Physically Guided Interpretation of Piezoelectric Signal Features for Concrete Curing Assessment Under Impact and Controlled Excitation

This study examines physically interpretable PZT signal features for concrete curing in two independent campaigns using rebound-hammer impact and controlled burst-sine actuation. Spearman correlations tracked feature medians across common Days 3, 7, 14, and 21 using campaign-specific compressive strengths as curing-progress references. Differences in batch, replication, and measurement configuration prevent isolation of excitation-method effects. Under impact excitation, PZT-B/Extra gave coefficients of 0.95, −1.00, 0.40, and 0.80 for peak count, time of maximum response, zero-crossing rate, and relative 2–4 kHz energy. Under controlled actuation, the side-actuation–Middle-sensor layout led screening across 70–110 kHz: RMS amplitude had a median coefficient of −0.80, meeting the criterion at three frequencies, while two-bin Fourier entropy had +0.87 and met it at all four. Additional within-method analyses showed Method 1 peak-count and response-timing directions persisted with Day 28, while zero-crossing-rate and relative-energy associations weakened; for Method 2, negative RMS persisted whereas Fourier entropy and layout ordering were age-set sensitive. The 120 kHz condition was retained for sensitivity analysis. These exploratory associations track curing-related mechanical development, not the complete curing state, and do not establish strength-specific sensitivity. PZT sensing complements standardized strength testing; no predictive or explainable-AI model was evaluated.