Piezoelectric Sensor Array with Lamb Wave Propagation and Hybrid CNN-SVM Classifier for Structural Health Monitoring of CFRP Aerospace Panels

Carbon Fibre Reinforced Polymer (CFRP) composites have become the dominant structural material in modern commercial aircraft, constituting 52% of the Boeing 787 airframe and 53% of the Airbus A350 by weight, driven by their superior specific stiffness, specific strength, and corrosion resistance relative to aluminium alloys. India's aerospace manufacturing sector, expanding rapidly under the Aatmanirbhar Bharat initiative with HAL's Light Combat Aircraft Mk2, Dornier-228 composite wing, and the DRDO's TAPAS-BH medium-altitude UAV all incorporating CFRP structures, faces an urgent requirement for cost-effective structural health monitoring (SHM) systems that can replace scheduled time-interval inspections with condition-based monitoring that detects damage before it reaches safety-critical severity.This paper presents an embedded SHM system for CFRP panels using a networked array of 16 PZT-5A piezoelectric transducers in pitch-catch configuration, generating 150 kHz five-cycle Hanning-windowed Lamb wave bursts and recording wave propagation responses that encode structural damage information in amplitude, time-of-flight, and frequency content changes. A hybrid CNN-SVM damage classifier processes wavelet scalogram features from the recorded Lamb wave signals to classify four structural states: healthy baseline, delamination, matrix cracking, and foreign object impact damage. The optimised 8-sensor configuration achieves damage detection efficiency of 97.2% and misclassification rate of 2.8%, outperforming threshold-index, pure SVM, Random Forest, and standalone CNN baselines. Electromechanical impedance (EMI) signatures independently detect bolt-loosening with 0.5 N·m torque resolution, providing complementary monitoring of fastener integrity.