【作者】 边博；

【导师】 白丹；

【作者基本信息】 西安理工大学 ， 环境工程， 2005， 硕士

【摘要】 宁夏石嘴山市地处干旱地区,水资源供需矛盾十分突出,地下水是该市主要水源之一,保护地下水水源,对当地社会和经济的发展具有重要意义,本文针对该市地下水水质变化实际情况,开展地下水质预测和评价研究,研究成果对地下水资源的合理开发、利用和保护具有重要的现实意义。 针对石嘴山市地下水水质监测资料比较少的特点,本文采用了多种方法预测地下水水质变化趋势。首先根据地下水水质的观测资料,建立了地下水水质二次曲线指数平滑模型,平滑系数采用跟踪信号自适应选取方法,较好地解决了确定平滑系数的问题;其次对灰色系统预测模型作了改进,建立了带有变参数自适应预测模型,根据需要选取不同的参数值,计算误差最小的参数值,从而建立预测公式,在此基础上,开发了基于MATLAB灰色系统程序GM.m;最后提出基于BP网络的组合预测模型,它以指数平滑模型和灰色系统模型的预测结果作为网络的输入,实测值相当于网络的输出,通过网络的学习获得输入模型的权重,从而建立预测模型。对上述三种预测模型的预测误差进行比较,分析了各预测模型产生误差的原因,同时也讨论了各预测模型在地下水水质预测中的优势和不足。 应用人工神经网络理论和内梅罗污染指数法对石嘴山市地下水污染的现状分别进行分级、分区评价。据此,分析了石嘴山市地下水污染的成因、途径以及地下水污染的变化规律,提出相应的地下水水质保护和管理措施。更多还原

【Abstract】 Shizuishan of Ningxia is seated in arid area, whose imbalance between supply and demand of water resources is very conspicuous, and groundwater is one of its main water sources. Protecting groundwater sources is important for local society and economic development, aiming at the actural change process of the city groundwater quality, this thesis does underground water quality prediction and evaluation research, and its results have important realistic value for reasonable development, utilizing and protection of groundwater resources.Because of fewer monitoring datas of groundwater quality in Shizuishan, the thesis applies many kinds of methods to predict water quality variety tendency of groundwater. First, Based on the observation datas of groundwater quality , it sets up groundwater water quadratic exponential smoothing model. Smooth parameter adopts the following signal and chooses the method adaptively, which solved the smooth parameter perfectly. Then gray system prediction model is made some progress, and an adaptive prediction model with varied parameter is established. According to needs , choosing different parameter values, and calculating the parameter values with minimum error, this thesis sets up prediction formula. On these basis, the thesis develops gray systematic procedure GM.m with the aid of MATLAB, and proposes the combined-forecast method based on BP network ultimately, which regardsprediction results of exponential smoothing and gray system model as input datas of the network, and survey value as the output datas, the weighting of the input model is obtained through the network study, thus, the prediction model is established. Compared with the prediction errors of three kinds of prediction models mentioned above, this thesis analyzes the reason that each model produces error, and discusses the advantage and deficiency of each model in predicting groundwater water quality.The evaluation of the grade and section of current situation of groundwater pollution with artificial neural network theory and Nemerowint pollution index method is presented respectively. Based on these, the original causes, ways and Change law of underground water pollution in Shizuishan are analyzed, corresponding groundwater water quality protection and management measures are put forward.更多还原