An adaptive threshold strategy based on empirical distribution functions and information entropy for battery abnormal diagnosis and fault alarm

The widespread deployment of batteries has driven the green transformation and sustainable development across various sectors, positioning battery fault diagnosis technology as a critical research priority. This paper proposes a method to determine adaptive thresholds for abnormal detection and fault alarm, based on empirical distribution functions (EDF) and information entropy (IE). The method preprocesses data from real-world vehicles to ensure data quality, calculates adaptive thresholds using EDF, and identifies abnormal states through the IE algorithm. Additionally, the paper introduces an abnormal fault coefficient to quantify the relationship between the abnormal state and the fault alarm mechanism. The effectiveness of the method is validated using real-world vehicle data, including thermal runaway, sensor failure, and normal data. The results demonstrate that the proposed algorithm effectively identifies battery abnormalities in a timely manner, achieving an 89 % reduction in false alarm rate, a 35 % reduction in missed detection rate, and a 55 % decrease in storage cost. The proposed algorithm enhances battery fault detection and reduces detection costs, offering significant value for the transportation and energy storage sectors. It enables timely diagnosis of battery abnormalities and activates the alarm mechanism, ensuring safe battery operation.