Smart web-based equipment management system

• Main Author: Zhou, Yin
• Other Authors: Wang Dan Wei
• Format: Final Year Project
• Language: English
• Published: 2018
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Computer hardware, software and systems
• Online Access: http://hdl.handle.net/10356/74774
• Institution: Nanyang Technological University

Machine deterioration has become a significant problem in manufacturing industries. It causes the reduction of production capacity, resulting in a delay of customer orders. For example, an overused etching tool for producing the semiconductor wafer may cause the machine to stop working that slows down the whole manufacture process. Cleaning maintenance can be conducted if it is relied on the equipment condition instead of scheduled maintenance. To meet this growing demand of Preventive Maintenance, various machine learning algorithms could be applied to predict the equipment condition, including Markov Chain, HMM, KRLS and ACO. The main objective of this final year project is to develop a smart web-based application that designed to predict the equipment condition based on HMM algorithm. To achieve this objective, the application had separately developed two HMM GUIs based on different kinds of the learning algorithm, HMM supervised learning (ML) and HMM unsupervised learning (Baum Welch). It worked as a toolbox that allowed the user to input the historical data and in results the user would be able to predict the hidden state path and the probability of the next-future state being on state N. The application was developed in Visual Studio with ASP.NET MVC framework. In addition, login session and file management through database had been designed with the implementation of Visual Studio with ASP.NET Web Form framework and Microsoft SQL Server Database. The prediction result shows that HMM supervised learning could be used when the user had the historical labeled data while HMM unsupervised learning was more support the historical data without knowing the hidden state. The accuracy of predicting results could be analyzed using the confusion matrix and different types of data could use the different algorithm in this application. The user could see the trend of the hidden state path and do the equipment maintenance according to the prediction.