Industrial optical networks based PCA for VLC application networks based intelligent recognition system

Abstract The optical communication architecture enables real-time image acquisition and reliable data transfer between sensing, processing, and sorting modules. Industrial automation has become a key part of modern manufacturing, driven by the need for faster production, intelligent monitoring, and quality inspection. This paper discusses a nut-and-bolt recognition and sorting system that uses Principal Component Analysis (PCA) and an Artificial Neural Network (ANN) in MATLAB. The system features a machine vision setup with a webcam, proximity sensor, conveyor belt, and ANN classifier for shape recognition. PCA extracts main features from images, which the ANN then uses to classify nuts and bolts. The architecture allows real time image capture and reliable data transfer between different modules. Analysis shows that the first principal eigenvalue for nut images is around 9.5, while bolt images are about 8.9, showing clear feature separation. The ANN achieved a recognition rate of nearly 98 % for nuts and 97 % for bolts, with error dropping from 0.42 to 0.01 after 10 training cycles. The system offers fast processing, with classification finishing within 18 ms, and is suitable for industrial optical networks, smart manufacturing, and quality control.