AI-DRIVEN DECISION FRAMEWORK FOR PED

This study introduces an AI-driven framework designed to optimize the operation of Positive Energy Districts (PEDs) through a high-resolution co-simulation environment. Utilizing Python and EnergyPlus the system dynamically adjusts thermal setpoints by processing real-time data in the PED such as energy demand profiles, occupancy levels, indoor temperatures and local PV energy production. The framework was validated across distinct climates, Trondheim and Bari, using diverse building profiles including Bestest 620 and ASHRAE 90.1 (mid-rise apartment and medium office). By comparing standard national policy setpoints (base case) with optimized schedules (flexible case), the impact is measured through three multi-objective evaluation metrics: environmental eustainability, evaluated through CO2 emissions; energy flexibility, quantified via the flexibility index and social sustainability, measured by thermal discomfort hours. This approach demonstrates how smart energy management can successfully balance grid interaction and carbon reduction while maintaining occupant comfort.