RESEARCH ON THE STATE OF CHARGE ESTIMATION OF LITHIUM-ION BATTERY USING TRANSFORMER REGRESSOR FOR INTELLIGENT BATTERY MANAGEMENT SYSTEMS

Povzetek

For the purpose of assuring the safety, dependability, and prolonged lifespan of energy storage systems in electric vehicles, it is vital to have an accurate estimation of the level of charge of the lithium-ion battery. Attempts to capture the complicated and non-linear behaviors of batteries are frequently unsuccessful when using traditional methods. From the perspective of Battery Management System (BMS) applications, this study gives a comparative analysis of a number of different artificial intelligence models, with a particular emphasis on how well these models perform in terms of state of charge estimate. Specifically, in order to estimate the State of Charge (SoC), models such as Random Forest, Support Vector Machine, Convolutional Neural Network, and Transformer are developed and assessed with the use of data pertaining to voltage, current, and temperature. The research endeavours an exhaustive performance comparison with the objective of determining which AI-based solution for BMS applications represents the highest level of certainty and robustness.