基于DEPSO混合智能算法的岩土体应力-渗流-损伤耦合模型多参数反演研究

Multi-parameters Inversion of Stress-Seepage-Damage Coupling Model Based on DEPSO Intelligent Algorithm

随着复杂岩土工程问题研究的深入,开展关于岩土体应力-渗流-损伤多参数反演问题的研究具有显著意义.引入差异进化算法（Differential Evolution Algorithm,DE）—粒子群算法（Particle Swarm Optimization,PSO）混合智能算法DEPSO到岩土工程应力-渗流-损伤耦合问题参数反演研究之中,该混合智能算法可以保证两个种群的相互协调开发能力和探索能力之间的平衡,维持整个种群的多样性,降低陷入局部最优或早熟停滞的风险.通过两个经典测试函数,针对DEPSO、PSO以及DE这3种算法的收敛性、鲁棒性进行对比分析,并讨论在DEPSO算法中DE控制参数、差异策略的控制作用.将岩体应力-渗流-损伤耦合模型与DEPSO混合智能算法结合起来,建立耦合模型多参数反演方法,并采用C＋＋编制程序,构建基于DEPSO混合算法的耦合多参数反分析平台.依据应力-渗流-损伤耦合正分析计算的结果作为＂假想实测值＂,进行多参数反演分析,以及方法和程序的验证.研究结果表明：所建立的基于DEPSO混合算法的耦合多参数反分析方法,可以同时反演多个力学参数和渗透参数,能够较好地解决岩土工程多参数反演问题,在计算效率方面高于单一智能算法,具有较好的反演精度和鲁棒性,是一种新颖和高效的反分析方法,可以更好地为复杂环境岩土工程动态施工提供帮助和依据.

Research on the stress-seepage-damage parameter inversion problem of the rock is a great significance with the development of the complex geotechnical engineering problems. DEPSO hybrid intelligent algorithm is introduced into the stress-seepage-damage coupling parameters inversion based on differential evolution algorithm and particle swarm optimization. The hybrid intelligent method can make sure the two interact with each other coordinated development of population of balance between ability and the ability to explore,to maintain the diversity of the whole population and reduce the risk of fall into local optimum. Two classical test functions are used in view of the convergence and robustness of DEPSO. PSO,DE algorithm and the control parameters and the importance of the difference strategy of DEPSO algorithm are discussed in details. The coupling multi-parameters method is set up combined the stress-seepage-damage coupling model with DEPSO hybrid intelligent algorithm,and the coupling parameters back analysis program is compiled using C ＋ ＋ programming based on DEPSO. A hypothetical measured values are calculated based on the stress-seepage-damage coupling program to analysis multi-parameter inversion for the method and program validation. Research results show that the multi-mechanics parameters and penetration can be inversion by the coupling parameter back analysis method based on the established DEPSO algorithm at the same time,and the problem of multi-parameters inversion can be well solved in geotechnical engineering. In terms of computation efficiency is higher than the single intelligent algorithm. It is a new and efficient back analysis method which has the good inversion precision and robustness. To provide the help and basis for dynamic construction in complex environmental geotechnical engineering.