Scholarlink

<p>This paper proposes an innovative iterative method that combines Kriging interpolation with the Sparrow Search Algorithm (SSA) to tackle the inverse problem of 3D defect reconstruction in Eddy Current Non-Destructive Testing (ECNDT). Firstly, the proposed approach utilizes Kriging as a surrogate model for ECNDT forward simulation, which significantly decreases the computational expenses associated with physics-based models. To be specific, in B-scan, an individual Kriging model is built with the impedance variations in each sampling position of the coil for mapping the relationships between them. It reduces the required number of physical simulations for the impedance variations while maintaining good accuracy. Secondly, in analyzing the ECNDT inverse problem, the SSA, which is less susceptible to being trapped in local optima compared with traditional optimization methods, such as Particle Swarm Optimization (PSO) etc., due to its incorporation of multiple strategies that mimic the behavior of sparrow populations, leads to fast convergence and it optimizes the agreements between predicted and actual responses by minimizing a discrepancy function. Once an optimal agreement is achieved, the input geometrical parameters of the physical model for the Kriging model identify the geometrical characteristics of the sought defects. Validation of this novel method involves testing with both simulation and experiment in several ECNDT cases. The findings confirm the high efficiency and accuracy in 3D defect geometrical reconstruction of the proposed method.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An Efficient Iterative Method for Defect Reconstruction in Eddy Current Non-Destructive Testing Based on Kriging Surrogate Model and the Sparrow Search Algorithm

  • 期刊论文

  • 发表日期：

    2026年3月7日

查看全文
  • Yang Bao,
  • Yiming Liu,
  • Jiuhao Ge,
  • Fei Xu,
  • Fangfang Wang

摘要

This paper proposes an innovative iterative method that combines Kriging interpolation with the Sparrow Search Algorithm (SSA) to tackle the inverse problem of 3D defect reconstruction in Eddy Current Non-Destructive Testing (ECNDT). Firstly, the proposed approach utilizes Kriging as a surrogate model for ECNDT forward simulation, which significantly decreases the computational expenses associated with physics-based models. To be specific, in B-scan, an individual Kriging model is built with the impedance variations in each sampling position of the coil for mapping the relationships between them. It reduces the required number of physical simulations for the impedance variations while maintaining good accuracy. Secondly, in analyzing the ECNDT inverse problem, the SSA, which is less susceptible to being trapped in local optima compared with traditional optimization methods, such as Particle Swarm Optimization (PSO) etc., due to its incorporation of multiple strategies that mimic the behavior of sparrow populations, leads to fast convergence and it optimizes the agreements between predicted and actual responses by minimizing a discrepancy function. Once an optimal agreement is achieved, the input geometrical parameters of the physical model for the Kriging model identify the geometrical characteristics of the sought defects. Validation of this novel method involves testing with both simulation and experiment in several ECNDT cases. The findings confirm the high efficiency and accuracy in 3D defect geometrical reconstruction of the proposed method.