摘要
针对大体积混凝土热力学参数反分析计算效率低的问题,提出了响应面遗传算法。该方法采用正交试验设计的拉丁方方法进行热力学参数组合,采用含交叉项的二次多项式作为温度响应面函数,根据有限元计算的不同参数组合的结构温度响应值,通过回归分析确定响应面函数的待定系数,利用响应面函数和实测温度建立参数反分析目标函数,最后应用遗传算法搜索最优参数组合。算例表明,采用响应面遗传算法反演的参数,计算温度与实测温度总体吻合较好;响应面遗传算法与单一采用遗传算法的计算结果比较接近,但计算量减少一半。
For solving the problems that back analysis of mass concrete thermal parameter is of low efficiency, the response surface genetic algorithm is put forward. In the proposed method, thermal parameters combination is designed by Latin Square method of orthogonal test, and complete quadratic polynomial is used as the temperature response surface function, then the structure temperature values under different parameter combinations are calculated using finite element method, based on which.the coefficients of the response surface function are obtained. The objective function of parameter back analysis is established by measured temperature and response surface function, finally the optimal parameter combination is searched using genetic algorithm. Numerical examples show that the calculated temperatures, whose parameters are inverted by response surface genetic algorithm, are in good agreement with measured ones; the results obtained by response surface genetic algorithm is close to the resuits obtained by single genetic algorithm, but the former can reduce about 50% of the calculation work.
出处
《人民长江》
北大核心
2013年第17期80-82,共3页
Yangtze River
关键词
热力学参数
反分析
响应面
遗传算法
混凝土
thermal parameters
back analysis
response surface
genetic algorithm
concrete
