.
Accurate day-ahead load forecasting (DALF) is essential for the reliable and economic operation of emerging regional power markets. Jeju Island, operating a pilot electricity market with high renewable energy penetration and non-industrial consumption characteristics, presents unique forecasting challenges due to irregular demand fluctuations and weather-sensitive load behavior. In this study, DALF is operated only for ordinary days, excluding holidays. This study proposes a hybrid ensemble forecasting framework tailored to Jeju’s operational context by integrating similar-day, statistical (ARIMAX), and artificial intelligence (LSTM, XGBoost) models. Each model is designed to capture distinct patterns and dependencies in demand data. The ensemble weights are dynamically optimized using the sequential least squares programming (SLSQP) algorithm based on historical forecasting performance. One year of operational data from Jeju Island is used for validation, showing that the ensemble model achieves substantial improvement in forecasting accuracy compared to standalone approaches. The results confirm the effectiveness of the proposed region-specific DALF framework for renewable-dominant and isolated power systems.
High‐performance polypropylene (PP) insulation is essential for next‐generation high‐voltage direct current (HVDC) systems. This study presents a scalable, solvent‐free melt grafting strategy to covalently incorporate thermally stable aromatic voltage stabilizers (VSs)—2‐vinylnaphthalene (VN), 1,1‐diphenylethylene (DPE), and 4‐vinylbiphenyl—into isotactic PP. VSs are rationally selected based on their high boiling points, π‐conjugated aromatic structures, and vinyl functionality, enabling compatibility with melt processing and uniform bulk functionalization of PP. Electrical tests showed that VN and DPE markedly improved volume resistivity, suppressed leakage current, and enhanced DC breakdown strength, while their effects diminished at loadings above 1.0 wt.%, respectively. Thermally stimulated depolarization current confirmed that these gains originated from deeper and well‐distributed trap formation. Quantum chemical and finite element simulations further revealed how geometry‐tailored VSs modulate trap formation and suppress space charge accumulation, linking molecular‐scale trapping to macroscopic charge dynamics. This work highlights a practical and recyclable route to engineer PP insulation with superior insulation reliability, offering new insights into stabilizer structure‐property relationships for HVDC applications.
Controlling where electromagnetic energy is ultimately absorbed is more challenging than controlling where it is focused, and this challenge is especially critical for wireless power transfer (WPT) in complex environments. Conventional beamforming and time-reversal (TR) approaches can achieve spatial focusing in such settings but suffer from low transfer efficiency, as energy is largely reradiated or redistributed after focusing. Coherent perfect absorption (CPA) offers a fundamentally different approach by enforcing complete absorption through engineered interference. However, prior experimental demonstrations have been limited to single-loss channel scenarios, preventing selective power delivery in realistic environments where multiple loss channels exist. Here, we introduce an ultraefficient, port-selective CPA-based WPT scheme in complex wave environments, which we experimentally demonstrate using a quasi-2-D wave-chaotic cavity. By employing rectifiers with distinct operating frequency characteristics, we deliberately break the degeneracy among competing loss channels, thereby establishing an effective single-loss channel CPA condition within an otherwise multiloss channel environment. As a result, CPA can be enforced selectively at a chosen rectifier port. Under each CPA condition, near-perfect absorption is achieved at the intended rectifier port, enabling position and thus frequency-selective WPT with high end-to-end efficiency.
Although previous studies have demonstrated that system inclination can influence the melting behavior of phase change materials (PCMs) in shell-and-tube latent heat thermal energy storage (LHTES) systems, the effect of independently inclining only the internal tube under a fixed gravitational direction has not been clearly elucidated. This study presents a three-dimensional numerical investigation to examine the influence of tube inclination on the melting characteristics of RT35 in a shell-and-tube LHTES system. A total of 12 cases were analyzed by varying the inclination angles of straight and nozzle-shaped tubes (theta = 0 degrees, 3 degrees, 4 degrees, 5 degrees, 6 degrees, and 7 degrees), with a focus on their effects on thermal performance throughout the melting process. The enthalpy-porosity method was employed to simulate the PCM melting process. The analysis considered liquid fraction evolution, total melting time, Nusselt number evolution, heat flux variation, and mean power. The results indicate that tube inclination significantly affects the melting behavior of PCM. At an inclination angle of 7 degrees, the straight and nozzle-shaped tubes reduced the total melting time by 18.62% and 21.75%, respectively, compared with the noninclined configuration. Furthermore, the mean power increased by 26% and 32.6% for the straight and nozzleshaped tubes, respectively. As the inclination angle increased, the surface-averaged Nusselt number increased after 2,700 s, indicating sustained enhancement of heat transfer performance during the later stages of melting.
In power systems, alternating current optimal power flow (AC-OPF) has been a challenging problem for decades due to its nonconvexity, but fast and efficient solutions are even more needed because of high penetration of large scale renewable generation and load growth. Recently, neural networks (NN) have gained attention in solving AC-OPF, but it is still in an early stage to be applicable for real and large-scale power system operation with topology-changing characteristics. To end this, we propose a novel framework called GraphOPF that considers topology-adaptability, scalability, NN training time, self-supervision, and feasibility altogether. Extensive experiments show that the proposed framework against the baselines is up to 200 times faster in NN training and up to 66 times faster in solving AC-OPF for large-scale power systems including the real Korean power system, while achieving more than 99
