Comparative Study of Machine Learning for Managing EV Energy Storage with Battery-Hydrogen Tank

This study utilized machine learning methods to manage the battery storage system and hydrogen tank, while taking into account the motor speed, average speed (57.6 km/h), motor consumption, and State of Charge (SOC) of the vehicle battery. The study was conducted over six hours of vehicle motor operation and compared the performance of four machine learning classification methods (K-nearest neighbors (k-NN), AdaBoost, Gaussian Naive Bayes, and Random Forest) to determine which power source could best supply energy to the vehicle’s Electric Motor (EM). The results show that the AdaBoost method outperformed the other methods with an f1-score accuracy of 0.98 for the fuel cell (FC) state and 0.90 for the battery state in decision-making regarding the operation of the FC and the battery.