Bayes方法在矿井突水水源判别中的应用  

The application of bayesian approach to discrimination of mine water-inrush source

作　　者：张春雷[1] 钱家忠[1] 赵卫东[1] 马雷[1] 

机构地区：[1]合肥工业大学资源与环境工程学院,安徽合肥230009

出　　处：《煤田地质与勘探》2010年第4期34-37,共4页Coal Geology & Exploration

基　　金：国家自然科学基金项目(40672154;40872166);新世纪优秀人才计划项目(NCET-06-0541)

摘　　要：快速有效地判别突水水源是矿井安全生产的重要保障。选取各含水层多项水质指标,应用Bayes方法建立适用于不同水质类型的矿井突水水源快速判别模型。结合SPSS软件,以淮南顾桥矿为例,并与模糊综合评判模型、神经网络模型进行分析比较。结果表明:贝叶斯多类线性判别模型能够有效地判别突水水源,比模糊综合评判有更高的准确性,与神经网络模型的判别准确率相同。Bayes多类线性判别模型又以其计算过程简单、模型结构稳定而优于神经网络模型。既提高判别准确率又提高判别速度,实现对突水水源快速有效判别。Discriminating the water-inrush source correctly and quickly is an important guarantee for mine safe production. The paper selected many items of water parameters for each aquifer and applied Bayes method to establish the model of rapidly identifying the water-bursting source for different types-water coal mine. Using SPSS, the example of Huainan Guqiao mine was compared with the fuzzy comprehensive evaluation and neural network model. The results show that Bayes multi-class linear discriminant model has a higher discrimination accuracy than the fuzzy comprehensive evaluation and has the same accuracy as neural network model. However, Bayes linear discriminant model is superior to neural network model because of its simple calculation and stable structure.

关 键 词：水源判别 Bayes模型 煤矿 

分 类 号：TD741[矿业工程—矿井通风与安全] P641.4[天文地球—地质矿产勘探]