摘要
降雨对土石坝渗流量监测值影响明显,评判土石坝的渗流性态必须在实测渗流量中扣除降雨因素。为此,将朴素贝叶斯分类算法、回归分析和LSTM模型综合应用到土石坝渗流量监控中,实现降雨条件下的土石坝渗流量推理预测。首先以是否出现降雨径流为条件,对数据进行离散化处理,并引入回归分析方法,构建降雨径流~降雨过程数据组;采用LSTM方法构造降雨径流~降雨过程定量模型,最后将上述成果用于不同降雨过程条件下的降雨径流预测。应用结果表明,该方法可以比较准确地剔除由于降雨径流对大坝渗流量测值造成的干扰,可以准确监控土石坝渗流状态。
The rainfall has a significant influence on the monitoring value of seepage flow in earth-rock dams.To judge the seepage behavior of earth-rock dams,the rainfall must be deducted from the measured seepage flow.For this reason,the Naive Bayes classification algorithm,regression analysis and LSTM model are comprehensively applied to the seepage flow monitoring of earth-rock dams to realize the reasoning prediction of seepage flow of earth-rock dams under rainfall conditions.Firstly,the data is discretized based on whether there is rainfall runoff as a condition,and the regression analysis methods are introduced to construct a rainfall runoff-rainfall process data set,then the LSTM method is used to construct a rainfall runoff-rainfall process quantitative model,and finally the above results are used for the rainfall runoff prediction under different rainfall process conditions.The application results show that this method can more accurately eliminate the interference caused by rainfall runoff to the dam seepage flow measurement value,and can accurately monitor the seepage state of the earth-rock dam.
作者
刘远财
林芝
黄孙
崔朋飞
任哲
卢林晶
LIU Yuancai;LIN Zhi;HUANG Sun;CUI Pengfei;REN Zhe;LU Linjing(Zhejiang Zheneng Beihai Hydropower Co.,Ltd.,Hangzhou 310009,Zhejiang,China;Large Dam Safety Supervision Center,National Energy Administration,Hangzhou 311122,Zhejiang,China)
出处
《水力发电》
CAS
2021年第6期91-94,119,共5页
Water Power
关键词
渗流量
预测模型
回归分析
朴素贝叶斯分类算法
LSTM模型
土石坝
seepage flow
prediction model
regression analysis
Naive Bayes classification algorithm
LSTM model
earth-rock dam
