鹤壁矿区利用水化学成分识别突水水源建模研究

发布时间：2019-01-01 13:54

【摘要】：水害是影响煤矿安全生产的主要威胁之一,突水后快速并准确地判断突水水源是矿井防治水工作的主要内容。本文以华北型煤田—鹤壁矿区为研究对象,对利用水质化验数据、建立数学模型、快速准确地判断突水水源类别问题进行了研究。开展了突水水源水质普查工作,采集了各充水含水层地下水及老空水标准水样,并收集了已有水质化验数据,归纳总结了各充水水源的水化学特征。其次,采用主成分分析法,从经过反复筛选的多项水质化验指标中,确定了能够准确反映各充水水源水质特征的“TDS”、“Na-Cl”、“Mg-HCO3-”主成分,作为识别突水水源的主要指标。在此基础上,建立了基于水质识别突水水源的灰色关联度模型、模糊数学模型、BP神经网络模型、Bayes判别模型和Fisher判别模型,并对已有的水质化验数据进行了验证性识别。识别结果显示:Fisher判别模型和BP神经网络的正确识别率最高,均为84.6%,灰色关联度模型和Bayes判别模型次之,模糊数学模型正确识别率较低。经过比较,Fisher判别模型不仅正确识别率高且操作简单,易于学习,从而选用Fisher判别模型作为鹤壁矿区基于水质识别突水水源的主选方法,其它模型作为辅助方法,这一结论已被我们承担的《鹤壁矿区基于水质自动识别矿井突水水源软件的研发》科研项目所采用。
[Abstract]:Water hazard is one of the main threats to the safety of coal mine production. It is the main content of mine water prevention and control work to judge the water inrush source quickly and accurately after water inrush. In this paper, taking the Huabei coalfield-Hebi mining area as the research object, a mathematical model is established by using the water quality test data to judge the water inrush water source category quickly and accurately. The general survey of water quality of water inrush source was carried out, the groundwater samples of each water-filled aquifer and the standard water samples of old empty water were collected, and the available water quality test data were collected, and the characteristics of water chemistry of each water-filled source were summarized. Secondly, the principal components "TDS", "Na-Cl" and "Mg-HCO3-", which can accurately reflect the water quality characteristics of each water-filled source, are determined by principal component analysis (PCA). As the main indicator to identify the source of water inrush. On this basis, the grey relational degree model, fuzzy mathematical model, BP neural network model, Bayes discriminant model and Fisher discriminant model are established based on water quality identification of water inrush water source. The recognition results show that the correct recognition rate of Fisher discriminant model and BP neural network is the highest, which is 84.6, followed by grey correlation model and Bayes discriminant model, and the correct recognition rate of fuzzy mathematical model is lower. Through comparison, the Fisher discriminant model is not only correct recognition rate, simple operation, easy to learn, so Fisher discriminant model is selected as the main selection method based on water quality identification of water inrush water source in Hebi mining area, and other models as auxiliary method. This conclusion has been adopted in the scientific research project of "automatic identification of mine water inrush source software based on water quality in Hebi mining area".
【学位授予单位】：河南理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P641.461;TD745

【相似文献】