Boosting Fiber Layup Defect Classification: A Comprehensive Analysis of Explainable AI Techniques for Neural Networks

This research study presents a machine learning model capable of accurately identifying and classifying various manufacturing defects. The model demonstrates exceptional performance in recognizing defects such as alignment issues, fluid leaks, welding defects, electrical system failures, and imperfections. The model achieves high precision and recall values, ranging from 81.68% to 93.33% and 87.50% to 89.38%, respectively, indicating its ability to correctly identify and capture positive examples of each defect type. The Fl-scores, ranging from 85.01 % to 90.32%, further validate the model's balanced performance. The study also provides support percentage values to standardize the data and illustrate the relative representation of each fault type. The consistent accuracy values across all classes, including macro, weighted, and micro averages, highlight the model's reliability and generalizability. The findings of this research demonstrate the potential of machine learning to improve manufacturing quality control processes, leading to enhanced product quality and operational efficiency.