This paper presents a fault-tolerant secondary and adaptive primary microgrid control scheme using a hybrid multiagent system (MAS), capable of operating either in a semi-centralised or distributed manner. The proposed scheme includes a droop-based primary level that considers the microgrid energy reserves in production and storage. The secondary level is responsible for: a) the microgrid units' coordination, b) voltage and frequency restoration and c) calculation of the droop/ reversed-droop coefficients. The suggested architecture is arranged upon a group of dedicated asset agents that collect local measurements, take decisions independently and, collaborate in order to achieve more complex control objectives. Additionally, a supervising agent is added to fulfill secondary level objectives. The hybrid MAS can operate either with or without the supervising agent operational, manifesting fast redistribution of the supervising agent tasks. The proposed hybrid scheme is tested in simulation upon two separate physical microgrids using three scenarios. Additionally, a comparison with conventional control methodologies is performed in order to illustrate further the operation of a hybrid approach. Overall, results show that the proposed control framework exhibits unique characteristics regarding reconfigurability and fault-tolerance, while power quality and improved load sharing are ensured even in case of critical component failure. PF min minimum power factor (-) m f-P droop coefficient (Hz/W) n V-Q droop coefficient (Hz/VAr) v ocd compensating d-axis voltage by virtual impedance (V) v ocq compensating q-axis voltage by virtual impedance (V) i od inverter output inductor current on the d-axis (A) i oq inverter output inductor current on the q-axis (A) R v resistive part of the virtual impedance (Ω) L v inductive part of the virtual impedance (H) M on exchanged agent messages during one operation cycle-MGCC operational (-) M off exchanged agent messages during one operation cycle-MGCC inactive (-) N ESS total number of ESSs in a microgrid (-) N PV total number of PVs in a microgrid (-) N Load total number of load buses in a microgrid (-) E nom Nominal Battery Capacity (Wh)

This paper proposes a new hybrid control system for an AC microgrid. The system uses both centralised and decentralised strategies to optimize the microgrid energy control while addressing the challenges introduced by current technologies and applied systems in real microgrid infrastructures. The well-known 3-level control (tertiary, secondary, primary) is employed with an enhanced hierarchical design using intelligent agent-based components in order to improve efficiency, diversity, modularity, and scalability. The main contribution of this paper is dual. During normal operation, the microgrid central controller (MGCC) is designed to undertake the management of the microgrid, while providing the local agents with the appropriate constraints for optimal power flow. During MGCC fault, a peer-to-peer communication is enabled between neighbouring agents in order to make their optimal decision locally. The initial design of the control structure and the detailed analysis of the different operating scenarios along with their requirements have shown the applicability of the new system in real microgrid environments.

Accurately forecasting power generation in photovoltaic (PV) installations is a challenging task, due to the volatile and highly intermittent nature of solar-based renewable energy sources. In recent years, several PV power generation forecasting models have been proposed in the relevant literature. However, there is no consensus regarding which models perform better in which cases. Moreover, literature lacks of works presenting detailed experimental evaluations of different types of models on the same data and forecasting conditions. This paper attempts to fill in this gap by presenting a comprehensive benchmarking framework for several analytical, data-based and hybrid models for multi-step short-term PV power generation forecasting. All models were evaluated on the same real PV power generation data, gathered from the realisation of a small scale pilot site in Thessaloniki, Greece. The models predicted PV power generation on multiple horizons, namely for 15 min, 30 min, 60 min, 120 min and 180 min ahead of time. Based on the analysis of the experimental results we identify the cases, in which specific models (or types of models) perform better compared to others, and explain the rationale behind those model performances.

As microgrids have gained increasing attention over the last decade, more and more applications have emerged, ranging from islanded remote infrastructures to active building blocks of smart grids. To optimally manage the various microgrid assets towards maximum profit, while taking into account reliability and stability, it is essential to properly schedule the overall operation. To that end, this paper presents an optimal scheduling framework for microgrids both for day-ahead and real-time operation. In terms of real-time, this framework evaluates the real-time operation and, based on deviations, it re-optimises the schedule dynamically in order to continuously provide the best possible solution in terms of economic benefit and energy management. To assess the solution, the designed framework has been deployed to a real-life microgrid establishment consisting of residential loads, a PV array and a storage unit. Results demonstrate not only the benefits of the day-ahead optimal scheduling, but also the importance of dynamic re-optimisation when deviations occur between forecasted and real-time values. Given the intermittency of PV generation as well as the stochastic nature of consumption, real-time adaptation leads to significantly improved results.