神经网络模型在坝基扬压力影响因子量化分析中的应用

本文应用改进的BP网络模型定量分析坝基扬压力的影响因子,赋于网络不同的权值来表示网络的输入变量(水位、温度、时效等因子)对网络的输出变量(扬压力)的影响程度,从而确定各影响因子分量对扬压力的影响比例.采用Lvenberg-Marquardt算法训练网络,网络达到一定的次数后收敛.实例计算结果表明,该模型具有计算精度高、简便实用等特点.因而认为,把神经网络模型应用于探讨诸如环境量对于效应量影响程度的一类问题,具有好的前景.

Application of improved neural network model into analysis of uplift pressure under dam-foundation

In general,there are lots of factors leading to the variation in uplift pressure under a dam foundation.As a result,it is difficult to use a existing model to describe the pressure.Neural network model is a new model with certain advantages compared to other models.In this paper,the improved BP model was used to quantify the factors which affect the variation in the uplift pressure under the dam foundation,illustrated by a case study.The results demonstrate that the improved BP model is characterized by high accuracy and convenience in use.Therefore,it is believed that application of the improved BP model to problems with a cause-effect will be prospective in the future.