Embedded sensing: The neural frontier and early-warning revolution in battery safety monitoring

The rapid proliferation of battery systems has positioned thermal runaway prevention as a crucial technological imperative. In-situ sensor-based monitoring frameworks enable real-time tracking of internal parameters, thereby providing early warnings and interventions for thermal management. However, conventional sensors, limited by their unidimensional architectures, struggle to accurately capture the intricate interplay among thermal, mechanical, and chemical fields. This limitation results in significant blind spots when predicting battery degradation under multiphysics conditions over the entire lifecycle. Consequently, advancing multi-parameter sensing technologies and developing multidimensional sensing architectures become essential for achieving comprehensive battery safety monitoring. From an embedded sensing perspective, this review systematically examines critical challenges related to chemical compatibility, measurement accuracy, and multi-parameter monitoring encountered during sensor integration. It provides a detailed elaboration on the operating principles and practical applications of thermocouples, optical fiber sensors, and thin-film sensors in batteries. To address technological bottlenecks, such as risks to structural integrity, electrolyte-induced performance degradation, and limitations in single-parameter monitoring, we propose strategies that include sensor miniaturization, the selection of chemically robust materials, integrated multidimensional in-situ platforms, and the incorporation of artificial intelligence (AI) technologies. This review advances comprehensive understanding of battery multidimensional sensing systems, significantly enhancing active safety engineering and multiphysics diagnostic frameworks.