Multisource information fusion based parameterization study of lithium-ion battery electrolyte leakage

Abstract

Electrolyte leakage can cause serious accidents and is a major hazard for battery thermal runaway due to the flammability of electrolyte. Currently, the method for diagnosing electrolyte leakage depends greatly on sensors, but its reliability remains questionable. The paper characterizes the electrolyte leakage based on multiple signals and proposes a multi-parameter fusion diagnostic framework for different conditions. Firstly, the electrolyte leakage behavior of commercial lithium-ion batteries is imitated on two aging conditions. Secondly, the evolution characteristics of the cyclic capacity of leaky and normal cells are compared. The features of leaky batteries are then extracted based on measurable electrical, thermal, and gas signals during aging period, and the diagnostic sensitivity and reliability of the characteristic parameters are analyzed. According to the analysis, the sudden capacity drop of leaky batteries is found to occur at the constant voltage charging stage, and the triggering mechanism of the sudden capacity drop is revealed. Furthermore, the parametric characterization for electrolyte leakage detection is elaborated. Finally, the electro-thermal-gas multi-parameter fusion framework for electrolyte leakage diagnosis is proposed based on the priority and combination relationships of each parameter. As the result, the proposed parametric characterization is proved to be of great significance for electrolyte leakage diagnosis.