Title: A Five-Layer Digital Intelligence Framework for Lifecycle-Aware Battery Management in Sustainable Energy Systems Abstract: This paper introduces a five-tier digital twin architecture designed to advance battery management within sustainable energy and smart manufacturing contexts. The framework integrates geometric digital modeling, real-time telemetry, physics-informed machine learning (PIML), prescriptive optimization, and autonomous control to enable full lifecycle oversight of lithium-ion batteries. Utilizing NVIDIA Omniverse for 3D visualization and PhysicsNeMo for accelerated multi-physics simulation, the system achieves high-fidelity electrochemical-thermal calibration via Bayesian optimization, reducing voltage and temperature prediction errors below 1.57% and 0.39% MAPE, respectively. A physics-informed neural network delivers state-of-health forecasts with under 3% error while quantifying uncertainty through deep autoencoding Gaussian mixture models. By unifying virtual modeling with operational intelligence, this framework supports predictive health monitoring, adaptive charging, thermal management, and end-of-life repurposing—key functions for next-generation battery manufacturing and sustainable energy storage deployment. The approach moves beyond conventional reactive BMS designs toward autonomous, interpretable, and lifecycle-optimized battery systems.