Approximation model of geometric factors for real-time assessment of on-site electrical resistivity of concrete

The measurement of electrical resistivity is widely used in civil engineering for structural inspection and material characterisation. Measuring devices employ electrodes to apply current and measure potential differences. The resistivity of concrete structures is determined from the applied current, measured voltage, and a geometric factor influenced by the structure’s geometry, electrodes spacing and positioning, and the presence of rebars. Several methods exist to determine this factor, including analytical approaches, finite element modelling, and experimental procedures. However, the latter two approaches are often time-consuming, and in practice—particularly during on-site inspections—a rapid evaluation of the geometric factor is required to adapt the measurement strategy in real time. This study introduces a fast approximation model for computing the geometric factor in four-probe devices using the Wenner array. The model is based on a neural network trained on a database generated by finite element simulations. It covers cylinders and prisms of various sizes, with and without rebars, and generalises results from finite-size slabs to large-scale structures. The model’s predictions were validated against published data. Furthermore, a graphical user interface (GUI) application has been developed to enable rapid, real-time evaluation of geometric factors, supporting researchers and inspectors in non-destructive testing (NDT).

电阻率测量在土木工程领域被广泛应用于结构检测与材料表征。测量设备通过电极施加电流并测量电位差。混凝土结构的电阻率可通过施加电流、测得电压，以及受结构几何形状、电极间距与布置方式、钢筋存在情况影响的几何因子计算确定。目前存在多种确定该几何因子的方法，包括解析法、有限元建模法与实验测试法。然而，后两种方法往往耗时较长；在实际应用中，尤其是现场检测场景下，需快速评估几何因子以实时调整测量策略。本研究提出了一种用于计算采用温纳阵列（Wenner array）的四探针装置几何因子的快速近似模型。该模型基于在有限元模拟生成的数据集上训练得到的神经网络，覆盖不同尺寸的带钢筋与不带钢筋的圆柱与棱柱结构，并可将有限尺寸板的结果推广至大型结构。该模型的预测结果已通过已发表数据验证。此外，本研究还开发了一款图形用户界面（Graphical User Interface, GUI）应用程序，可实现几何因子的快速实时评估，为从事无损检测（Non-Destructive Testing, NDT）的研究人员与检测人员提供支持。