位移反分析的粒子群优化-高斯过程协同优化方法

针对采用随机全局优化技术进行岩土工程位移反分析存在数值计算量大、效率低的问题,将粒子群优化算法与高斯过程机器学习技术相结合,提出了位移反分析的粒子群优化-高斯过程协同优化方法。该方法利用全局寻优性能优异的粒子群优化算法进行寻优的基础上,采用高斯过程机器学习模型不断地总结历史经验,预测包含全局最优解的最有前景区域,通过提高粒子群搜索效率并降低适应度评价次数,进而有效地降低位移反分析过程中的数值计算工作量。多种测试函数的数学验证和工程算例的研究结果表明该方法是可行的,与传统方法相比较,可显著地降低位移反分析的计算耗时。

A cooperative optimization method based on particle swarm optimization and Gaussian process for displacement back analysis

Aiming to the problem about expensive cost and low efficiency of stochastic global optimization technology for displacement back analysis in geotechnical engineering, a novel cooperative optimization algorithm based on particle swarm optimization (PSO) algorithm and Gaussian process (GP) machine learning for back analysis is proposed. In order to reduce the cost of numerical calculation during displacement back analysis, the new method not only takes advantage of the global optimization performance of particle swarm optimization with quick convergence, but also uses GP model to summarize the historic experience during searching optimum solution and predicting the most perspective zone for guiding the flying of particle swarm. The results of performance analysis for different benchmark functions and application to a tunnel engineering show that the method is feasible to reduce remarkably computational time-consuming for solving problem of displacement back analysis based on numerical calculation.