Towards a probabilistic error correction approach for improved drone battery health assessment

Abstract

Health monitoring of remote critical infrastructure, such as offshore wind turbines, is complex and expensive. For the offshore energy sector, the accessibility for on-site asset inspection is hampered due to their harsh and remote location. In this context, inspection drones are crucial assets. They can perform multiple tasks, which are benefitial for the industry and society, including the improved reliability of critical and remote infrastructure. However, the reliability and safety assurance of inspection drones is complex, as they are autonomous systems and they require incorporating run-time operation and degradation knowledge. Focusing on the health assessment of inspection drones, their battery is a key component, which is a single point of failure and determines the probability of a successful operation. In this context, this paper presents a novel concept for inspection drone battery health assessment through a probabilistic hybrid approach which combines physics-based battery discharge models with data-driven error forecasting strategies. Results are validated with real data obtained through different offshore wind inspection flights of drones.