用Fisher判别法评价矿井通风系统安全可靠性  

Reliability Assessment for Mine Ventilation System Safety Using Fisher Discriminant Analysis

作　　者：史秀志[1] 周健[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：《采矿与安全工程学报》2010年第4期562-567,共6页Journal of Mining & Safety Engineering

基　　金：国家"十一五"科技支撑计划项目(2006BAB02A02);中南大学学位论文创新项目(2009ssxt230)

摘　　要：针对传统的通风系统安全可靠性评价方法存在的问题,应用统计学理论并结合对矿井通风系统经济性、安全性的要求,选取16项矿井通风系统评价指标作为判别因子,建立矿井通风系统安全可靠性的Fisher判别分析(FDA)模型.以15组具体数据作为学习样本进行训练和检验,建立相应线性判别函数并利用回代估计方法进行回检,误判率为0;并对3组检测样本进行评判,识别率100%.此外利用该模型对3个实际生产矿井通风系统测定数据作为预测样本进行测试,预测结果与实际情况吻合较好,并与神经网络结果一致.研究结果说明:该模型在研究矿井通风系统的安全可靠性中具有较高的可信度,从而为评价矿井通风系统的安全可靠性探索了一条新思路.Aiming at the problems of the traditional method of assessing security and reliability of ventilation system,we built the Fisher discriminant analysis（FDA） model according to the economic and safety requirement for assessing the safety reliability of mine ventilation system.In building this model,16 indexes of the ventilation system were used as the discrimination factors.A FDA model is obtained through training 15 sets of measured data from some coal mines,the re-substitution method is introduced to verify the stability of FDA model and the ratio of mis-discrimination is zero.Therefore,the feasibility of the proposed model is validated.The FDA model is used to discriminate other three samples and the prediction results are identical with actual situation.Moreover,the proposed model has been used to predict some new samples of three actual coal mine ventilation reliability and the prediction results agree well with actual situation and are consistent with the artificial neural network（ANN）.The results show that the FDA model has a high credibility in assessing mine ventilation reliability of risk classification,so it is a new approach to forecast mine ventilation reliability,which can be applied to practical engineering.

关 键 词：矿井通风 安全评价 可靠性 预测 Fisher判别分析(FDA) 回代估计方法 

分 类 号：TD724[矿业工程—矿井通风与安全]