水源识别的核心判别指标,建立基于RS的矿井突水水源识别的最小二乘支持向量机(LSSVM)模型。选用约简处理后的13组煤矿数据对模型进行训练,再用训练好的模型对另外12组突水数据进行水源判别,并与未进行属性约简的LSSVM模型及Fisher判别分析法、随机森林方法进行对比。结果表明,利用属性约简方法可以很好地排除原始数据中的冗余信息干扰,因而能有效判别矿井突水水源,使矿井突水水源模型的误判率降低至0;而且指标约简过程可以降低LSSVM运算的复杂度,也能够提高判别效率。
[Abstract]:In order to distinguish mine water inrush water accurately and efficiently, considering the characteristics of water chemistry, the paper selects Ca~(2 (Ca~(2) as the initial discriminant index of water inrush water source with seven indexes (mass concentration (mg / L) and HCO-3HCO-3HCO-3SO2-4).The attribute reduction of RS (rough set) theory is used to screen the hydrochemical characteristic index, which is used as the core discriminant index of water source identification, and the LSSVM model of mine water inrush water source identification based on RS is established by least squares support vector machine (LSSVM).The model was trained by 13 groups of coal mine data after reduction, and the other 12 groups of water inrush data were identified by using the trained model. The model was compared with the LSSVM model and Fisher discriminant analysis method without attribute reduction.The random forest method was compared.The results show that the method of attribute reduction can eliminate the interference of redundant information in the original data, so it can effectively distinguish the source of mine water inrush.The misjudgment rate of mine water inrush water source model is reduced to 0, and the process of index reduction can reduce the complexity of LSSVM operation and improve the discriminant efficiency.
【作者单位】：辽宁工程技术大学系统工程研究所;
【基金】：国家自然科学基金项目(71371091) 辽宁省社科基金项目(L14BTJ004)
【分类号】：TD745.2;TP18

【相似文献】