Generative Adversarial Network for Synthesizing Multivariate Time-Series Data in Electric Vehicle Driving Scenarios DOI Creative Commons
Shyr-Long Jeng

Sensors, Journal Year: 2025, Volume and Issue: 25(3), P. 749 - 749

Published: Jan. 26, 2025

This paper presents a time-series point-to-point generative adversarial network (TS-p2pGAN) for synthesizing realistic electric vehicle (EV) driving data. The model accurately generates four critical operational parameters—battery state of charge (SOC), battery voltage, mechanical acceleration, and torque—as multivariate Evaluation on 70 real-world trips from an open dataset reveals the model’s exceptional accuracy in estimating SOC values, particularly under complex stop-and-restart scenarios across diverse initial levels. delivers high accuracy, with root mean square error (RMSE), absolute (MAE), dynamic time warping (DTW) consistently below 3%, 1.5%, 2.0%, respectively. Qualitative analysis using principal component (PCA) t-distributed stochastic neighbor embedding (t-SNE) demonstrates ability to preserve both feature distributions temporal dynamics original data augmentation framework offers significant potential advancing EV technology, digital energy management lithium-ion batteries (LIBs), autonomous comfort system development.

Language: Английский

Generative Adversarial Network for Synthesizing Multivariate Time-Series Data in Electric Vehicle Driving Scenarios DOI Creative Commons
Shyr-Long Jeng

Sensors, Journal Year: 2025, Volume and Issue: 25(3), P. 749 - 749

Published: Jan. 26, 2025

This paper presents a time-series point-to-point generative adversarial network (TS-p2pGAN) for synthesizing realistic electric vehicle (EV) driving data. The model accurately generates four critical operational parameters—battery state of charge (SOC), battery voltage, mechanical acceleration, and torque—as multivariate Evaluation on 70 real-world trips from an open dataset reveals the model’s exceptional accuracy in estimating SOC values, particularly under complex stop-and-restart scenarios across diverse initial levels. delivers high accuracy, with root mean square error (RMSE), absolute (MAE), dynamic time warping (DTW) consistently below 3%, 1.5%, 2.0%, respectively. Qualitative analysis using principal component (PCA) t-distributed stochastic neighbor embedding (t-SNE) demonstrates ability to preserve both feature distributions temporal dynamics original data augmentation framework offers significant potential advancing EV technology, digital energy management lithium-ion batteries (LIBs), autonomous comfort system development.

Language: Английский

Citations

0