【作者】 高瑞忠；

【导师】 朝伦巴根；

【作者基本信息】 内蒙古农业大学 ， 农业水土工程， 2006， 博士

【摘要】 随着人类经济活动范围的日益扩大,地表水资源污染日益严重,地下水的开发规模不断增大,出现地下水的大量开采,要做到充分、有效和持续地利用地下水资源,把由于开采地下水资源酿成的危害限制到最小程度,就必须对地下水资源进行科学的管理,地下水系统计算是地下水资源管理的前提与基础,正确地进行地下水系统计算为合理的配置和开采地下水资源提供依据和指导。关于地下水系统计算的人工智能应用研究内容极为丰富,难以一一涉及,本文以FORTRAN90计算机语言为载体,就加速遗传算法(AGA)和自适应神经网络算法(A-BP)在地下水系统计算的某些方面的应用进行了研究。在各种类型含水层水文地质参数的推求上,通过对描述含水层特性的水文地质模型的数值分析,将AGA和A-BP融于水文地质模型中,给出了AGA和A-BP确定承压含水层系统参数的过程,提出了以AGA和A-BP确定越流含水层系统参数的方法及确定潜水含水层系统参数的Boulton-AGA、Boulton-A-BP、Neuman-AGA及Neuman -A-BP方法,系统化了各类型含水层水文地质参数的推求,简化计算过程;以气象数据为预测因子,建立A-BP模型和基于AGA的相似预测模型、投影寻踪预测模型、组合预测模型,实现了对地下水位的短期预测;以地下水水质标准产生建模样本,建立A-BP模型和基于AGA的相似评价模型、投影寻踪评价模型、S曲线评价模型,实现了对地下水水质的分类评价,并对多种方法的评价结果进行了对比;将AGA应用于地质统计学,以AGA估计理论变差函数模型的参数、优化确定估计方差最小和无偏性约束条件下的克立格、协克立格权重系数,以A-BP模型实现类似于协克立格的空间估值;以AGA求解了基于响应矩阵法的地下水管理模型,取得了较好的计算效果。文章最后概括总结本文主要研究的内容及取得的成果,并对存在问题进行了探讨及对有待进一步研究的问题进行了展望。更多还原

【Abstract】 With the gradual enlargement of human economy activity, the surface water pollu-tion become more and more seriously. The exploitation of groundwater increases quickly and cause environment worse. So, in order to completely, effectively, sustainable utilize the groundwater and minimize the danger caused by a great quality of exploitation of groundwater, the groundwater must be managed scientifically. The groundwater system calculation is the thesis of the management of groundwater. The reasonable groundwater system calculations offer the guidance for the distribution and exploitation of groundwa-ter.The contents of application of artificial intelligence method in the groundwater sys-tem calculation are very widely and it is unrealistic to research all. So, the author applies Accelerating Genetic Algorithm (AGA) and Adaptive Back-Propagation Algorithm (A-BP) in some aspect of the groundwater system calculation. The application processes are realized by one kind of computer language, FORTRAN90.Firstly, in the process of identifying the hydrogeology parameters of different type aquifers, the AGA and A-BP are combined in hydrogeology model. The procedures of identifying parameters of confined aquifer, leaky and confined aquifer, and unconfined aquifer are presented in detail. The method is proposed for identifying the leaky and con-fined aquifer. The Boulton-AGA, Boulton-A-BP, Neuman-AGA, Neuman-A-BP are proposed to identify unconfined aquifer. All of these methods make the estimation of aq-uifers parameters become simple. Secondly, take the weather datum as the prediction factors, the short-term groundwater table is predicted by A-BP prediction model, Shepard similarity prediction model, Projection pursuit prediction method, Combined forecasting model. Thirdly, by producing the model random samples in accord to the Groundwater Quality Norm, the A-BP evaluation model, Shepard similarity evaluation model, Projec-tion pursuit evaluation model, and Logistic curve evaluation model are established to evaluate the groundwater quality. Fourthly, AGA is applied in the geostatistics,for ex-ample, the estimation of variograms model parameters, the calculation of the weight co-efficient of Kriging and CoKriging methods. Similarity to CoKriging estimation method,更多还原