The global transition toward sustainable energy is accelerating due to rising environmental concerns and the need to minimize fossil fuel dependency. Microgrids (MGs), which integrate distributed energy resources (DERs) such as solar and wind, provide a flexible and sustainable solution. However, renewable generation variability and dynamic operating conditions challenge energy balance and cost efficiency. This paper presents a machine learning (ML)-based energy management system (EMS) designed to improve forecasting accuracy, optimize scheduling, and enhance MG reliability in both grid-connected and islanded modes. The proposed EMS employs advanced ML models trained on historical data, weather conditions, and real-time parameters to accurately forecast solar and wind power generation. These forecasts feed into a dynamic optimization framework solved using the modified hippopotamus optimizer (MHO), which determines the optimal DER operation and grid interaction. Additionally, a real-time energy trading mechanism is incorporated to enable cost-effective energy exchange decisions based on market prices, load demand, and renewable availability. By maximizing renewable utilization and ensuring seamless operational transitions, the framework reduces reliance on external power sources, lowers operational costs and emissions, and enhances system resilience. Case studies demonstrate the system’s ability to maintain energy balance under varying operating conditions, highlighting its potential for advancing sustainable MG management.

Recent Italian regulation rewards Local Energy Communities (LECs) through two distinct channels: an incentive for virtual shared energy and market access for distributed batteries that provide up-regulation. These incentives often conflict, as charging batteries to maximize the shared energy limits the capacity to provide ancillary services, and vice versa. Currently, quantitative tools for effectively balancing these objectives are lacking respecting the electrical constraints of the low-voltage grid. To fill this gap, a multi-objective optimization is proposed that co-maximizes the revenue from up-regulation and the virtual shared energy reward, under the constraint that the daily energy bill does not exceed a predefined baseline. The implemented mathematical programming formulation utilizes multi-objective second-order cone programming (SOCP) with linear constraints to incorporate the network’s physical constraints. Linearization and decomposition techniques are employed to simplify the problem. By adjusting the physical constraints of the network, the impact of energy communities on the distribution network can also be evaluated with different objectives. The model allows the representation of real peer-to-peer trading, quantifying its effects on both revenue streams and voltage profiles as well as power losses. Trade-off analyses performed on an 84-bus radial distribution network, under both constant and variable prices, show that the framework adapts smoothly to market volatility, highlighting when it is advantageous to prioritize up-regulation and when it becomes preferable to maximize the virtual shared energy incentive.

Encouraging citizens to invest in small-scale renewable resources is crucial for transitioning towards a sustainable and clean energy system. Local energy communities (LECs) are expected to play a vital role in this context. However, energy scheduling in LECs presents various challenges, including the preservation of customer privacy, adherence to distribution network constraints, and the management of computational burdens. This paper introduces a novel approach for energy scheduling in renewable-based LECs using a decentralized optimization method. The proposed approach uses the Limited-memory Broyden–Fletcher–Goldfarb–Shanno (L-BFGS) method, significantly reducing the computational effort required for solving the mixed integer programming (MIP) problem. It incorporates network constraints, evaluates energy losses, and enables community participants to provide ancillary services like a regulation reserve to the grid utility. To assess its robustness and efficiency, the proposed approach is tested on an 84-bus radial distribution network. Results indicate that the proposed distributed approach not only matches the accuracy of the corresponding centralized model but also exhibits scalability and preserves participant privacy.