Online learning for equipment monitoring

• Grant number: 25/01895-9
• Support Opportunities: Scholarships in Brazil - Master
• Start date: August 01, 2025
• End date: July 31, 2027
• Field of knowledge: Physical Sciences and Mathematics - Computer Science
• Principal Investigator: Lilian Berton
• Grantee: Juan Carlos La Rosa Paredes
• 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
• Company: Secretaria de Desenvolvimento Econômico (São Paulo - Estado). Instituto de Pesquisas Tecnológicas S/A (IPT)
• Associated research grant: 20/09850-0 - Applied Artificial Intelligence Research Center: accelerating the evolution of industries toward standard 5.0, AP.PCPE

Abstract

From a data science perspective, Industry 4.0 represents a paradigm shift in the Fourth Industrial Revolution, enabling the extraction of valuable knowledge through the adoption of intelligent monitoring and data fusion strategies. Additionally, it allows for the application of advanced machine learning and optimization methods. One of the main objectives of data science in this context is to effectively predict anomalous behaviors in machines, tools, and processes, anticipating critical events and potential failures that could lead to economic losses and compromise operational safety. Maintenance is one of the key application areas of Industry 4.0, as machines must be constantly available, while repair time and costs must be minimized. Balancing these factors has become a major challenge for technicians and specialists. In a descriptive model, analysis is based on system observations, and the selected data depend on a specific objective. However, if the objective or data changes, the model may become obsolete, requiring constant updates. Despite the growing demand for adaptable models, only a small portion of studies has explored descriptive models from the perspective of data stream processing, where data arrives continuously and dynamically, requiring flexible computational environments that comply with resource constraints. Data stream classification aims to categorize examples from data streams into multiple classes, a significant challenge as new classes may emerge and existing ones may change over time. Therefore, this work aims to develop predictive prognostic models using data stream processing techniques applied to online learning, promoting more effective and adaptable solutions for intelligent monitoring and maintenance.