1. Charging Current Estimation of Electric Vehicle DC Fast Chargers Using Long Short-Term Memory Networks
- Author
-
Wenwen Zhou, Peilei Zhao, and Xiaopeng Zhu
- Subjects
business.product_category ,Mean squared error ,Computer science ,business.industry ,020209 energy ,Deep learning ,Process (computing) ,020302 automobile design & engineering ,Topology (electrical circuits) ,02 engineering and technology ,Power (physics) ,0203 mechanical engineering ,Electric vehicle ,0202 electrical engineering, electronic engineering, information engineering ,Electronic engineering ,Waveform ,Artificial intelligence ,Current (fluid) ,business - Abstract
This paper presents a long short-term memory network for estimating the charging current in electric vehicle DC fast chargers. From the prospective of testing, current estimation with deep learning techniques can provide an approach for predicting health state and remaining useful life of EV chargers in the future. A typical charging data is comprised of three control signals, two power signals, and BMS messages. Data set collection has been finished by a testing tool from five electric vehicle models and six charger models. The LSTM theory is introduced briefly, then the topology of the designed network for current estimation and the training process with root mean square error and training loss are given. The experimental results show the charging current has been estimated by the designed network with a best RMSE of 8.59 A. We also discuss the error of an estimated current waveform in comparison to a tested current waveform.
- Published
- 2019
- Full Text
- View/download PDF