基于POD-DNN代理模型的闸墩锚索有效预应力反演

Inversion of Effective Prestress of Gate Pier Anchor Cables Based on POD-DNN Surrogate Model

为掌握锚索的有效预应力,确保混凝土预应力闸墩的安全可靠,建立了结合PSO算法的POD-DNN高效反演方法。使用LHS方法在设计空间中采样,并通过有限元计算相应的响应获得数据样本集;利用本征正交分解对数据集进行降维,建立POD-DNN代理模型;结合PSO算法,对混凝土预应力闸墩锚索有效预应力进行了反演。结果表明,该方法能够准确快速地反演锚索的有效预应力,适用于大体积混凝土水工建筑物的反演问题。

In order to understand the effective prestress of anchor cables and ensure the safety and reliability of concrete pre-stressed gate piers,an efficient inversion method based on Proper Orthogonal Decomposition(POD)-Deep Neural Network(DNN)surrogate model and Particle Swarm Optimization(PSO)algorithm is constructed.The Latin Hypercube Sampling(LHS)method is used to allocate sample points in the design space,and the data sample set is obtained by calculating the corresponding response with finite elements.The proper orthogonal decomposition is used to reduce the dimensionality of the dataset,which is used to build the POD-DNN surrogate model.Combining with the PSO algorithm,this model is applied to invert the effective prestress of concrete pre-stressed gate pier anchor cables.The results show that this method can accurately and quickly reverse the effective prestress of anchor cable,and is suitable for the inversion of large mass concrete hydraulic structures.