Print Email Facebook Twitter Improving the Quality Control Process for the Roll-to-Roll Production of ThinFilm Modules at HyET Solar Title Improving the Quality Control Process for the Roll-to-Roll Production of ThinFilm Modules at HyET Solar Author Rajasekhar, Rajesh (TU Delft Electrical Engineering, Mathematics and Computer Science) Contributor Smets, A.H.M. (mentor) Degree granting institution Delft University of Technology Date 2021-08-31 Abstract The quality of a product is an important factor in the manufacturing industry. Maintaining a standard product quality ensures customer satisfaction and loyalty to the brand and reduces the risk and cost involved in the production. A system that reviews the product quality in production is the quality control system. Quality control ensures that the product that reaches the customer is defect-free and safe. At HyET Solar B.V, a thin-film PV manufacturing company located in Arnhem, The Netherlands, a potential to improve/upgrade the quality control process was identified as the production facility aims to increase its production capacity. This thesis aimed to provide such a system for the industry. The current limitations of the quality control process and the defects that affect the module quality in the production were studied. Thorough research was carried out on the industrial practices for quality control in two main industries, namely the photovoltaic industry and the manufacturing industry. Machine vision, a subpart of computer vision, was chosen as the technology to realise a vision system to detect defective modules in the production line. Machine vision with deep learning algorithms has proven to solve complex industrial problems. A deep learning model to perform binary image classification has been developed and tested. The deep learning model employs various layers of neural networks to classify the defective and defect-free modules by mapping the features on their own. The results obtained from the model promises that a vision system with a deep learning algorithm is effective in improving the production quality. Subject Deep Learning, Convolutional Neural Network, Re-gression, Sensitivity Analysis, Batch SizeQuality Control To reference this document use: http://resolver.tudelft.nl/uuid:478cd290-8be4-4103-aaae-176afcb3af39 Embargo date 2024-08-31 Part of collection Student theses Document type master thesis Rights © 2021 Rajesh Rajasekhar Files file embargo until 2024-08-31