Embedded Sensing-enabled Distributed Thermal Modeling and Nondestructive Thermal Monitoring of Lithium-ion Battery

Accurate modeling and estimation of the internal temperature distribution is of great significance to the thermal management of lithium-ion batteries (LIBs). Existing control-oriented models generally assume a uniform temperature distribution along the axial direction of LIB. The ignorance of thermal inhomogeneity however challenges the refined thermal monitoring of LIB. To remedy this deficiency, this paper proposes for the first time a novel distributed thermal model for LIB, by hybridizing the thermal transfer law and the artificial intelligence approach. Relying on the spatial temperatures of LIB obtained by a distributed sensing technique, a lumped-parameter thermal network model is developed to capture the general thermal behavior of LIB. In a cascaded manner, the long short-term memory (LSTM) neural network is proposed to compensate for the thermal inhomogeneities that cannot be explained. The proposed cascaded distributed thermal (CDT) model further proves to be compatible with commonly-used observers for online internal temperature distribution estimation. Experimental results suggest that the proposed distributed model and the associated estimation framework can give space-resolved inner temperature estimation with remarkably-improved accuracy compared with the existing methods.