Technical gazette, Vol. 31 No. 5, 2024.
Original scientific paper
https://doi.org/10.17559/TV-20230708000789
An Electric Vehicle Charging Control System using LSTM Encoding-GRU Decoding
Taofang Xia
; State Grid Fu Jian Marketing Service Center, China
Shian Zhan
; State Grid Fu Jian Marketing Service Center, China
*
Jianmin Xu
; Huali Technology Co., China
Xu Ren
; Huali Technology Co., China
* Corresponding author.
Abstract
Electric vehicle (EV) charging is random in time and space, and a large number of electric vehicles connected in a short period of time will cause the phenomenon of "peak on peak" in the community power grid load, which has certain safety risks. The microset metering system can effectively control the charging and discharging of electric vehicles, greatly reducing the transformer load and ensuring safety. However, existing technology only regulates electricity consumption based on the current moment, and this method does not provide an estimate of what may happen in the future. We have made improvements to address this issue. The total electricity consumption of a common community consists of unchangeable public infrastructure electricity and changeable residential electricity. We divide residential electricity consumption into two categories: residential domestic electricity consumption and charging pile electricity consumption. Long Short Term Memory (LSTM) encoding- Gate Recurrent Unit (GRU) decoding is used to predict short-term residential electricity consumption, and then the transformer full load value minus the difference between the unchanged public infrastructure electricity consumption and the residential electricity consumption is used as the upper threshold for charging at the charging piles, which is finally utilized to stagger the EV charging. After constructing the complete system, we conducted relevant simulation experiments to verify our ideas, and the simulation results show that our method has some effectiveness.
Keywords
LSTM-GRU; staggered charging of electric vehicles; time series data
Hrčak ID:
320383
URI
Publication date:
31.8.2024.
Visits: 164 *