PERIOD AND LOCATION
The course will be held from 16 to 18 november 2026, from 9.30 to 17.30 (CET).
It will be held in hybrid setting (both physical and virtual attendance is allowed) at the Department of Energy – Politecnico di Milano, Campus Bovisa La Masa, 20156, Milano.
COURSE PARTICIPANTS
The course is mainly dedicated to control, process, quality and maintenance engineers, asset managers, data scientists, data miners, researchers and PhD students in the areas of Reliability, Availability, Maintainability (RAM), and fault diagnostics and Prognostics and Health Management (PHM).
MISSION AND GOAL
In recent years, the volume of data and information collected by the industry has been growing exponentially, and more sophisticated and performing analytics have been developed to exploit their content.
This offers great opportunities for optimized, safe and reliable productions and products, including optimal predictive maintenance for “zero-defect” production with reduced warehouse costs, and improved system availability, with “zero unexpected shutdowns”.
To grasp these opportunities, new system analysis capabilities and data analytics skills are needed. The goal of this course is to provide participants with advanced methodological competences, analytical skills and computational tools necessary to effectively operate in the areas of reliability, availability, maintainability, diagnostics and prognostics of modern industrial equipment and systems. The course presents advanced techniques and analytics to improve safety, increase efficiency, manage equipment aging and obsolescence by setting up condition-based, predictive and prescriptive maintenance and asset management strategies.
TRAINING FORMAT
The course is focused on advanced methods for the availability, reliability and maintainability (RAM) analysis of complex systems, and Prognostics and Health Management (PHM) for condition-based and predictive maintenance. Data analytics methods, including Artificial Neural Networks, Deep Learning, Convolutional Neural Networks, Autoencoders, Physics-Informed Machine Learning, Domain Adaptation, Large Language Models are illustrated, and hands-on sessions are carried out in which the participants directly apply to practical case studies the methods explained in the lectures (MATLAB and/or PHYTON are used). Also, real applications of the advanced methods illustrated in the course are presented.
Lectures are held in English.
All participants will receive a complete set of the presentation slides with specific examples and case studies, selected reference lists and resources in electronic format.
CONTENTS
Methods:
Machine learning techniques for RAM and PHM (Artificial Neural Networks, Deep Learning, Convolutional Neural Networks, Autoencoders, Physics-Informed Machine Learning, Domain Adaptation, Large Language Models);
Bayesian filtering for prognostics (Particle Filtering).
Exercise sessions:
Autoencoders for fault detection; Artificial Neural Networks and Convolutional Neural Networks for fault diagnostics; Large Language Models for maintenace planning; Particle Filter for failure time prediction;
Applications:
Monte Carlo Simulation for system reliability/availability analysis and condition-based maintenance management;
Regression and classification techniques for fault detection, classification and prognosis in industrial equipment.
CERTIFICATE OF ATTENDANCE
At the end of the course, the participants will receive a certificate of attendance, provided that they have attended at least 80% of the course lectures.