Abstract
Electric cars are spreading fast, so smart systems to track battery charge have become essential. Getting the charge level right helps prevent damage from too much charging or draining, while also making batteries last longer. Still, lithium-ion cells behave unpredictably - changing with temperature, age, and usage - which complicates accurate readings. Because of this, figuring out exact levels demands more than basic methods. Six recent approaches to estimate charge are examined here, each shaped by current studies. What emerges is not a single fix but varied paths through shifting chemical responses. A look at different ways to manage battery charging starts with physical devices using fuzzy logic, built into circuits that run on STM32 chips and respond instantly to sensor readings. Moving away from hardware, some systems rely on patterns found in data, like the Tsetlin Machine, which shows how it reaches conclusions while still predicting well - key when mistakes could be dangerous. Comparisons show how basic techniques stack up: counting charge flow, measuring voltage when idle, and refining estimates with Kalman math, each tested inside simplified electrical designs. Temperature shifts and wear over time affect voltage behaviour, making fixed assumptions less reliable as batteries age. Instead of preset rules, certain algorithms learn from examples; SVR and tree models test how stable predictions stay when conditions change. Wrapping up, the study dives into a combined Dual Extended Kalman Filter setup that tracks both system states and internal values at once. Instead of treating methods separately, it blends insights from physical tests, simulation models, and pattern- based approaches - giving a clear path through different SoC estimation options depending on how electric vehicles are used.