| Summary: | The rapid development of electric vehicles also demands the development of supporting tools for EVs, namely batteries. For the EV to work optimally as support, it requires monitoring and maintaining the battery using the Battery Management System (BMS). One of the battery conditions that needs to be observed is the State of Charge (SOC). Observations on SOC can be done in various ways, one of which is by determining the estimated SOC value using data-driven. This study used public data's voltage, current, and temperature data for training and data testing using deep learning. The architecture used in this study is a combination of Long Short-Term Memory (LSTM), Gate Recurrent Units (GRU), and recurring neural network (RNN). Observation of the results was carried out on data with a temperature of 25° C and the hidden nodes of each architecture were 25 (LSTM), 50 (GRU), and 50 (RNN) obtained an MAE value of 1.8% and an MSE of 0.0561%.
|
|---|