论文部分内容阅读
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。
Aiming at the problem of large amount of numerical calculation and low efficiency in the geo-engineering displacement back analysis by stochastic global optimization, the combination of particle swarm optimization algorithm and Gaussian process machine learning technology is proposed. The particle swarm optimization-Gaussian process Collaborative optimization method. Based on the optimization of particle swarm optimization algorithm with global optimization, this method uses the machine learning model of Gaussian process to continually summarize the historical experience and predict the most promising region containing the global optimal solution. By improving the efficiency of particle swarm optimization, And reduce the number of fitness evaluation, and thus effectively reduce the numerical analysis of displacement back analysis process workload. The results of mathematical verification and engineering studies of various test functions show that this method is feasible and can significantly reduce the time-consuming calculation of displacement back analysis compared with the traditional methods.