Operando decoding of thermal runaway precursors in lithium-ion batteries: from multimodal sensing to intelligent early warning

Lithium-ion batteries (LIBs) are extensively utilized in electric transportation and large-scale energy storage systems. Thermal runaway (TR) is considered the most critical safety hazard, with associated concerns becoming increasingly prominent. To address the critical challenges posed by the lack of unified evolution laws for TR across diverse abuse scenarios and the limitations of existing engineering-oriented early warning architectures, it is imperative to elucidate TR precursor mechanisms and establish robust risk identification models and safety protection strategies. Accordingly, this study systematically reviews three typical abuse conditions and innovatively delineates a universal, cross-condition TR precursor evolution pathway characterized by the sequence of structural damage–performance degradation–signal manifestation. Building on this foundation, this paper reviews five categories of operando monitoring technologies. With online applicability and mechanistic insight as core dimensions, a comparative analysis is conducted regarding their strengths and limitations in early warning response, spatiotemporal resolution, and engineering deployment. To address the limitations of existing single-modality monitoring technologies, the Mechanism–Monitoring–Action integrated framework is proposed. By establishing a robust mapping between structural degradation characteristics and multi-physics signals, and incorporating artificial intelligence (AI) models to resolve nonlinear relationships, this framework aims to achieve precise inference of internal battery states and enable hierarchical active safety responses.