Serving condition assessment of steel spring isolators using force-sensing methods based on embedded smart aggregates and electromechanical impedance (EMI) technique

Abstract

Smart aggregates (SAs) based on electromechanical impedance (EMI) technique demonstrate outstanding performance in structural health monitoring, yet there are few reports on their application in the field of serving condition assessment of steel spring isolators. This paper proposes a force-sensing method based on SA and EMI technique. The serving condition of steel springs is analyzed by embedding the SAs into the concrete base and testing the conductance signals. A structural condition assessment model has been developed to enable baseline-free evaluation of conductance signals. The model first employs an adaptive wavelet threshold denoising algorithm to reduce noise in the original conductance signals. Subsequently, baseline-free classification evaluation of structural states is performed through feature extraction and deep mining modules, dual attention mechanism and multi-layer perceptron. Three data partitioning strategies (random data partitioning, independent loading-unloading cycles and independent test specimens) are used to evaluate the model’s generalization ability. The results show that the model achieves accuracy, F1-macro and F1-weighted scores of 0.9574, 0.9571 and 0.9575, respectively, on the reserved test set for strategy 1. For strategy 2, these scores are 0.9063, 0.9033 and 0.9060, respectively. In the eight outer-layer experiments for strategy 3, the average accuracy is 0.7625 ± 0.0291. The ablation results show that removing the dual attention mechanism reduces the model’s accuracy by 0.0852. Furthermore, visualizations of the feature space indicate that the model is capable of applying type-specific, non-uniform weighting to feature locations and feature channels based on the input signal. The research findings validate the feasibility of using the proposed method to conduct baseline-free assessment of the in-service condition of steel spring isolators under laboratory conditions, providing a reference for related research works.