This study proposes a high-precision framework for predict
ing Unmanned Aerial Vehicles (UAVs) batteries’ remaining useful life
(RUL). Utilizing a multilevel data processing approach and a decay
regularization-enhanced stochastic configuration network (SCN), this frame
work aims to improve battery health monitoring (BHM) safety and effi-
ciency. The framework is validated using NASA’s UAV battery dataset
and compared against existing techniques, demonstrating notable perfor
mance improvements. Specifically, the study starts with an initial analy
sis of the battery’s key health indicator (HI) using variational modal
decomposition (VMD), followed by a secondary decomposition using
complete ensemble empirical modal decomposition (CEEMDAN) and
signal-to-noise ratio (SNR). Finally, the signals are further processed by
Fast Fourier Transform (FFT) and Power Spectral Density (PSD) anal
ysis and refined by band-pass filters. Performance evaluation across four
datasets (LLF, ULA, LRF, URA) shows that the proposed model outper
forms comparative models in terms of Root Mean Square Error (RMSE)
and Coefficient of Determination (R 2 ) metrics, particularly on the LRF
dataset, achieving an RMSE of 0.054 and an R 2 of 0.8137, indicating
very high prediction accuracy and model fit.