다양한 외기 온도를 고려한 리튬 이온 배터리의 트랜스포머 기반 충전 상태 추정 연구

Abstract

In this paper, we propose a transformer-baseddeep learning network that accurately estimates thestate of charge (SOC) of a lithium-ion battery for arange of ambient temperature conditions.Theinternal chemical characteristics of lithium-ionbatteries change as the temperature changes.Therefore, existing studies are limited to accuratelyestimating SOC only for trained temperatureconditions.To overcome this limitation, we proposea neural network that accurately estimates SOC atvarious temperatures, even for untrainedtemperatures.The experimental validation of theproposed method at various temperatures shows thatthe maximum error is 2.5% and theroot-mean-square error is 0.9195%, indicating thatthe SOC is well estimated.We also performvalidation at sub-zero temperatures, where thecharacteristics of the battery change significantly,proving that the proposed method is practical for awide temperature range.