Self-adaptive indirect health indicators extraction within prognosis of satellite lithium-ion battery

摘要

Lithium-ion battery has already become the third generation of space energy storage with higher specific energy, lower ratio of self-discharge and other remarkable advantages. Robust and accurate battery prognosis is indispensable not just in satellite autonomous mission planning but also in self-determined maintenance. Prognosis based on indirect health indicators (HIs) provide an effective non-invasive approach to identify the battery degradation state. But identification and extraction of effective HIs in different working conditions is still a challenge. In this paper, a self-adaptive non-invasive HIs extraction approach is explored, which benefits enhancing the robustness and adaptability of the indirect health indicators under different working situation. A cluster of potential HIs are extracted from the data samples of the battery. Principle component analysis (PCA) is taken into consideration to select and fuse those different HIs. Grey Relational Analysis (GRA) is utilized to evaluate the effectiveness of the hybrid HI. The verifications show that the proposed self-adaptive indirect HIs extraction approach can achieve promising results with high relativity with direct HIs, even operating with diverse conditions.