Adaptive predictive Maintenance for the Batteries of Electric Vertical Take-off and Landing (eVTOL) Aircraft using Remaining Useful Life prognostics

In this paper we develop probabilistic Remaining Useful Life (RUL) prognostics for Lithium-ion batteries using Mixture Density Networks (MDNs). We integrate these prognostics into a reliable and cost-efficient linear program model that identifies optimal battery replacement moments while limiting the risk of the batteries becoming inoperable during operations. Over time, as more measurements become available, the RUL prognostics are periodically updated, and the battery replacement strategy is adapted. We apply our approach for electric Vertical Take-off and Landing (eVTOL) aircraft, a promising emerging technology for mobility in congested urban areas. The results show that the RUL is accurately estimated using MDNs. The results also show that prognostics benefit the planning of battery replacement, leading to 80% less yearly unscheduled battery replacements compared with maintenance planning approaches when point estimates (average values) of the RUL are predicted.

Keywords: Predictive maintenance, Probabilistic RUL prognostics, eVTOL aircraft, Mixture density networks.