基于Bayes判别分析方法的叙岭关隧道溶洞水源识别

基于Bayes判别方法,选取水化学常量组分作为判别指标,利用叙岭关隧道地区9个动态监测点的53个水样样本建立了该地区的水源判别模型.模型检验结果表明,其回判准确率为96.23％,具有较高的识别精度和工程推广能力.利用建立的判别模型,对叙岭关隧道1号溶洞内两个出水点RQ1和RQ2的水源进行了识别,并结合其流量动态变化特征、同位素分析结果,以及1号溶洞发育位置,推断1号溶洞内两个出水点的水源为P1m+q含水层中的岩溶地下水.根据判别结果,建议1号溶洞采取“以排为主”的原则加以处治,并尽量保留溶洞水的过水通道.

Water Source Identification of Karst Cave in Xulingguan Tunnel Based on Bayes Discriminant Analysis

A Bayes discriminant model for water source identification of kasrt cave in Xulingguan tunnel is established using 53 water samples of 9 dynamic monitoring points.It chooses hydrochemical major constituents as discriminant index and bases on Bayes discrimninant theory.51 of all the samples are classified correctly with 96.23% accuracy rate,when the model is adopted to test the training samples,indicating that this model has perfect performance and good generalization ability.P1m+q aquifer is identified as the water source of RQ1 and RQ2 in No.1 karst cave when using the above Bayes model,combining with its flow characteristic,isotopic analysis result and development position.According to the results,the water of No.1 karst cave is suggested to be drained from its own channel.