Multi-parameter inverse analysis of concrete dams using kernel extreme learning machines-based response surface model

Inverse analysis by finite element model (FEM) based on measured displacement data is a popular approach for parameter identification of concrete dams. FEM-based inverse analysis method is economical and effective, but with some limitations such as complex computation and time-consuming. This paper presents a kernel extreme learning machine (KELM)-based response surface model (RSM) for parameter inverse analysis of concrete dams. The FEM is replaced by KELM-based RSM to explore the relationships between material parameters and displacement response of dam-foundation systems. Sample data of KELM-based RSM are generated by the efficient sampling technique Latin hypercube sampling. Subsequently, a novel optimization algorithm Jaya is adopted to minimize the objective function for material parameters identification. The effectiveness and practicability of the proposed approach are verified by two cases and two real concrete gravity dams in operation with sufficient monitoring data. Comparative studies with direct finite element method and several optimization algorithms were also performed to demonstrate the superiority of the proposed approach over commonly used techniques. Results show that the proposed KELM-based RSM is a simple and efficient way to achieve high accuracy in parameter inverse analysis of concrete dams at a low computation cost.

Automated summary

This paper proposes a response surface model based on kernel extreme learning machine for parameter inverse analysis of concrete dams. The effectiveness of this method is verified through experiments, showing high accuracy at a low computation cost.