摘要
对大坝变形情况进行预测,明确大坝的实际状况是保证其长期安全稳定运行的关键之一,目前研究中普遍存在预测精度不足以满足实际需求的问题。为此,将长短时记忆网络(Long and Short-term Memory Network,LSTM)模型引入大坝变形预测的研究,并利用自适应人工鱼群算法(Adaptive Artificial Fish School Algorithm,AAFSA)对模型的参数进行优化,以实际工程的数据对模型进行了实例验证,并将该模型与LSTM模型的性能进行对比,结果表明,优化后模型的平均绝对误差、平均相对误差、平均绝对百分比误差、均方根误差以及拟合度分别为0.225 9、0.031 6、0.289 2、0.054 7以及94.51%,即优化后的模型预测精度最高且误差最小,稳定性最好,从而为大坝的安全分析提供了新的借鉴。
The actual condition of the dam is one of the key indicators to ensure its long-term safe and stable operation.The deformation prediction of the dam is the current research hotspot. The current research generally has the problem of insufficient prediction accuracy to meet actual needs. For this reason,the Long and short-term memory network( LSTM) model is introduced into the study of dam deformation prediction,and adaptive artificial fish schools algorithm( AAFSA) are used to optimize the parameters of the model,which is then validated with actual engineering data. The results were compared with the LSTM model. The results show that the average absolute value of the he error,average relative error,average absolute percentage error,root mean square error,and degree of fit are 0. 225 9,0. 031 6,0. 289 2,0. 054 7,and 94. 51%,respectively. Which means the optimized model has the highest prediction accuracy,the smallest error,and the best stability. This paper can provide a new reference for dam safety analysis.
作者
杨宗仁
杨凯
王健
YANG Zongren;YANG Kai;WANG Jian(Shaanxi Water Conservancy Electric Power Reconnaissance Design Research Institute,Xi'an,Shaanxi 710001,China;Xi'an University of Architecture and Technology,Xi'an,Shaanxi 710055,China)
出处
《水利与建筑工程学报》
2022年第1期98-102,176,共6页
Journal of Water Resources and Architectural Engineering
关键词
大坝变形预测
人工鱼群算法
长短时记忆网络
dam deformation prediction
artificial fish school algorithm
long and short-term memory network
