# Electromechanical Impedance Sensing Under Humid Conditions: Experimental Insights and Compensation Using Machine Learning

## TL;DR

This study shows how humidity affects EMI sensing in concrete structures and proposes a machine learning method to compensate for humidity effects.

## Contribution

A novel machine learning approach using 1D-CNN is introduced to estimate humidity from EMI data for improved structural health monitoring.

## Key findings

- Humidity significantly affects EMI RMSD scores, impacting damage detection accuracy.
- A 1D-CNN model estimated humidity with a MAE of 2.14%RH using EMI's reactive component.
- The imaginary part of EMI signatures can detect humidity effects in concrete structures.

## Abstract

This work investigates the effect of ambient humidity on the Electromechanical Impedance (EMI) signatures of steel-reinforced concrete (RC) for structural health monitoring (SHM). The influence of varying relative humidity (%RH) is quantified using three RC blocks containing piezoelectric sensors bonded to the steel reinforcements of the RC blocks. We show that the the Root Mean Squared Deviation (RMSD) score is strongly affected by humidity, highlighting the need to address humidity effects to achieve robust damage detection using EMI. Using the reactive component of the EMI (X) in the range of 20 kHz and 120 kHz, a three-layer one-dimensional convolution neural network (1D-CNN) was able to estimate ambient %RH between 20% and 80%, with a Mean Absolute Error (MAE) of 2.14%RH. The results highlight the significant impact of humidity on EMI-based SHM and suggests that the imaginary part of the EMI signature can be used to detect the effect of humidity. This work provides a foundation for more robust SHM systems in humidity-varying environments applicable to a wide range of concrete infrastructure.