混凝土浇筑仓一期冷却通水措施优选初探  

Research on Optimization of First Stage Cooling Measures of Concrete Casting Warehouse

作　　者：练迪 黄耀英[1] 郑晓晖[1] 高祥泽 陈义涛[1] 吕晓曼[1] 

机构地区：[1]三峡大学水利与环境学院,湖北宜昌443002 [2]温州市水利电力勘测设计院,浙江温州325011

出　　处：《中国农村水利水电》2014年第11期90-92,97,共4页China Rural Water and Hydropower

基　　金：国家自然科学基金项目(51209124)

摘　　要：针对混凝土浇筑仓一期冷却是一个与通水水温、通水流量和通水时长等有关的复杂多因素问题,而现有文献报道的多是关于单个因素的敏感性分析。基于西南某混凝土坝施工期的实际施工条件,采用有限元温度场仿真和神经网络技术建立了混凝土浇筑仓一期冷却通水措施优选模型。首先采用均匀设计原理对通水措施进行组合设计,然后在设计通水措施组合下进行温度场仿真计算,从而获得学习样本,进而建立描述混凝土浇筑仓最高温度和最大降温速率与通水措施非线性关系的广义回归神经网络模型,将已确定的浇筑仓最高温度和最大降温速率输入该模型中,即可实现一期冷却通水措施智能优选。算例分析表明,所建立的通水措施智能优选广义回归神经网络模型是可行的。Concrete cooling is a complex multi-factor issue related to water temperature ,water flow and water period .Cooling measures optimization model is established by using temperature simulation and neural networks technology ,under the construction condition of a southwest concrete dam .First ,the principle of uniform design is used to combine cooling parameters .Then ,temperature field simulation analysis under cooling combination is proceeded to make up of neural network learning samples .On the basis of the above steps ,network models are established to describe the nonlinear relationship between highest temperature and maximum cooling rate and cooling measures .Casting warehouse highest temperature and the maximum cooling rate input of trained network ,the cooling measure can be automatically selected .Example analysis shows that the neural model established to select cooling measures is effective .

关 键 词：一期冷却 均匀设计 广义回归神经网络 通水措施优选 

分 类 号：TV431[水利工程—水工结构工程]