基于自适应混合遗传算法的 岩石类材料动态参数反演

将多元函数的下山单纯形法嵌入遗传算法,加强了算法的局部搜索能力,并且将交叉概率、变异概率以及 局部搜索概率改进为随遗传个体的适应度值改变和进化代数的增加而调整的自适应型。在此基础上针对岩石 类材料的特性,将数值计算值与冲击压缩实验结果之间的相对均方差作为适应度函数,编制了反演分析算法来 确定岩石类材料考虑损伤的非线弹性动态本构模型的待定特征参数。结果表明:由动态本构方程得到的再生应 力-应变曲线与实验曲线之间具有较好的一致性,从而验证了该反演分析方法是合理可行的以及动态本构模型 对于岩石类材料的适用性。

Inverse Analysis of Dynamic Parameters of Rock Materials Based on Adaptive Hybrid Genetic Algorithms

The hybrid genetic algorithms is composed with the down hill optimization methods embedded in basic genetic algorithms by coding in real number, so capabilities of the local search are enhanced. The crossover probability and mutation probability of the genetic algorithms can be adjusted with the change of fitness functions of individuals. Furthermore, the local search probability is changed with the progress of generations. Characteristic parameters of the nonlinear dynamic constitutive model of granite and concrete are ascertained by the inverse analysis method based on the adaptive hybrid genetic algorithms. The results indicate that the calculated stress-strain curve matches with the experiment curve, and the approach presented here satisfactorily accounts for the features of dynamics of rock and other brittle materials.