An improved temporal data pipeline with foundational model for battery State of Health (SOH) prediction (R²->0.99) using advanced time series decomposition (D3R, CEEMDAN) and transformer-based methods. Utilized 100-150 features (ARIMA-based, Rolling statistics, Degradation indicators)
rolling-statistics foundational-models gru-neural-network ceemdan-lstm-arima emd-gru-arima-hybrid d3r-decomposition arima-based-features grid-scale-energy-storage-monitoring predictive-maintenance-scheduling battery-lifecycle-optimization
-
Updated
Jan 25, 2026 - Python