Advanced Technologies for Renewable Energy Systems and Their Applications

1. Introduction

The energy transition requires advanced technologies to plan, manage and operate systems with high penetration of renewables, facing the stochastic variability of sources, the massive integration of electric vehicles (EV), coupling with storage and the emergence of new market models and energy communities, in order to ensure flexibility, resilience and economic efficiency in the short and long term [1,2]. The rapid expansion of solar and wind-based generation has boosted the share of renewables but has exposed limitations in the systems’ ability to cope with rapid variations and uncertainties, requiring new forecasting, operational, and market methodologies to maintain balance and quality of service. The literature highlights that the variable nature of renewables imposes stochastic planning and operation problems, which require stochastic programming, robust scenery, and the enhancement of demand-side and storage flexibility. In parallel, digitalization and electrification (EV, heat pumps) and decentralization create new actors and business models, putting pressure on market design and sector integration to accommodate distributed resources and system services. Stochastic programming and scenario generation approaches have been developed to represent the uncertainty of renewables in long-term systemic models, improving the selection of representative days and the evaluation of solution stability [3,4]. These methodologies allow quantifying the value of end-use flexibility, storage, and demand response in the low-carbon transition, informing investment decisions and multi-scale operation. In local networks and hybrid systems, integrated optimization of generation, storage, and load management across islands or isolated networks reduces the impacts of intermittency and improves reliability and economics [5,6].

The reinforcement of short- to medium-term forecasting methods, with probabilistic scenarios, is central to reducing reserves and diversion costs in markets, mitigating the effects of variability. Smart grids with bidirectional metering, automation, and power quality control are essential to accommodate distributed injections, manage congestion, and ensure voltage and frequency standards under high penetration of renewables [6,7,8].

The integration of electrical and thermal storage is critical to smoothing profiles, shifting power, and providing ancillary services, with trends including phase-change materials, molten salts, and data-driven strategies for performance and cost optimization. Concepts such as “heat commoditization” and international heat markets, with cross-border storage and sharing, increase the systemic efficiency and decarbonization of the thermal sector [9,10].

EV massification introduces mobile flexible loads, enabling V1G/V2G as distributed flexibility features and local grid support, especially when coordinated with PV generation and local markets. The integration of smart charging into energy grids and communities requires management models that internalize dynamic pricing, grid constraints, and service values to the system [11,12,13,14,15]. Local markets and renewable energy communities demonstrate energy and economic benefits by coordinating prosumers with complementary profiles, increasing average self-sufficiency when they combine local generation, storage, and demand response based on price signals [16]. These arrangements depend on favorable regulatory frameworks, IoT for monitoring/control, and the design of rules that promote social inclusion and active participation, ensuring a just transition [17,18,19,20].

Recent market insights underscore the need for flexible products, finer temporal granularity, locational signals, and DER integration to ensure efficiency during network transformation. In hybrid and remote systems, policies, institutional design, and infrastructure investment strategies are as crucial as technological innovations for operational reliability and resilience. PV exceeded 1.6 TW operational in early 2024, producing around 2135 TWh (8.3% of global electricity), reflecting cost declines, manufacturing expansion, and integration into rooftops, utility-scale, and communities. These advances reinforce the urgency of integration solutions, from forecasting and storage to smart markets and grids, to convert capacity growth into stable systemic value [21,22,23]. Case studies in smart communities, second-generation heat networks with geothermal pumps, and seasonal storage show efficiency, cost, and flexibility gains that enable high shares of renewables. Demonstrations in hybrid systems show that load and storage management, combined with intelligent control, sustain continuous and safe operation in isolated networks and microgrids [24,25,26].

Together, the priorities include: stochastic and probabilistic forecasting models; multi-vector integration of electrical and thermal storage; intelligent EV charging with PV coupling; local markets and the design of flexibility incentives; and digitalization of networks for optimization and quality of energy, ensuring an efficient, inclusive, and resilient transition. These are exactly the topics that this Special Issue has aimed to bring together, as a means to foster discussion and exchange of ideas in the closely related and complementary topics.

2. A Short Review of the Contributions in This Article Collection

In this Special Issue it is demonstrated that the orderly management of EV charging in microgrids, when guided by hourly tariffs (TOU), reduces peaks, costs and improves system reliability, as demonstrated by Weng et al., Contributor 1, when modeling EV charging behavior (start/end/charging time and daily mileage via Monte Carlo) and optimizing two objectives—minimizing the peak-to-valley ratio and the total charging cost—under power balance constraints, EV and storage system (ESS) SOC limits, and PV/MV/WTG operation, solved by an elitist genetic algorithm; with 100 EVs, the peak load fell by ~10% (1.90 → 1.71 MW), the peak-valley ratio went from 54.63% to 41.92% and the total charging cost reduced by ~15.8% (6562.54 → 5525.76 CNY), while reliability indicators improved (LOLP 0.017% → 0.005%; ASAI 99.97% → 99.98%), and for larger scales (300–1000 EVs) the “peak shaving and valley filling” trend continued; Furthermore, for an ASAI target of ≥99.5%, the EV acceptance capacity increased from ~590 (disordered charging) to ~770 (ordered charging), demonstrating gains in hosting capacity with price signaling and coordinated dispatch (RBTS Bus6 F4, detailed data and parameters).

