National Project

PRIN 2022 E3DM is a Project of Significant National Interest funded by the FIRST (Fund for Investments in Scientific and Technologic Research). It studies innovative decision-making frameworks for accelerated degradation tests (ADT) and predictive maintenance (PM), when information about the state of the system, represented by prior knowledge and experimental data, is encapsulated in a degradation model. The developed methods are aimed at the testing of electronic components and of lithium-ion batteries, and at the maintenance plans of lithium-ion batteries, also using new experimental datasets

The main objective of E3DM is to propose innovative decision-making frameworks for accelerated degradation tests (ADT) and predictive maintenance (PM), when information about the state of the system, represented by prior knowledge and experimental data, is encapsulated in a degradation model. The developed methods will be applied to the testing of electronic components and of lithium-ion batteries and to the maintenance plans of lithium-ion batteries, also using new experimental datasets.

Among degradation models, general path models (GPM) will be considered. Previous literature will be used to identify the most suitable model families and different estimation methods will be examined so as to include information from the data in the decision-making procedures. ADT test plans will be obtained by considering novel approaches based on optimal design theory. The optimal test plans will be developed by considering real-life applications, also allowing for the presence of several stress variables and/or levels, time constraints and costs.

Maintenance policies vary widely, depending on system type, available data, degradation model used and policy selection method, but a unifying framework is lacking. It has recently been suggested in the literature that RL offers such a framework. A sequential PM planning method, using the RUL as a key indicator, will be built by bringing together degradation modelling and RL, also including the degradation model update as new information on the state of health of the battery become available.

Every single method composing this approach is well represented in the literature but considering them all together is a novel contribution from E3DM. In particular, it seems promising because batteries are often used in uncontrolled environments, hence the importance of the joint policy/design and model learning aspects.

The results of E3DM will be disseminated via a dedicated web site, a final conference, and open access articles.