Enhancing Battery Management Systems with Machine Learning and 5G Connectivity beyond Maritime Transport Software Architectures
Constantin-Laurentiu Paraschiv, Larisa-Mihaela Tufeanu, Marius Vochin · 2024
The growing advancement in battery technology has revolutionized various domains and services, notably transportation and energy storage. This research focuses on improving battery management systems (BMS) for electric vehicles (EV) and hybrid vehicles (HEV) by leveraging advanced machine learning models and 5G connectivity. This study proposes an integrated software architecture designed for both industrial and road applications, aiming to optimize battery performance and safety by analyzing collected data and predicting the anomaly events by training Random Forest and Gradient Boosting machine learning (ML) models. Throughout the research, key battery performance parameters and their normal operating ranges, such as voltage, current, temperature, charge/discharge rates, internal Resistance and State of Charge (SoC) are analyzed, according to the official regulations and utilized in Python scripting. The Python script implemented simulates battery operations by updating sensor values and logging incidents with factors such as latency, jitter, and encryption time. The script trains and compares machine learning models to determine the most accurate one for anomaly detection. The final part of the script simulates battery performance under different connectivity modes (5G Non-Standalone (NSA) and 5G Standalone (SA)) to assess the impact on data transmission and real-time monitoring. Ultimately, obtained results demonstrate not only advanced, sustainable, mobility solutions by improving the safety and performance of Battery Management Systems in electric and hybrid vehicles, but also the importance of leveraging cutting-edge technologies (such as ML) and connectivity (5G).