Prediction of the health and charge status of batteries used in electric vehicles through machine learning

Globally, the interest in electric vehicles (EVs) has increased due to the reduction of fossil fuels and the environmental, economic, and grid-related benefits they offer. Numerous research studies are currently being conducted on EVs. The aim of this thesis is to contribute to the Turkish literature on the development and technological advancements of EVs in our country. Therefore, this study provides information about the historical progression of electric vehicles, battery systems used in EVs, types of electric motors, and converter topologies. Additionally, a detailed examination is conducted on the status of EV charging stations in our country, charging technologies, charging methods, and charging modes. Furthermore, a focused investigation is carried out on the grid integration of EVs, specifically exploring the Vehicle-to-Grid (V2G) technology within the context of charging technologies. In addition, the thesis extensively analyzes the crucial issues of EVs, such as battery health (SOH) and charging (SOC) state, along with machine learning predictive algorithms. Real-world data is collected from autonomous mobile robots functioning as electric vehicles, without employing any simulated datasets. These collected data are processed, subjected to machine learning algorithms, and tests are conducted, followed by a comparative analysis of these tests. The objective is to contribute to the development of future battery health and charging prediction methods through these analyses. Furthermore, raising awareness for ongoing indigenous electric vehicle projects and their development in our country is also one of the objectives of this study.