基于SCM与K-means聚类算法的矿工不安全动作分类特征研究

Study on characteristics of unsafe behaviors in coal mines based on SCM and K-means clustering analysis

为减少托管运营煤矿不安全行为导致的安全事故,针对此种运营模式下人员的不安全动作进行科学分类研究;现场调研了2017—2018年内蒙古、宁夏、新疆地区6个煤矿1996名煤矿从业者的“三违”行为情况,从时间、工种及SCM行为产生特点3个方面对全部“三违”行为进行统计分类;基于k-means聚类算法建立了8个指标、4个子类的数据集合,并通过PCA降维绘制了可视化聚类散点图。分析表明:采用SCM和k-means的聚类算4种不安全动作分类占比关系与人工分析均得出了相同的结论;在所有不安全动作中,违章占比最大,错误占比最小;研究结果对于减少煤矿从业人员不安全动作行为,以及分级、分类预防安全事故的发生具有一定的指导意义。

In order to reduce the safety accidents caused by the unsafe behaviors in the managed operation of coal mines,we conducted a scientific classification study on the unsafe behaviors of personnel under such operation mode.The field investigation on the“three violations”behaviors of 1996 coal miners in six coal mines in Inner Mongolia,Ningxia and Xinjiang from 2017 to 2018 was conducted.All the three violations were statistically classified from three aspects of time,job type and characteristics of SCM behavior generation.A data set of 8 indicators and 4 subcategories was established based on K-means clustering algorithm,and a visual clustering scatter diagram was drawn through PCA dimension reduction.The analysis shows that:using SCM and Kmeans clustering to calculate the proportion of four kinds of unsafe action classification,the same conclusion was obtained as manual analysis,in all unsafe action,lawless accounted for the biggest,errors accounted for the smallest proportion.The research results have certain guiding significance for reducing unsafe behaviors of coal mine employees and preventing safety accidents by classification.