Jiangbo Wu

Deep learning has become an important tool for wind power forecasting because it can help improve wind energy utilization and support reliable grid-connected operation. For wind farms, accurate turbine-level forecasting depends on spatial interactions among turbines and temporal evolution of historical operating data. In this study, a spatio-temporal forecasting framework is developed by combining a Graph Attention Network with a Temporal Convolutional Network. The graph attention module describes the neighborhood relations among turbines and learns their influence strengths adaptively, while the temporal convolution module extracts temporal patterns from multivariate SCADA sequences for multi-step prediction. On this basis, the learned attention weights are further used to define a node influence metric. This makes it possible to identify a small set of key turbines and use only their historical data to predict the future power output of the whole wind farm. The proposed framework is evaluated using one year of SCADA data from 134 turbines. A sliding-window dataset is constructed, and the model is tested on the training, validation, and test sets. The results show that the method can capture the spatio-temporal dependencies within the wind farm and still provide effective farm-wide forecasting when only limited observation nodes are available. The value of this work lies in organizing existing techniques around a practical wind farm forecasting task and in providing an interpretable prediction strategy based on key turbine selection, rather than in proposing a fundamentally new theoretical model.

Bionic self-propelled droplet transport technology holds immense potential in microfluidic systems. However, existing passive strategies are often constrained by contact line pinning at vertical structures and insufficient durability of functional coatings, while frequently overlooking the kinetic decay induced by solid–liquid triboelectric charging. To address this, this study proposes a self-propelled droplet platform capable of velocity regulation. On one hand, femtosecond laser etching was employed to construct a dual-slope coupled, hierarchical wedge structure. This reconfigures vertical steps into smooth inclines to mitigate droplet pinning, while simultaneously providing physical anchorage points for nano-hydrophobic coatings on the etched micro/nano-textured surface. On the other hand, to counteract the electrostatic deceleration effect during long-distance transport, a back-grounded conductor structure was introduced. Leveraging electrostatic induction and shielding mechanisms, this design significantly reduced the cumulative surface potential from −198 to −98.7 V after 60 consecutive transport cycles. Furthermore, by switching the mode of the back conductor, the platform enables droplet velocity control while simultaneously outputting stable triboelectric signals. The average droplet movement speed on the platform reaches 44.77 mm s−1. This synergistic biomimetic platform unifies wear-robust, residue-free, self-propelled transport with programable triboelectric velocity gating, enabling residence-time-defined droplet microreactors/digital microfluidics, and in situ self-powered sensing.

Tip leakage flow is a critical factor constraining the efficiency and stability of supercritical carbon dioxide (SCO2) centrifugal compressors. To address this challenge, this study proposes a non-uniform blade tip clearance design and establishes a parametric closed-loop optimization framework based on the NSGA-III algorithm to investigate its aerodynamic performance enhancement and unsteady flow dissipation suppression mechanisms. The results demonstrate that the optimized non-uniform clearance configuration significantly improves compressor performance, increasing the pressure ratio and isentropic efficiency by 3.55% and 5.65%, respectively, at the design point. Unsteady flow analysis reveals that the non-uniform clearance actively modulates the pressure gradient distribution in the tip region, effectively suppressing the scale and intensity of the tip leakage vortex and weakening the shear mixing between the leakage flow and the main flow. Consequently, the turbulent kinetic energy dissipation rate in the diffuser region is reduced by 33.08%. Entropy generation analysis further confirms a substantial reduction in irreversible energy losses in the tip region, with the average entropy generation rate at 0.9 BH (Blade Height) decreasing by 46.78%. Furthermore, the area-weighted average turbulent kinetic energy on the S1 stream surface is reduced by 10.4% globally, indicating a comprehensive improvement in flow stability. This study elucidates the physical mechanism by which non-uniform blade tip clearance enhances SCO2 compressor performance through flow structure reconstruction, providing a new theoretical basis and optimization pathway for high-performance turbomachinery design.

Background: Conjugate heat transfer in supercritical hydrocarbon fuels within microchannels is strongly influenced by sharp thermophysical property variations and chemical reactions, posing significant challenges for accurate numerical prediction. To address this, a high-fidelity solver is developed within the OpenFOAM framework, incorporating detailed reaction mechanisms and demonstrating robust stability under steady supercritical conditions. In particular, to mitigate numerical oscillations and accuracy loss in the pseudo-critical region, a high-order variable-property transport model, based on an eight-segment, seventh-order polynomial formulation, is introduced and integrated in the solver. This model is tightly coupled with the Peng–Robinson equation of state and a simplified cracking mechanism, enhancing both stability and predictive capability for highly nonlinear supercritical reacting flows. The proposed approach is applied to compare the coupled thermal–hydraulic–chemical behavior of a baseline straight channel and a trapezoidal-ribbed configuration under supercritical pressure. Unlike conventional fully distributed rib arrangements, the proposed design achieves heat transfer enhancement using a limited number of rib elements. The ribs promote elevated turbulent kinetic energy in their wake and intensify near-wall mixing. As a result, peak wall temperature is reduced by 30–40 K under identical conditions, effectively suppressing localized hot spots. Although improved cooling slightly decreases the cracking conversion rate, the design markedly lowers the risk of fuel coking by eliminating high-temperature regions, thereby enhancing overall thermal management. The performance evaluation criterion (PEC) remains close to or slightly above unity across different mass flow rates, indicating a modest but meaningful thermo-hydraulic benefit and practical engineering potential.

To investigate transient flow instabilities in parallel-channel regenerative cooling systems subjected to nonuniform heat flux, a three-dimensional transient numerical model was developed to couple variations in supercritical fluid thermophysical properties with endothermic pyrolysis kinetics. The spatiotemporal evolution of RP-3 fuel within parallel channels was analyzed, and the role of a midstream interconnection structure in mitigating flow maldistribution was clarified. During the initial heating stage, the viscosity reduction of the supercritical fuel produced a drag-reduction effect that temporarily maintained a nearly uniform flow distribution. As the wall temperature increased and the pseudocritical region approached, the sharp decrease in density markedly increased the acceleration pressure drop, disrupting the pressure balance between channels. In combination with the progressive accumulation of pyrolysis products, this process led to a positive feedback loop characterized by flow rate reduction, insufficient heat absorption, and increasing flow resistance. The introduction of a midstream interconnection enabled pressure-driven lateral mass transfer between channels. The resulting crossflow was directed predominantly from the high-heat-flux channel toward the low-heat-flux channel, providing a release path for overheated, low-density, and strongly cracked fluid in the high-heat-flux channel and thereby weakening the downstream accumulation of thermal and compositional nonuniformities as well as the associated resistance amplification. Compared with the configuration without interconnection, the stage-averaged maximum flow-deviation coefficient decreased by 17.4% during the pseudocritical transition stage and by 48.3% during the deep-pyrolysis stage.