分数阶模型模拟混凝土徐变特性的多参数识别

Fractional-order model simulation and multi-parameter identification for the creep behavior of concrete

本研究主要研究混凝土徐变特性的分数阶模型拟合及其多参数识别问题。在模型方面,提出利用修正的分数阶Maxwell模型和分数阶Poynting-Thomson模型两种模型模拟混凝土的徐变实验数据,并将结果进行对比,通过数据拟合和误差分析验证两种模型的有效性。在多参数识别方面,分别采用贝叶斯算法和布谷鸟搜索算法两种算法识别两个模型中的多个参数。研究表明:修正的分数阶Maxwell模型和分数阶Poynting-Thomson模型在刻画混凝土徐变特性中均是有效的;贝叶斯算法和布谷鸟搜索算法在分数阶模型的多参数估计问题中均是可行的,但布谷鸟算法搜索速度更快、误差更小、效率更高,在分数阶模型的多参数识别问题中性能更优。

Focusing on the creep behavior of concrete,the fractional-order model simulation and the corresponding multi-parameter identification are studied and analyzed in this paper.Both the modified fractional Maxwell model and the fractional Poynting-Thomson model are proposed to simulate experimental data of concrete creep.The results are compared,and the validity of the two models is verified by data fitting and error analysis.As for the corresponding multi-parameter identification,both the Bayesian algorithm and the cuckoo search algorithm are used to identify the multiple parameters in the two models.Results show that the modified fractional Maxwell model and fractional Poynting-Thomson model are both effective in describing the creep characteristics of concrete.Both the Bayesian algorithm and the cuckoo search algorithm are feasible in the fractional multi-parameter estimation problem,but the cuckoo algorithm has higher search speed,smaller error,higher efficiency,and better performance in the multi-parameter identification problem of the fractional order model.