"A real-world battery dataset comprising operational data from battery packs of 300 mass-produced new energy vehicles."

"RVD is a large-scale, real-world electric vehicle battery dataset comprising 104,621 charge\u2013discharge cycles collected from 300 vehicles, with a cumulative driving distance exceeding 30 million km. For each cycle, the dataset records key battery and vehicle variables, including voltage, current, state of charge (SOC), state of health (SOH), vehicle speed, acceleration, and accumulated mileage, providing longitudinal observations of battery degradation throughout actual vehicle use. The large vehicle population also captures substantial diversity among vehicle users, reflecting differences in travel frequency, vehicle-use intensity, charging habits, and long-term usage patterns. This user-level heterogeneity results in diverse battery ageing trajectories across vehicles, making RVD particularly valuable for SOH estimation, time-series forecasting, transfer learning, robustness analysis, and evaluating model generalization to previously unseen vehicles and users."