Hybrid AI–Taguchi–ANOVA for Thermographic Monitoring of Electronic Devices #WorldResearchAwards
Introduction
Printed circuit boards (PCBs) are fundamental to modern electronic systems, and their reliability directly affects the performance and safety of critical applications. Undetected defects in PCBs can evolve gradually, leading to unexpected failures and costly downtime. Conventional monitoring techniques, often limited to simulations or surface-level measurements, lack the capability for real-time fault detection and predictive maintenance. This research addresses these limitations by introducing an integrated framework that combines infrared thermography (IRT), artificial intelligence (AI), and Taguchi–ANOVA statistical methods to enable accurate, real-time diagnosis of thermal anomalies in operating PCBs.
Infrared Thermography for Real-Time PCB Monitoring
Infrared thermography serves as a non-contact, non-destructive technique for capturing thermal signatures of PCBs during normal operation. By visualizing heat distribution and thermal stresses, IRT reveals hidden defects such as short circuits, overloads, and material degradation that are not detectable through conventional electrical testing. In this research, real-time IR acquisitions form the foundation of the monitoring system, ensuring continuous observation of PCB health under actual operating conditions.
AI-Based Thermal Anomaly Detection Framework
The acquired infrared data are used to construct a specialized dataset for training AI models. A U-Net architecture is employed for precise segmentation of thermal anomalies, enabling accurate localization of abnormal heat patterns on PCBs. Subsequently, a Multilayer Perceptron (MLP) classifier analyzes the segmented regions to classify heat distribution behaviors associated with potential faults. This hybrid AI pipeline enhances detection accuracy while maintaining computational efficiency suitable for real-time deployment.
Taguchi Method for Hyperparameter Optimization
To systematically optimize the AI model performance, the Taguchi design of experiments is applied. This method efficiently explores the influence of multiple hyperparameters with a reduced number of experimental runs. By identifying the optimal combination of model parameters, the Taguchi approach ensures robustness and consistency of the AI system, minimizing variability in performance when exposed to different thermal operating conditions.
ANOVA-Based Statistical Validation of AI Performance
Analysis of Variance (ANOVA) is used to quantify the contribution of each selected factor to the F1-score response. This statistical evaluation provides clear insight into which hyperparameters significantly influence model performance. ANOVA not only validates the optimal configuration identified by the Taguchi method but also enhances the interpretability and credibility of the AI-driven diagnostic framework.
Novelty, Impact, and Research Significance
The key novelty of this research lies in the seamless integration of real-time infrared thermography, an interpretable AI pipeline, and a Taguchi–ANOVA statistical framework. This combination enables both optimization and rigorous validation of AI performance under real operating conditions. The proposed approach significantly advances predictive maintenance strategies for PCBs, offering a reliable, scalable, and statistically validated solution for early fault detection in electronic systems.
Global Particle Physics Excellence Awards
Get Connected Here:................
Twitter: x.com/awards48084Blogger: www.blogger.com/u/1/blog/posts/7940800766768661614?pli=1Pinterest: in.pinterest.com/particlephysics196/_created/Tumbler: www.tumblr.com/blog/particle196
Hashtags
#infraredthermography #pcbmonitoring #artificialintelligence #thermalimaging #predictivemaintenance #realtimefaultdetection #machinelearning #unet #mlp #taguchimethod #anova #electronicsreliability #smartmanufacturing #qualityengineering #thermalanomalies #industrialai #digitaldiagnostics #embeddedystems #researchinnovation #worldresearchawards

Comments
Post a Comment