基于LMBP神经网络的土石坝渗流压力预测

发布时间：2018-04-27 19:07

本文选题：土石坝 + 渗流压力　；参考：《人民黄河》2017年08期

【摘要】：渗流压力是反映大坝工作状态的重要物理量,对渗流压力进行预测分析可以及时了解大坝渗流状况和趋势。为克服标准BP算法收敛速度慢、泛化能力弱和计算量大等不足,引入LM算法优化标准BP神经网络的权值和阈值,提高BP神经网络对土石坝渗流压力的预测效果。根据渗流分析,给出了渗流压力的统计模型,由统计模型选取上下游水位、降雨和时效作为神经网络输入层因子,以渗流压力作为输出层因子,建立了3层LMBP神经网络大坝渗流压力预测模型。利用MATLAB进行了多组仿真试验,确定了使本次渗流压力预测效果更好的训练样本数据量区间。以渗流压力实测数据及同期库水位和降雨资料作为训练样本,在选取适当数据量的训练样本的基础上,运用LM算法对BP网络进行训练,利用测试样本对训练好的神经网络进行测试。将同结构的LMBP神经网络和标准BP神经网络应用于某土石坝渗流压力的预测中,应用结果表明,LMBP神经网络收敛速度更快、拟合和预测精度更高,在土石坝渗流压力分析和预测应用方面是可行的。
[Abstract]:Seepage pressure is an important physical quantity reflecting the working state of the dam. The prediction and analysis of seepage pressure can understand the seepage condition and trend of the dam in time. In order to overcome the shortcomings of the standard BP algorithm, such as slow convergence speed, weak generalization ability and large amount of calculation, LM algorithm is introduced to optimize the weight and threshold of the standard BP neural network to improve the prediction effect of BP neural network on the seepage pressure of earth-rock dam. According to the seepage analysis, the statistical model of seepage pressure is given. The upstream and downstream water level, rainfall and aging are selected as input layer factors of neural network, and seepage pressure is taken as output layer factor. A three-layer LMBP neural network model for predicting seepage pressure of dams is established. A series of simulation experiments are carried out with MATLAB to determine the interval of the training sample data which can make the prediction effect of seepage pressure better. Taking the measured seepage pressure data, the water level and rainfall data of the same reservoir as the training samples, the LM algorithm is used to train the BP network on the basis of selecting the appropriate training samples. Test samples are used to test the trained neural network. The LMBP neural network with the same structure and the standard BP neural network are applied to the prediction of seepage pressure of a certain earth-rock dam. The application results show that the neural network converges faster, and the fitting and forecasting accuracy is higher. It is feasible to analyze and predict seepage pressure of earth-rock dam.
【作者单位】：南京水利科学研究院;水利部大坝安全管理中心;江苏省水利厅;
【基金】：水利部公益性行业科研专项(201501033) 国家重点研发计划项目(2016YFC0401600) 江苏省水利科技项目(2015010) 广西水利厅科技项目(201614)
【分类号】：TV223.4;TV641

【相似文献】