This paper presents a multiobjective optimization method for blackstart restoration in power distribution systems, aiming to minimize both Energy Not Supplied (ENS) and total recovery time. ENS measures the cumulative energy not delivered during the outage, while recovery time reflects the duration required to re-energize all network components. The proposed approach uses a Multiobjective Genetic Algorithm (MOGA) applied to the IEEE 123-bus feeder, with dynamic co-simulation using MATLAB and OpenDSS under realistic operating conditions. The results show a clear trade-off between objectives, with Pareto-optimal solutions achieving up to 35% ENS reduction and up to 40% recovery time improvement, providing practical guidance for restoration planning in modern distribution networks.

This paper presents the design and evaluation of RuralVPP, a decentralized Virtual Power Plant (VPP) architecture designed for rural energy communities. The system integrates ten semi-autonomous municipalities into a coordinated structure based on a dual-layer market framework, consisting of Local Energy Markets (LEMs) and a Supra-Municipal Market. Energy transactions within and between communities are managed through smart contracts implemented on a permissioned blockchain platform using Hyperledger Fabric, ensuring secure, transparent, and auditable settlements. The communication infrastructure is based on the IEC 61850 standard, enabling interoperability among distributed energy resources (DERs), smart meters, and flexible loads. To support efficient market operation and grid management, the system incorporates advanced forecasting techniques using deep learning models, including Long Short-Term Memory (LSTM) networks and Transformer architectures. These models are trained on real energy generation and demand data collected from the participating communities. Results show high forecasting accuracy, effective automation of energy trades, and enhanced local energy utilization. The proposed solution improves energy resilience, lowers operational costs, and provides a scalable reference model for decentralized rural energy systems based on blockchain and artificial intelligence.

Photovoltaic energy is very sensitive to the variation of the climatic parameters such as the solar radiation and the ambient temperature, which are characterized by their intermittent behavior. Thus, to optimize the use of the electrical power generated by the PV plant, it is necessary to have a good prediction of the PV generation. In this paper, 15-minute interval experimental measurements of the solar radiation, the ambient temperature and photovoltaic generation that correspond to the four seasons (winter, spring, summer and autumn) are used to develop models for the prediction of the PV power of a 10 kW polycrystalline photovoltaic plant installed in Mostoles (Madrid, Spain). Hence, a combination of prediction models which are the double exponential smoothing and autoregressive integrated moving average (ARIMAX) are applied to predict the PV generation. The results obtained illustrate the good performance of the proposed prediction models as a response for several weather measurements.