Tri-Fused AI Architecture for Real-Time PCB Defect Intelligence Using Cross- Modal Agreement of Visual, Thermal, and Electrical Signatures

Printed Circuit Board (PCB) inspection is a vital process to ensure safety, safety and performance of electronic devices in industries ranging from automotive, the aerospace industrytothemanufacturingconsumerelectronicssector.Tradi- tionalinspectiontechniques,suchasmanualinspectionandAOI, remain to limited surface-level anomalies and are susceptible to humanerrorenvironmentalconstraint.Inthispaper,wepropose a multi-modal AI-integrated PCB defect detection framework based on visual, thermal and electrical sensors for improved accuracyaswellasindustrialrobustness.Thesystemisbased on a model deep learning detector, YOLOv8, applying it to an ESP32-CAM for visual inspection while a MLX90614 infrared sensor populates thermographic anomalies detection and the ArduinoUnoallowingelectriccontinuityverification.Integration is achieved throughough a Flask-based centralized dashboard, enabling real-time defect visualization, analytical reporting, and automated Pass/Fail classification. Experiments have validated such a system with an average detection accuracy of 91.88% under various circumstances, which indicates that the system satisfiesIt meets the requirement of zero-compliant smart man- ufacturing environments. The work also discusses comparisonof the currently available PCB inspection methods, system stem design for future expansions and cross-modal fufusion sioniming to lower false alarms and incropoperationallustness.