A machine learning framework for predicting moisture effects on active sensory piezoelectric flax composite laminates

The increasing use of Flax Fiber-Reinforced Composites (FFRCs) in sustainable engineering applications—particularly in marine environments—demands a deeper understanding of their long-term performance under environmental exposure. In this context, our previous study [1] experimentally investigated how absorbed moisture affects active sensory wave propagation in flax composite laminates. Significant alterations were observed in key signal characteristics, including Time of Flight (ToF), amplitude, and dispersion behavior. Building upon this prior work, the current study extends pitch-and-catch experiments using piezoelectric devices to assess the degradation of mechanical properties over time due to moisture uptake. The experimental campaign includes systematic measurements of stiffness degradation, thickness swelling, and mass gain, which are used to quantify the extent of environmental deterioration in FFRCs. These degradation metrics are then used as inputs to machine learning regression models that aim to predict stiffness loss and dispersion characteristics as functions of moisture uptake. The models are trained on time-resolved experimental data encompassing both moisture diffusion and mechanical response, achieving high accuracy in correlating active-sensory features with material degradation. The proposed methodology establishes a predictive framework for designing environmentally resilient FFRCs with online structural health monitoring capabilities. This work offers valuable insights into moisture-induced degradation mechanisms and contributes to the development of smart, bio-based composite materials for sustainable engineering applications.