库坝区地下水渗流场模式识别模型研究

STUDY ON PATTERN RECOGNITION MODEL OF GROUNDWATER SEEPAGE FIELD IN RESERVOIR AND DAM SITES

阐述应用模式识别方法研究库坝区地下水渗流场特征的优点。基于模糊优选相关理论建立混合比模型,根据不同样本之间的区别和联系从理论研究及工程应用角度给出水样类别的定量化界定标准,为目标模式的数学表达提供理论支持。基于模式识别基本思想并结合多元函数及BP网络相关理论建立多因素增量模型,其目标模式识别值存在且唯一确定,该模型能够筛选并整合有效信息,针对未知样本与已知样本基元值之间的增量关系来实现样本目标模式的确定,完成模式识别。通过龙羊峡坝区地下水渗流场工程实例进行验证,结果表明,模式识别方法对于研究地下水渗流场特征和补给关系具有避免过多主、客观因素干扰,便于程序实现,计算过程清晰,结果明确的特点。

The advantages with the use of the pattern recognition method to study the characteristic of the reservoir and dam groundwater seepage field are described briefly. According to the theory of fuzzy selection, the mixing ratio model is built. It presents the criteria to classify the water samples quantitatively through application based on the difference and relation among various samples, and offers the precondition in order to describe target pattern. According to the basic concept of pattern recognition and combining with the theory of multifunction and BP network, the multi-factor increment model is established. The target value of each pattern is unique. The model can filter and harmonize valuable information, and deduce the target model value by analyzing the increment between known and unknown variables. An instance of Longyangxia dam groundwater seepage field is provided. It shows that there are so many merits about the pattern recognition method within the research domain of groundwater seepage field characteristic and supply relationship, such as the interfere from subjective or impersonality is avoidable, it is easy to program, the process and results are clear.