摘要
针对混凝土面板坝渗透系数反演维数高、计算复杂、耗时长的问题,采用正交试验设计构建渗透系数组合与测压点水头组成的学习样本,通过广义回归神经网络(GRNN)建立渗压监测点水头与渗透系数之间的非线性映射关系,并引入粒子群优化算法(PSO)搜寻适合特定工程的光滑因子σ值,提高模型的泛化性和收敛速度,建立了混凝土面板坝渗透系数反演的PSO-GRNN模型,并应用于工程实例。结果表明,基于该模型反演得到的渗透系数取值合理,渗流分析得到的渗压监测点水头与实测值相对误差最大为3.64%,精度满足工程需要。
Considering the characteristics of inversion problems of concrete face dam,including high dimensionality,complex calculation and excessive calculation time,the orthogonal experimental design was used to construct the learning sample composed of the combination of permeability coefficient and the water head of pressure measuring point.The nonlinear mapping relationship between the water head at monitoring points and permeability coefficient was established by general regression neural network(GRNN),and the particle swarm optimization(PSO)algorithm was used to search for the smoothing factorσsuitable for the specific project to improve the generalization and convergence speed of the model.The PSO-GRNN model for the inversion of the permeability coefficient of concrete face dam was established,and was applied on an engineering example.The results show that the value of permeability coefficient obtained by inversion based on the model is reasonable,and the maximum relative error between the calculated value of water head at monitoring points obtained by seepage analysis and the measured value is 3.64%,and the accuracy meets the needs of engineering.
作者
李皓璇
沈振中
张文兵
LI Hao-xuan;SHEN Zhen-zhong;ZHANG Wen-bing(College of Water Conservancy and Hydropower Engineering,Hohai University,Nanjing 210098,China;College of Ocean Science and Engineering,Shanghai Maritime University,Shanghai 201306,China)
出处
《水电能源科学》
北大核心
2023年第5期67-70,80,共5页
Water Resources and Power
基金
国家自然科学基金项目(52179130)
中央高校基本科研业务费专项资金项目(B210203065)。
