Compensation for the effect of temperature in the electromechanical impedance method using polynomial regression through parametric inference

Compensating for temperature effects in the electromechanical impedance method is crucial for structural monitoring. Therefore, this work introduces a data processing protocol using polynomial regression modeling to establish baseline impedance signatures at different temperatures from those collected, helping to develop a damage metric. Polynomial models were defined for each analyzed frequency point and evaluated through residual normality, independence, homogeneity, and a minimum coefficient of determination (R²). The ratio between the frequency points that have valid polynomials and the total frequency points per impedance signature corresponded to 89.6% and 92.1% for the storage system profile experiment. For the silo profile experiment, the total frequency points per impedance were 96.1% and 97.6%. It was observed that the number of peaks and valleys in the impedance signatures and the degree of the polynomial influence the number of frequency points with valid data polynomials. The Correlation Coefficient Deviation (CCD) and threshold values derived from the valid polynomials were higher than the measured counterparts and showed consistency with the effective frequency shift (EFS) method. These findings demonstrate that the number of valid polynomials was sufficient to support polynomial regression as a robust and reliable strategy for establishing baselines at unmeasured temperatures.