Energy management systems (EMS) play a vital role in maintaining the reliability and efficiency of power systems, especially with the growing integration of renewable energy sources and potential grid disturbances. This paper introduces an innovative framework for a resilient EMS that combines Model Predictive Control (MPC) with an adaptive learning mechanism. This integration ensures effective energy resource management under both normal and outage conditions. The proposed system adapts to evolving conditions by learning from past events, considering critical load management, outage probabilities, dynamic electricity tariffs, and the status of energy storage systems. Through this approach, the system can make informed decisions to optimize performance and enhance resilience. Simulation results indicate that this advanced EMS significantly boosts system resilience and cost efficiency, providing a robust solution for modern power systems facing increasing complexity and variability. By leveraging adaptive learning, the EMS continuously improves its predictive capabilities, ensuring optimal energy distribution and minimizing disruptions, thereby contributing to the overall stability and sustainability of the power grid.