Data Driven Prognostics and Diagnostics for SMT Machine Logs

Ng, Jung Hong (2020) Data Driven Prognostics and Diagnostics for SMT Machine Logs. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text NgJungHong Full Text.pdf Restricted to Registered users only Download (1MB)

Abstract

In recent years, Surface-mount Technology (SMT) has become the industry norm for electronic device assemblies and replaced the Through-hole technology (THT). The SMT is a very mature technology for mass production of electronic devices, but the maintenance engineers still suffer for part drop issue that causes by mechanical limitation of the machine. It is a time-consuming effort to troubleshoot the root cause of the part drop event. Consequently, the productivity decreases and the company had to bear more cost for the operation. In the past, maintenance is carried out periodically, in order to prevent the machine to break down. However, a new trend in maintenance technology which is the Prognostics and Health Maintenance (PHM). PHM is an engineering discipline that provides a better solution for maintenance. PHM contains two important components which are the diagnostics and prognostics. Prognostics is to forecast the health status of the machine, meanwhile, Diagnostics is the process to detect the fault of the machine. In this project, the data-driven Prognostics and Diagnostics is developed for SMT machine logs. The part drop features are extracted out based on the machine logs provided by Hotayi Electronics (m) Sdn. Bhd. The Prognostics and Diagnostics model is evaluated using real industrial data. The threshold of the WARNING and ERROR state is determined. When exceeding the threshold, the maintenance is suggested to be carried out. At the same, the Diagnostics model can diagnose the malfunction of the nozzle and feeder. From the result, it showed that the Prognostics and Diagnostics model can work for the Machine logs.