Unsupervised data clustering for production mode identification through energy consumption monitoring and analysis

• Main Author: Toh, Zi Kai
• Other Authors: Ng Wee Keong
• Format: Final Year Project
• Language: English
• Published: 2016
• Subjects: DRNTU::Engineering::Computer science and engineering::Mathematics of computing::Numerical analysis
• Online Access: http://hdl.handle.net/10356/66408
• Institution: Nanyang Technological University

High competitive pressure in the manufacturing industry has contributed in ensuring manufacturing processes need to be closely monitored for any deviation in the process. Proper analysis of control charts that are used to determine the mode of the process not only requires a thorough knowledge and understanding of the underlying theories but also the expertise for decision making. In this paper, a methodology is adapted from the Fayyad model in searching for the most appropriate algorithm that could identify and interpret the production operational modes based on various patterns of variation energy measurement that can occur in a manufacturing process. This methodology uses both internal cluster validity measures and external cluster validity measure in evaluating the most appropriate clustering algorithm. To justify the proposed model, experiment is conducted on an industrial application, an injection moulding system. Experimental result show that the hierarchical agglomerative (complete-link) clustering is more effective in labeling the production operational modes using the energy patterns.