Heuristic Kalman optimized particle filter for remaining useful life prediction of lithium-ion battery

• Published in: Microelectronics and reliability Vol. 81; pp. 232 - 243
• Main Authors: Duong, Pham Luu Trung, Raghavan, Nagarajan
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01-02-2018
• Subjects: Heuristic Kalman algorithm; Particle filtering; Particle swarm optimization; Prognostic health management; Remaining useful life; Particle filtering; Prognostic health management; Remaining useful life; Heuristic Kalman algorithm; Particle swarm optimization
• ISSN: 0026-2714; 1872-941X
• DOI: 10.1016/j.microrel.2017.12.028

Accurate prediction of the remaining useful life of a faulty component is important to the prognosis and health management of any engineering system. In recent times, the particle filter algorithm and several variants of it have been used as an effective method for this purpose. However, particle filter suffers from sample degeneracy and impoverishment. In this study, we introduce the Heuristic Kalman algorithm, a metaheuristic optimization approach, in combination with particle filtering to tackle sample degeneracy and impoverishment. Our proposed method is compared with the particle swarm optimized particle filtering technique, another popular metaheuristic approach for improvement of particle filtering. The prediction accuracy and precision of our proposed method is validated using several Lithium ion battery data sets from NASA® Ames research center.
•Heuristic Kalman algorithm (HKA) is combined with particle filter (PF) framework.•HKA helps tackle sample impoverishment and degeneracy of standard particle filter.•Our proposed method is tested on Li-ion battery test data from NASA®.•RUL predictions were more accurate for HKA approach compared to standard PF.•HKA has fewer tuning parameters and is less sensitive than particle swarm method.