Page 89 - 2022年第53卷第2期

［ 26］ FERREIRA，PEDRO M，FARIA E A，et al . Neural network models in greenhouse air temperature prediction
［J］. Neurocomputing，2002，43（1）：51-75 .
［ 27］ SMITH Brian A，RONALD W McClendon，GERRIT Hoogenboom . Improving air temperature prediction with arti⁃
ficial neural networks［J］. International Journal of Computational Intelligence，2006，3（3）：179-186 .
［ 28］ HEIDARI Elham，MOHAMMAD Amin Sobati，SALMAN Movahedirad . Accurate prediction of nanofluid viscosi⁃
ty using a multilayer perceptron artificial neural network（MLP-ANN）［J］. Chemometrics and intelligent laborato⁃
ry systems，2016，155：73-85 .
［ 29］ GARDNER，MATT W，DORLING S R . Artificial neural networks（the multilayer perceptron）—a review of appli⁃
cations in the atmospheric sciences［J］. Atmospheric environment，1998，32（14）：2627-2636 .
［ 30］ LI D，HUANG F，YAN L，et al . Landslide Susceptibility Prediction Using Particle-Swarm-Optimized Multilayer
Perceptron：Comparisons with Multilayer-Perceptron-Only，BP Neural Network，and Information Value Models
［J］. Applied Sciences，2019，9（18）：3664 .
［ 31］ QIAN Q，JIN R，YI J，et al . Efficient distance metric learning by adaptive sampling and mini-batch stochastic
gradient descent（SGD）［J］. Machine Learning，2015，99（3）：353-72 .
［ 32］ 张平 . 基于实时监控的高心墙堆石坝施工仿真与进度控制研究［D］. 天津：天津大学，2011 .
［ 33］ 李 泽 椿 ，毕 宝 贵 ，金 荣 花 ，等 . 近 10 年 中 国 现 代 天 气 预 报 的 发 展 与 应 用［J］. 气 象 学 报 ，2014，72（6）：
1069-1078 .

Study on the simulation model of high-core rockfill dam construction
in alpine region considering the influence of low temperature

ZHANG Jun，YU Jia，REN Bingyu，WANG Xiaoling，YU Peng，LIN Weiwei
（State Key Laboratory of Hydraulic Engineering Simulation and Safety，Tianjin University，Tianjin 300072，China）

Abstract： Low temperature is a key influencing factor that impedes the construction progress of high-core
rock-fill dams in alpine region. However，most of the existing simulation models of construction merely con⁃
sider the effect of temperature on construction indirectly through parameters such as effective construction
time which are acquired from engineering experience or statistical analysis methods. As a result，it is diffi⁃
cult to accurately quantify the impact of low temperature on the construction progress. Besides， such mod⁃
els put little consideration on the heat preservation measures in alpine region，which is incapable of meet⁃
ing the high accuracy demand of the construction simulation of rockfill dam. To solve the above problems，
this paper proposes a simulation model of the high-core rockfill dam construction in alpine region consider⁃
ing the influence of low temperature. First，a temperature time series prediction method based on the Parti⁃
cle Swarm Optimization Multilayer Perceptron （PSOMLP） is established. In this process， the hyperparame⁃
ters of the multilayer perceptron （MLP） are optimized through the Particle Swarm Optimization algorithm
（PSO） to solve the problems encountered in traditional MLP training process，including difficult determina⁃
tion of hyperparameters， low training efficiency and poor accuracy. Then， the prediction method is embed⁃
ded in the construction simulation model to determine the accurate low-temperature shutdown time. Second⁃
ly，based on the Bootstrap method，the activity time of heat preservation measures is sampled and used to
construct the simulation model of the high-core rockfill dam in alpine region which simultaneously consid⁃
ers the influence of the low-temperature shutdown and the added heat preservation measures. Engineering
application results show that the proposed PSOMLP model has higher prediction accuracy than the tradition⁃
al temperature prediction models， reducing the average error rate from 19.74% to 1.21% ， which demon⁃
strates that the qualification of the proposed method in quantifying the influence of low-temperature show⁃
down and added heat preservation measures on construction progress. Therefore， the proposed model pro⁃
vides a new idea for the simulation of high-core rock-fill dam construction in alpine region.
Keywords：construction in cold area；high core rockfill dam；construction simulation；particle swarm opti⁃
mization （PSOMLP）；temperature forecast

（责任编辑：李福田）
— 211 —