Multiple Spatiotemporal Broad Learning for Real-Time Temperature Estimation of Lithium-Ion Batteries

• Published in: Industrial & engineering chemistry research Vol. 62; no. 15; pp. 6251 - 6261
• Main Authors: Xu, Kangkang, Fan, Yajun, Zhu, Chengjiu, Tian, Guangdong, Hu, Luoke
• Format: Journal Article
• Language: English
• Published: American Chemical Society 04.04.2023
• Subjects: Thermodynamics, Transport, and Fluid Mechanics
• ISSN: 0888-5885; 1520-5045
• DOI: 10.1021/acs.iecr.2c04349

The temperature estimation of lithium-ion batteries (LIBs) is of great significance to their thermal management and intelligent operation. Due to different working conditions and unknown external disturbance, batteries often need to work at a large operating range with multiple working points. However, direct global modeling and persistently exciting experiment in a large working region are very costly in practical. Complex spatiotemporal coupling and infinite-dimensional nature further make the problem more difficult. To address the above problems, a novel multiple spatiotemporal modeling method is proposed based on incremental spatiotemporal broad learning (ST-BL) and adaptive ensemble learning (EL) for the temperature prediction of batteries. First, the thermal process is identified through several local spatiotemporal domains using density peak clustering. To solve the complex spatiotemporal coupling problem, a novel ST-BL that introduces a spatial kernel function into BL is developed to model each local spatiotemporal domain. Further, the multiple ST-BL model is obtained by adaptive EL of all local models. In addition, to improve the adaptive ability of the model to the current state, incremental learning is performed on local models using newly arrived samples. Since the proposed multiple spatiotemporal modeling method involves a multi-modeling mechanism, it can achieve higher accuracy and efficiency than the traditional global single spatiotemporal modeling method, which is validated by an LIB experiment.