基于BP神经网络模型的水库塌岸预测研究

Collapse Prediction of reservoir based on BP neural network model

考虑水库塌岸预测的非线性,构建基于BP神经网络模型的水库塌岸非线性预测模型,以漳河流域内岳城水库为研究对象,结合野外实地勘察数据,对岳城水库塌岸进行预测。研究结果表明:所建水库塌岸非线性预测模型预测结果与实际情况较为吻合,预测的左右岸水库塌岸和实地调查塌岸相对误差分别为1.92%和2.15%,岳城水库左右岸塌岸宽度预测平均值分别为31.8m和36.9m;水库下游预测塌岸速度小于1m/a,中上游塌岸速度大于1m/a,下游库岸态势较为稳定,中上游塌岸趋势将加剧。研究成果对于水库塌岸非线性预测及岳城水库塌岸防治和岸坡加固设计提供参考价值。

Consider the nonlinear reservoir collapse prediction to construct nonlinear prediction based on BP neural network model of reservoir bank collapse model,Yue Zhang River Valley within the city reservoir for the study, combined with data on field survey, conducted for Yuecheng reservoir bank collapse forecast.The re- suits show that：the construction of the reservoir bank collapse nonlinear predictive model predictions with the actual situation is more consistent ,predictable and field surveys of both banks reservoir bank collapse bank collapse relative error of 1.92% and 2.15% ,both banks Yuecheng reservoir bank collapse the width of the av- erage forecast was 31.8m and 36.9m;predicting reservoir bank collapse downstream speed is less than 1m/ year,in the upper reaches of bank collapse faster than 1m/year downstream reservoir bank situation is more stable, the trend will be exacerbated by the upstream bank collapse.Research for nonlinear prediction of reser- voir bank collapse and yuecheng reservoir bank collapse prevention and slope reinforcement designed to pro- vide a reference value.