Data-based manufacturing process optimization using image data : case : veneer and LVL

Abstract

Manufacturing industries are continually exploring ways to optimize their manufacturing processes to improve production by maximizing raw materials usage and minimizing operational costs. Industry 4.0 brings a digital revolution to the manufacturing industries in which Machine learning, IoT, Big Data, Data Mining, and Image processing play a vital role in optimizing manufacturing processes. This thesis aims to examine how the manufacturing process in the veneer/LVL industry can be optimized by using data from veneer sheets and applying machine learning methods. To achieve the thesis objective, a comprehensive literature review has been conducted of the existing methodologies used in the veneer/LVL industry.

In the empirical part of this thesis textural image data of veneer sheets is extracted and utilized to identify similar veneer sheets images after the drying process. Veneer sheets contain different variations in the textures; this thesis studied approaches to select the sheets containing high textures as well as low textures. A model has been developed in this thesis to select the candidate dry sheets with high and low textures after the drying process. Three approaches have been studied with Gray-Level Co-Occurrence Matrix (GLCM) and Canny edge detection methods on the image data set, and 20 features have been extracted from the image's texture. Future work and research required for further analyse the veneer sheet images with different edge detection methods are discussed.