Affordable Access

Access to the full text

Big Data Analysis Methodology for Smart Manufacturing Systems

Authors
  • Sim, Hyun Sik1
  • 1 Kyonggi University, Department of Industrial and Management Engineering, Suwon-city, Gyeonggi, 154-42, Republic of Korea , Suwon-city (South Korea)
Type
Published Article
Journal
International Journal of Precision Engineering and Manufacturing
Publisher
Korean Society for Precision Engineering
Publication Date
May 07, 2019
Volume
20
Issue
6
Pages
973–982
Identifiers
DOI: 10.1007/s12541-019-00136-7
Source
Springer Nature
Keywords
License
Yellow

Abstract

The goal of smart factories is to improve productivity and reduce production costs, but it is more important to attain manufacturing competitiveness through improvements to product quality and yield. As product functions become more advanced and processing becomes increasingly miniaturized, the yields of micro-manufacturing processes have become an important management factor, determining the production cost and quality of a product. Micro-manufacturing processes generally pass through many stages to produce a product; therefore, it is difficult to find the process or piece of equipment where a fault has occurred. As such, it is difficult to realistically ensure high yields. This paper presents an S-EES (smart-equipment engineering system) construction and big data analysis methodology for manufacturing to increase product yield and quality in a smart factory environment. It also presents plans for acquiring the data needed for big data analysis of a manufacturing site and for constructing the system. To improve product yield, it is necessary to analyze the fault factors causing low yield; similarly, the critical processes and equipment that affect these fault factors must be identified and managed. However, interrelations exist between pieces of equipment, and complex faults are caused by the downstream as well as upstream in the processing sequence that a certain lot passes through. Because of this, yield management is important but also difficult. This study finds the fault-responsible processes and machines that affect yields by using a method that utilizes PLS-VIP (partial least squares with variable importance of projection) and association rules in micro-manufacturing line processes, and it classifies these processes and machines as single factors or cumulative factors. In addition, it applies the specific methodology to an actual site, extracts the fault-responsible processes and machines, and confirms the effects of important processes and equipment on yields.

Report this publication

Statistics

Seen <100 times