ML-Enhanced Self-Healing Fiber-Reinforced Polymer Composites with Embedded IoT Sensors for Damage Prediction

Fiber-reinforced polymer (FRP) composites are widely used in aerospace and structural systems;
nevertheless, the potential for microcracking and fatigue-induced performance degradation remains an
obstacle with respect to improved service life. Traditional self-healing methods, while performing well
on a chemical level, often lack real-time diagnostic awareness and adaptive control. To circumvent
this, we developed a machine-learning augmented self-healing FRP composite, in which a DCPD–
Grubbs catalytic matrix was combined with IoT sensor network capability and a hybrid CNN–LSTM
predictive model. This framework identifies, interprets, and performs in-situ actions, enabling the
material to become an intelligent closed-loop system capable of self-healing and managing damage
automatically instead of passively. Our work differs from previous studies, which focus on the simulated
static analysis of FRP composites. The synergy between IoT and CNN–LSTM learns continually from
the multi-sensor data and predicts failure mechanisms and severity of internal damage based on load
cycles prior to mechanical failure. The reported experiments demonstrate an average healing capacity
of 90.6%; an R² of 0.989 prediction accuracy; and 11 ms as the latency to decision outcomes, exceeding
state-of-the-art indications for action in thermally and magnetically initiated self-healing composites.
The model's adaptive retraining preserved accuracy across multiple healing cycles without increasing
energy consumption; thus, it is appropriate for use over the long term. These results constitute a
paradigm shift in polymer composites design – from passive structural materials to active, data-based
agents capable of self-diagnosing and repairing. The attribute proposed overcomes barriers to
sustainable, low-power, and intelligent composite
systems and is aligned with the Journal of Polymer
Composites' current vision for follow-on research
on multifunctional and self-adaptive polymer
composite architectures.