基于GPR代理模型和GA-APSO混合优化算法的软基水闸底板脱空反演

Inversion study of soft foundation sluice bottom plate emptying based on a GPR surrogate model and a GA-APSO hybrid optimization algorithm

软基水闸底板脱空是水闸在长期服役期间受水流侵蚀等环境因素影响所产生的一种危害极大且难以察觉的病害。由于其病害部位于水下,传统方法难以检测,该研究提出一种基于高斯过程回归(Gaussian process regression,GPR)代理模型和遗传-自适应惯性权重粒子群(genetic algorithm-adaptive particle swarm optimization,GA-APSO)混合优化算法的水闸底板脱空动力学反演方法,用于检测软基水闸底板脱空。首先,构建表征软基水闸底板脱空参数和水闸结构模态参数之间非线性关系的GPR代理模型;其次,基于GPR代理模型与水闸实测模态参数建立脱空反演的最优化数学模型,将反演问题转化为目标函数最优化求解问题;最后,为提高算法寻优计算的精度,提出一种GA-APSO混合优化算法对目标函数进行脱空反演计算,并提出一种更合理判断反演脱空区域面积和实际脱空区域面积相对误差的指标—面积不重合度。为验证所提方法性能,以一室内软基水闸物理模型为例,对两种不同脱空工况开展研究分析,结果表明,反演脱空区域面积和模型实际设置脱空区域面积的相对误差分别为8.47%和10.77%,相对误差值较小,证明所提方法能有效反演出水闸底板脱空情况,可成为软基水闸底板脱空反演检测的一种新方法。

The floor voiding of a sluice on soft foundation is a very harmful and undetectable disease caused by environmental factors such as water erosion during long-term service of sluice gates et al.Since the damage area is underwater,it is difficult to be detected by traditional methods.A dynamic inversion method based on a Gaussian process regression(GPR)surrogate model and a genetic algorithm-adaptive particle swarm optimization(GA-APSO)hybrid optimization algorithm was proposed to detect the voiding of the soft foundation sluice floor.The GPR surrogate model was constructed to characterize the nonlinear relationship between the emptying parameters and the modal parameters of the sluice structure.Then,a mathematical optimization model for emptying parameter inversion was established based on the GPR surrogate model and the measured modal parameters of the sluice,and the inversion problem was transformed into an objective function optimization solution problem.In order to improve the accuracy of the algorithm,the GA-APSO hybrid optimization algorithm was proposed to perform the inversion calculation of the objective function for the emptying,and a more reasonable index to judge the relative error between the inversion voiding area and the actual voiding area—area non-coincidence was proposed.In order to verify the performance of the proposed method,the physical model of an indoor soft foundation sluice was used as an example,and two different emptying conditions were set for analysis.The results show that the relative errors of the inverse emptying area and the actual set emptying area of the model are 8.47%and 10.77%,respectively,which are rather smaller relative error values.The proposed method can effectively inversely perform the emptying of the sluice bottom plate,which can be a new method for inversion detection of the soft foundation sluice floor.