Abstract:Based on the offline mission planning model and results for the application of dynamic scenario algorithms, an improved hierarchical distributed mission planning framework was proposed to give the heterogeneous multi-UAV (unmanned aerial vehicle) collaborative mapping system the decision-making capability to face the dynamic environment. Among these, the mission valuation method based on pre-planned trajectory took the global cost into account, and the valuation results were updated synchronously by the local auction algorithm with restricted communication, avoiding the mission conflict and local optimum. The joint correction method of trajectory based on rolling time-domain predictive control satisfied the requirements of dynamic mapping and obstacle avoidance. Through numerical simulation in a variety of circumstances, the applicability and dependability of the planning algorithm were confirmed.

Abstract:To mitigate the degradation of tracking accuracy induced by numerous false association ghost points, a dual-level ghost point elimination and target tracking algorithm integrating angular measurement with target motion characteristics was introduced. The proposed algorithm employed a cooperative localization strategy that prioritizes association before estimation. By establishing a mapping relationship between angular measurement noise and localization error, a field-of-view grid map and an energy accumulation matrix were constructed. Through a detailed analysis of the spatial geometric distribution characteristics of real targets and false association ghost points within the field of view, a novel elimination criterion based on Hough transform theory was developed, facilitating the primary coarse elimination of ghost points. Additionally, by examining the distribution characteristics of target localization ambiguity regions and motion features, a predictive tracking gate was constructed using motion parameter identification, enabling the secondary fine elimination of ghost points at the kinematic level. Experimental results demonstrate that the proposed algorithm significantly enhances target tracking accuracy, achieving a ghost point elimination rate of 91.82%, thereby effectively addressing the false association problem in multi-target tracking.

Abstract:The pipeline-injector component, located downstream of the main valve of a liquid rocket engine, was simplified to an exit-contracted pipe. In order to study the liquid oxygen chill down process of the pipe, two groups of experiments were conducted with high mass flux (3750 kg?m-2?s-1) and low mass flux (1800 kg?m-2?s-1), respectively. Based on the experimental data, flow pattern development diagrams of the internal fluid were plotted and analyzed during the chill down process. Moreover, the heat transfer coefficients at Leidenfrost points (hLFP) were fitted. The detailed conclusions are presented as follows. There are three liquid rewetting patterns during the chill down process including I, II, and III, which are controlled by the quenching fronts at the inlet and the outlet, the quenching fronts in the middle, and the high pressure filling-in of the liquid, respectively. While the rewetting patterns at the front 1/4 of the pipe are always I for the experimental conditions, the rewetting patterns at the other sections of the pipe change with increasing pressure. For the middle and the rear sections, when the pressure is lower than 1.181MPa, the rewetting patterns are I or II. And when the pressure is equal to or higher than 1.181MPa, the rewetting patterns of these sections tranform into I at low mass flux and III at high mass flux. With an error of less than 34%, certain correlation is employed to predict the hLFP for 4 measurement points at the 0.15 and 0.30 cross-sections.

Abstract:Computer-aided aerodynamic design is crucial for aircraft geometry optimization. To further improve the efficiency of aerodynamic characteristic modeling, an aircraft-oriented intelligent aerodynamic coefficient prediction method, AeroPointNet, was proposed. A three-dimensional point cloud representation of geometric models was employed as input, and a neural network architecture was constructed to efficiently extract both local and global geometric features. To capture variations in flow conditions, physical information was fused with geometric features, and two weighted attention mechanisms were introduced to dynamically adjust the weights, by which the problem of weight imbalance was effectively addressed. Experimental results show that AeroPointNet achieves a computational efficiency improvement of over three orders of magnitude in aerodynamic coefficient prediction compared with traditional numerical methods. The mean relative errors of lift and drag coefficients are kept below 5%.

Abstract:To achieve swept-back, variable camber, and torsional deformation of the aircraft's deformable wing, Therefore, a rotating re-entry superstructure with adjustable elastic parameters is proposed. The rotational re-entrant metastructure is composed of an inwardly concave octagon rotated 90° and extended ligaments. It relies on the extended straight arm ligaments to topologically fill the airfoil section in space to form a skeleton deformation structure. Based on the Mohr's theory, a theoretical model of the relative elastic modulus and Poisson's ratio of the rotational re-entry metastructures along three directions in space is established. The finite element model of the rotational re-entry metastructures were established by ANSYS software, and five rotational re-entry metastructures prototypes were processed by 3D printer technology. The theoretical, simulation and experimental results were compared respectively. The maximum relative error of the relative elastic modulus along the x, y and z directions was 9.88%, indicating the accuracy of the theoretical model and the simulation model. The effects of geometric parameters on the elastic parameters of metastructures were analyzed, and it was found that the aspect ratio and the structural Angle had a great influence on the mechanical parameters of Poisson's ratio, which could provide a theoretical basis for the application of deformable wing skeleton of aerospaceplane.

