The increasing penetration of renewable energy sources and energy storage systems introduces significant uncertainty and complexity in modern smart energy systems. Traditional optimization methods often fail to provide efficient real-time solutions under high-dimensional and stochastic operating conditions. This study proposes a multi-level hybrid energy management framework that integrates deep learning, reinforcement learning, and quantum optimization techniques. The architecture combines neural-network-based forecasting, reinforcement learning-driven decision-making, and quantum-assisted optimization to address complex resource allocation and control tasks. The energy management problem is formulated as a stochastic optimization task with operational and network constraints. The proposed hybrid AI-quantum approach enhances system adaptability, improves energy efficiency, reduces operational losses and carbon emissions, and supports scalable deployment in smart grids, microgrids, and intelligent urban energy infrastructures.