基于BP网络和混合遗传算法的岩土工程位移反分析

Displacement Back Analysis of Geotechnical Engineering Based on BP Network and the Hybrid Genetic Algorithm

岩土工程涉及到的地质条件和岩体特性参数是不完全定量的,采用传统的方法进行岩土工程位移反分析难以得到优化。此文根据技术的发展和自然进化的全局搜索优化算法,经研究提出遗传算法与局部搜索能力很强的单纯形法相结合的混合遗传算法,将大大改善遗传算法的局部性能。在此基础上,又将混合遗传算法与神经网络相结合,建立起智能位移反分析方法,采用MATLAB6.5开发出智能位移反分析程序,并以工程实例验证该位移反分析方法的可行性。

The geological conditions and characteristics of rock mass parameters involved in geotechnical engineering is not entirely quantitative,the displacement back analysis in geotechnical engineering is difficult to be optimized using traditional methods. According to the development of technology and the natural evolution of global search optimization algorithm,by the genetic algorithm with simplex method is presented,which will greatly improve local performance of genetic algorithm. Based on this,combining the hybrid genetic algorithm with neural network,the intelligent displacement back analysis method is established,the intelligent displacement back analysis program is developed using MATLAB6. 5. The feasibility of the displacement back analysis method is verified with an engineering example.