Aesha, Umme Habiba and Bhuiyan, Reyan Hridoy and Khan, Sumaya Mahajabin and Ayan, Yeasir Arafat (2026) INDUSTRIAL MACHINE FAILURE PATTERN ANALYSIS FOR AUTOMATED DEFECT DIAGNOSIS IN ADVANCED MANUFACTURING SYSTEMS. JOURNAL OF ENGINEERING, MECHANICS AND MODERN ARCHITECTURE, 5 (10). ISSN 2181-4384

Text Jemma_INDUSTRIAL+MACHINE+FAILURE.pdf Download (497kB)

Abstract

This study investigates the application of exploratory data analysis as a tool to facilitate automated defect diagnosis and reliability engineering of advanced industrial automation systems. The research is based on a quantitative exploratory design with the help of the Predictive Maintenance Dataset with Irregularities to explore the machine operating conditions and failure patterns. The focus of the analysis is on the distribution of machine failures, variations in operating parameters, and relationships between the operating parameters of the machines. These results show that the dataset is dominated by normal operating conditions, and the failure events make up a relatively small fraction of the dataset. The most common fault categories are heat dissipation failure and overstrain failure. It is discovered through comparative analysis of the operating conditions that various types of failure have different characteristics in operation, specifically in terms of torque, rotational speed, tool wear, temperature, etc. The air temperature and process temperature are highly positively correlated, and the rotational speed and torque are highly negatively correlated. Failure status is correlated with air temperature and torque, with weak to moderate positive correlations, which could be used as diagnostic criteria. Air temperature and torque are correlated with failure status with weak to moderate positive correlation, indicating that they could be used as diagnostic variables. The results suggest the importance of exploratory analysis to detect patterns of operation and factors that repeat as failure in an industrial setting. The study is an important contribution to reliability engineering by providing evidence-based information that could help maintenance personnel in detecting problems early and making informed decisions, as well as help design future automated defect diagnosis frameworks.

Item Type:	Article
Uncontrolled Keywords:	Defect, Diagnosis, Reliability, Engineering, Industrial, Automation, Predictive, Dataset
Subjects:	H Social Sciences
Depositing User:	admin eprints
Date Deposited:	16 Oct 2026 14:45
Last Modified:	16 Oct 2026 14:45
URI:	http://eprints.umsida.ac.id/id/eprint/17207

Actions (login required)

View Item