Exploring Interpretability for Predictive Maintenance Models in Industry 4.0

• Grant number: 23/11418-8
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: November 01, 2023
• End date: October 31, 2024
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computer Systems
• Principal Investigator: Lilian Berton
• Grantee: João Victor Assaoka Ribeiro
• Host Institution: Instituto de Ciência e Tecnologia (ICT). Universidade Federal de São Paulo (UNIFESP). Campus São José dos Campos. São José dos Campos , SP, Brazil

Abstract

As Machine Learning (ML) gains popularity in many domains like healthcare and industrial equipment monitoring, the need for transparent decision-making models becomes paramount. In industrial settings, timely maintenance of equipment is crucial for uninterrupted operations and cost savings. Leveraging machine learning, predictive and prescriptive maintenance endeavors to predict and prevent system failures. This research aims to study interpretability to present insights derived from predictive maintenance models trained on time-series data. We will explore dimensionality reduction, statistical measures, trend components, or frequency domain characteristics, and explore simpler models like decision trees or linear regression. The ability to explain the relationships within the ML models empowers the technicians to comprehend the factors influencing maintenance predictions and subsequent decisions. Transparent interpretability not only fosters trust in the model's outcomes but also facilitates the identification of key variables and patterns that drive maintenance recommendations. This becomes particularly essential in industrial settings, where actionable insights from these models can drive efficient resource allocation and enhance overall operational reliability.