Mathematical programming, meta/math-heuristic, data analysis, Big Data, machine learning
Maintenance plays a key role for increasing a company’s competitiveness. In addition, especially in complex production systems, for which the breakage of a component may lead to the stoppage of the whole system and in which all the resources involved in the maintenance activities (e.g., budget and human resources) are very limited, it may become very crucial defining optimized maintenance plans. It means deciding which components have to be maintained, by respecting the imposed constraints, e.g., on both the available budget and the available human resources, for optimizing a performance criterion, e.g., the system’s reliability.
Moreover, the use of on-board sensors, for example, makes also possible the collection of Big Data regarding the breakages occurred in the past. Advanced techniques for analysing the collected data allows deducing knowledge that may be used for predicting future possible breakages and in some way, estimating the breakage probability of the components. This research activity aims at integrating advanced Big Data analysis techniques (e.g., based on machine, deep and reinforcement learning) with mathematical optimization approaches in order to both plan and schedule the maintenance activities. To this aim, multi-objective optimization approaches are also investigated in order to optimize more than one criterion simultaneously.
DAISY – Co-working Lab in Data Analytics, Artificial Intelligence and CyberSecurity
LORA – Laboratory for Operations Research Applications
Metodi e strumenti innovativi per il REACTive Product Design and Manufacturing (REACT)”, PON FESR FSE Ricerca e Innovazione, Area di specializzazione fabbrica intelligente. November 2018, duration 30 months
