A cloud-edge collaborative strategy for capacity prognostic of lithium-ion batteries based on dynamic weight allocation and machine learning. (15th January 2022)

• Record Type: Journal Article
• Authors: Sun, Tao; Wang, Shaoqing; Jiang, Sheng; Xu, Bowen; Han, Xuebing; Lai, Xin; Zheng, Yuejiu
• Is Part Of: Energy. Volume 239:Part C(2022)
• Publication Date: 2022-01-15
• Subjects: Capacity prediction -- Induced ordered weighted averaging operator -- Neural network -- Cloud-edge collaboration; Power resources -- Periodicals; Power (Mechanics) -- Periodicals; Energy consumption -- Periodicals; 333.7905
• DOI: 10.1016/j.energy.2021.122185
• Languages: English
• ISSNs: 0360-5442

Abstract

Abstract: To avert the degradation information island brought by a single model for state estimation of lithium-ion batteries (LIBs), multi-model integration is well worth being extended for health evaluation. A cloud-edge collaboration strategy that integrates multi-model adaptation and machine learning is proposed for battery capacity prediction in lifespan. In this respect, multiple individual algorithms are promoted relying on online data from battery management system (BMS) and induced ordered weighted average (IOWA) operator is introduced for joint capacity estimation. Taking the massive data storage and computing power into account, the operation of the neural network optimized by genetic algorithm (GA) is elevated to the big data platform for strong robustness and accurate prediction, accompanied by the historical state message from BMS. With the validation of NASA battery aging data, the error of the collaborative prediction strategy is kept within 5%, and the feasibility of cloud-edge collaboration is confirmed for future battery management. Highlights: A novel "Induced ordered weighted average operator" is applied to capacity estimation. Sequential cloud-edge collaboration strategy that integrates multi-model adaptation and machine learning is proposed. Simulations relying on experiments confirm ideal accuracy and convergence rate. Excellent anti-interference ability against system uncertainties is achieved.