Abstract:For the problem of trajectory planning for multi-vehicle collaborative formation in the gliding mid-flight phase, a two-phase cooperative formation trajectory planning method based on the "Coordinated Assembly and Formation Maintenance" strategy was proposed. In the coordinated assembly segment, a trajectory planning method based on a coordination-execution dual-layer framework was designed. The coordination layer included three modules: spatiotemporal capability boundary forecasting, rendezvous point information calculation and distribution, and adaptive correction of rendezvous point information, to quickly determine the rendezvous point information while considering the vehicles' control capabilities. The execution layer then designed a trajectory planning method considering spatiotemporal full-state constraints to achieve high-precision assembly of multiple vehicles, providing a favorable initial situation for formation maintaining. In the formation maintenance segment, using virtual altitude and heading angle as coordination information, a trajectory planning method based on fixed-time consistency was designed to realize long-range formation maintenance. Simulation results show that the proposed trajectory planning method demonstrates excellent high-precision assembly capability, long-range formation maintaining ability, and adaptability to multiple tasks.

Abstract:Data-driven generalizable aerodynamic analysis models demonstrate strong capability in performing real-time reliable aerodynamic analysis for any geometry under arbitrary aerodynamic conditions. It represents a key technology for achieving high-speed intelligent optimization design of aircraft. However, the construction of high generalizable analysis models for complex aerodynamic geometry remains severely constrained by the curse of dimensionality, which necessitates large number of training data, thereby impeding practical implementation and broader application. This paper presents two studies in data-driven airfoil and wing optimization design. By establishing a rational parametric characterization of the aerodynamic shape design space, the adverse effects of the curse of dimensionality were effectively circumvented. A data-driven aerodynamic analysis model with demonstrable generalizability was constructed utilizing approximately 100,000-scale CFD training datasets, enabling high-efficiency optimization design of relevant aerodynamic shapes. As a summary of the author"s preliminary work, this paper aims to inspire the academic community to conduct further research in the field of data-driven aerodynamic optimization design and to promote the advancement of intelligence in the field of aircraft design.

Abstract:To address the challenge of rapid flow field prediction for supercritical airfoils, this paper proposes a hybrid deep learning model, termed TransCNN-FoilNet, based on two main approaches in current deep learning flow field prediction models: convolutional neural networks and Transformers. The model is capable of predicting the flow fields of supercritical airfoils with varying thicknesses at different angles of attack, achieving up to a 79.5% reduction in the mean absolute error compared to the baseline model. Additionally, the study introduces a new combined loss function for training the flow field prediction model, referred to as the weighted L1SSIM loss function. The results demonstrate that this loss function can improve the prediction of lift and drag coefficients, with the relative error in drag coefficient reduced by up to 17.8%. The proposed model achieves improved prediction accuracy and generalization performance while reducing complexity, providing a promising tool for fast and reliable flow field prediction of supercritical airfoils.

Abstract:Hypersonic internal compression inlets are typically designed using streamline tracing technology based on the basic flowfield, and the quality of the basic flowfield design directly determines the performance metrics of the inlet. In this study, the quasi-uniform B-spline method is applied to achieve the parametric design of the internal compression basic flowfield. A fast prediction model for the internal compression basic flowfield is established based on the residual neural network architecture of deep learning, realizing the goal of "parametric design-flowfield prediction." The predicted flowfield cloud images are quantitatively evaluated using image quality assessment methods, and key flowfield characteristic parameter distributions are extracted to enable rapid acquisition of flowfield cloud images and characteristic parameter distributions based on design parameters. The results show that the developed basic flowfield fast prediction model achieves high prediction accuracy for flowfields corresponding to given geometric parameters. The average peak signal-to-noise ratio (PSNR) of the predicted total flowfield is 42.51 dB, and the average structural similarity index (SSIM) is 0.9973. Additionally, the model can effectively extract key flowfield characteristic parameter distributions from the predicted results, providing robust support for the rapid design and optimization of internal compression basic flowfield.

Abstract:In order to improve the heat dissipation capacity of the electronic equipment system of the near space vehicle and solve the problem of high heat flux of integrated and miniaturized electronic devices，the paper had a study on the heat transfer performance of dual synthetic jets actuators, by making use of two-phase flow composed of micron-particle and water and their combination. The mechanism of enhancing heat transfer capacity with dual synthetic jets actuator is was studied and analyzed. The flow process of the dual synthetic jets actuator and Cu-water micron particle fluid in the tube is was modeled, and the influence of five particle concentrations on the enhanced heat transfer capacity of the fluid is was simulated by the single Euler model. CUO-water and Al2O3-water micron particle fluids are were simulated. The results show that the heat transfer capacity of fluid can be enhanced by the dual synthetic jets. The heat transfer capacity in-creases with the increase of micron particle concentration. The two-phase flow heat transfer capacity of different metal particles varies with the thermal conductivity of metal particles. In the example, a microparticle fluid of Cu particles with a dual synthetic jets actuator on and a particle concentration of 8% is synthesized, and the chip temperature is reduced from 328.225K to 303.816K.