论文部分内容阅读
由于土体的复杂性,在目前士坝设计中,尚不能对影响土坝渗流的各种因素完全加以考虑,加上施工等因素的影响,大坝的实际运行状况与设计往往具有差别。因此大坝观测是了解大坝运行状况的有效途径。在土坝渗流观测资料分析中一般采用回归分析方法分析库水位和测压管水位关系曲线,用位势进行测压管水位的预测。本文用介绍一种简单易行的神经网络方法,进行库水位和测压管水位关系曲线分析及测压管水位预测。并用山东省引黄济青工程棘洪滩水库土坝观测的实际资料进行了神经网络的训练,得到了训练后的权值和阀值,通过神经网络模拟和预测,用神经网络进行大坝渗流观测分析。
Due to the complexity of soil, in the current design of Shi dam, the various factors that affect the seepage of earth dam can not be fully considered. Due to the influence of construction and other factors, the actual operation status of the dam often differs from the design. Therefore, dam observation is an effective way to understand the operation of the dam. In the analysis of seepage observation data of earth dam, the regression analysis method is generally used to analyze the relationship between reservoir water level and piezometric water level, and the potential water pressure is used to predict the piezometric water level. In this paper, a simple and easy neural network method is introduced to analyze the relationship between reservoir water level and piezometric water level and the piezometric water level prediction. The neural network was trained with the actual data of earthen dam observation in Jiuxintan Reservoir of Yellow River Diversion Project in Shandong Province. The weights and thresholds after training were obtained. Through the neural network simulation and prediction, the seepage analysis of dam was performed by using neural network.