您的位置:首页  > 论文页面

混凝土热学参数最优选取方法研究

发表时间:2012-07-31  浏览量:1180  下载量:542
全部作者: 段亚辉,吴家冠
作者单位: 武汉大学水利水电学院;广州市水务工程质量安全监督站
摘 要: 现实施工中由于人为、自然等因素的干扰,随机因素也会导致计算参数的不稳定性,因此必须综合考虑随机因素影响下的概率统计最优的参数取值。根据反分析情况,统计出各参数概率分布,并通过神经网络训练出各参数与计算结果的非线性映射关系,将各监测值作为输入神经元,反算出每个参数的值,由此可以计算出每个参数在考虑参数间交互作用影响情况下的取值波动范围。然后按一定的变化率改变主参数(自变量),根据交互作用系数计算其他影响因素(因变量)。在各参数概率分布型的基础上,就可以计算出每次调整参数所对应的各参数概率密度值,对比每次调整的情况便可确定出最理想的参数组合。通过实例验证了参数组合最优选取后,数值计算结果与监测误差减小。
关 键 词: 水工结构;概率统计;温度场;BP神经网络;敏感度
Title: Study on optimal selection method of concrete thermal parameters
Author: DUAN Yahui, WU Jiaguan
Organization: College of Water Resources and Hydroelectric Engineering, Wuhan University; Supervisory Station of Guangzhou Waterworks Construction
Abstract: Human and natural factors interference will lead to instability of calculate parameters in real construction. Therefore, it is necessary to take the parameters’ optimization value based on probability statistic into account under the influence of random factors. The paper figures out each parameter probability distribution based on back analysis data, and trains the nonlinear relationship between parameters and results under the influence of various parameters interaction. Considering field monitoring value as the input neurons, the paper calculates the value of each parameter, and gets the value range for each parameter considering their interaction. Then, it changes the main parameters by a certain rate, calculates other parameters on the basis of interaction coefficients. So that, parameters joint probability can be got for each parameters adjustment case according to each parameter probability distribution, and the parameters to be optimized can be determined by contrasting each parameters joint probability case. Example illustrates that the difference between numerical calculate value and field monitoring value decreases by adopting the parameters optimization method.
Key words: hydraulic structure; probability statistic; thermal field; BP neural network; sensitivity
发表期数: 2012年7月第14期
引用格式: 段亚辉,吴家冠. 混凝土热学参数最优选取方法研究[J]. 中国科技论文在线精品论文,2012,5(14):1376-1383.
 
0 评论数 0
暂无评论
友情链接