In the operational control of real microgrids, Swibki et al., Contributor 2, propose an energy management scheme using “imitation Q-learning” that combines linear programming (LP) as an “expert” to generate demonstrations and reinforcement learning (RL) with sparse rewards (+1/−1 when the action imitates/does not imitate the expert) to decide the ESS setpoint in real time. Applied to the campus microgrid of the Catholic University of Lille (PV 189 + 28 kW, ESS 250 kWh/40–80 kW, 4 buildings, 6 EV chargers), the method was ~80× faster than conventional Q-learning (≈1 min vs. 84 min) and achieved daily costs equivalent to LP under winter TOU (peak €18.04 c/kWh; idle €12.26 c/kWh), robust to faults (PV inverter out of service) and load shedding, demonstrating the practical advantage of imitation learning over pure RL in EMS without explicit predictions.

In terms of power quality and measurement robustness, Song et al., Contributor 3, present an optimal SVD filtering method based on a Hankel matrix that improves the accuracy of current samples in grid-connected inverters under harmonic/interharmonic noise and oscillations. They prove that choosing the optimal dimension l = (N + 1)/2 maximizes the useful signal energy in the first 2p singulars and minimizes the noise energy in this subspace. They also introduce a reconstruction order criterion (“singular threshold difference spectrum”) consistent with the theoretical rank (2p) to avoid losing high-frequency oscillation content. Case studies with fs = 100–400 kHz show a reduction in the overall relative error to ~2.5% in the optimal scenario and preservation of oscillations (e.g., 1 kHz) that would be suppressed by difference spectrum methods, making the method suitable for increasing the robustness of control and measurement in renewable energy converters.

The problem of supraharmonics (2–150 kHz) in LV/MV networks grows with power electronics; Pinto, Grasel and Baptista, Contributor 4, analyze emissions from photovoltaic inverters in a distribution network laboratory (UASTW), distinguishing primary/secondary emissions and observing both narrowband (e.g., 20 kHz—switching frequency) and broadband emissions, in addition to intermodulation when devices coexist (e.g., PV with EV), which increases the amplitude in bands (and can generate peaks ~117–120 kHz, including PLC) and affects PLC equipment and communications; show that topologies and internal impedances (capacitors, skin effect) affect the absorption/propagation of SHs and that measurements according to IEC 61000-4-7 (200 Hz bands) and 61000-4-30 (2 kHz bands) lose information, arguing for risk-weighted aggregated metrics (TSHCw) and the need for specific regulations and mitigation/compatibility techniques (positioning, filters, impedance design).

Regarding the sizing and reliability of interconnected single-phase PV converters, Adamas-Pérez et al., Contributor 5, show that the L filter imposes a phase shift between the grid and inverter voltages, resulting in energy return to the DC link and more demanding sizing of the link capacitor. They derive closed-form expressions for L as a function of ripple current (%riL), considering the main harmonics of unipolar SPWM (fsw = 15 kHz, harmonics at 2β ± 1) and for minimum Vdc as a function of m, ωnsw, Vgrid, and %riL, as well as for the value of the Clink capacitor as a function of ripple voltage (%Δrvdc) and the angle φinv. They demonstrate through simulation and prototype (60 W and 1 kW) that operating with Vdc close to the grid peak minimizes L, and that ignoring the L feedback leads to capacitor undersizing and higher ripple. They validate this with THDi of ~0.23% and efficiency of ~94–98%.

Concluding the dossier, Tian et al., Contributor 6, present a hybrid method for predicting the remaining useful life (RUL) of PEM batteries, combining empirical modal decomposition (EMD) for denoising, a multi-kernel relevance vector machine (MK-RVM) with Bayesian optimization of kernel weights, and an empirical model of voltage recovery in start-stop cycles. Validated on public FCLAB data, the method achieves a 95.35% AR with confidence intervals, outperforming variants without BO and without recovery, and demonstrating the relevance of explicitly and separately capturing the global degradation dynamics and the effects of periodic recovery.

In summary, the articles demonstrate a common thread: the convergence between multi-objective optimization, learning (imitation/reinforcement/Bayesian), advanced metrological techniques (SVD, EMD), and careful modeling of high-frequency and reliability phenomena. By integrating these fronts, the set offers the reader a comprehensive and pragmatic vision of how to design, operate, and monitor microgrids rich in power electronics—maintaining power quality, reducing operating costs, extending asset life, and increasing supply reliability in scenarios with high penetration of renewables and EVs (Contributor 1–6).