Abstract
The manufacturing industry stands at a crossroads, facing the dual challenge of meeting growing global demand while addressing environmental concerns. Sustainable practices have emerged as a paramount focus, and the integration of deep learning techniques offers a promising avenue for achieving sustainability goals during the manufacturing of parts A deep learning approach for online fault recognition via automatic image processing is developed to identify defects and thereby prevent non-conformities in the Computer Numerically Controlled (CNC) manufacturing process. Analytical research was conducted wherein in-process images of tool wear acquired during the CNC manufacturing process are analyzed via a bi-stream Deep Convolutional Neural Network-based model. Experimental evaluations confirmed the effectiveness of the deep learning methods for the detection and ResNet was identified as the best Deep Learning (DL) algorithm to predict the quality of the part produced with a batch size of 8 epoch 50 learning rate .0001 together with RMS prop optimizer, to hyper-tune the model. This deep learning framework, together with machine learning models like X.G.Boost incorporating real-time data acquisition of input parameters, was able to predict the final quality of the parts manufactured with an accuracy of 96.58% fostering sustainable practices within the manufacturing environment directly impacting 14 KPI’s and indirectly 7KPI’s of the sustainability index.