Hangar-based maintenance, repair and overhaul (MRO) operations expose aircraft to close-proximity interactions with ground support equipment (GSE), where operator errors can cause structural damage and disruption. This study presents a monocular vision framework for collision-risk monitoring in a smart hangar, coupling instance segmentation, domain-adapted monocular depth inference and deterministic clearance reasoning. A paired dataset comprises synthetic top-down hangar renderings with pixel-aligned red-green-blue (RGB)-depth supervision and a matched miniature testbed containing scaled aircraft/GSE models and 3D-printed calibration objects, imaged by custom-mounted RGB cameras. The dual-head pipeline comprises YOLOv8-seg for six-class instance masks and DepthAnything-V2 for metric depth under the trained model convention. On the scene-disjoint synthetic test split, the custom-trained depth model achieves 0.135 m MAE, compared with 8.385m for pretrained weights, without materially increasing depth-inference time. WIthout dense RBG-camera depth ground truth, testbed-image transfer is reported as an ordinal/structural calibration-object proxy: relative step ordering is often preserved, whereas inter-step metric spacing is compressed and deemed inadequate for direct clearance-threshold decisions. Custom YOLOv8-seg variants achieve mean Intersection over Union (IoU) of 0.88–0.92 across synthetic and testbed splits, while the off-the-shelf baseline yields near-zero overlap in transfer tests. Collision inference is evaluated on deterministic 30-image samples per synthetic/testbed scenario, combining depth-derived height envelopes and boundary-only ground-plane distances to produce clear/warning/alert states and overlays. Median no-render compute latency is 0.736–1.037 s; profiling identifies post-processing, particularly GSE clustering and geometry construction, as the dominant runtime component. The study provides an auditable perception-to-geometry pipeline and highlights the need for uncertainty-aware threshold calibration before deployment.

This study investigates the effects of burnishing force, feed rate and the number of passes on surface roughness, microhardness and fatigue life of conventionally turned Inconel 718 alloy via the roller burnishing (RB) method. The process parameters were evaluated at four levels using a Taguchi L16 experimental design. To improve the statistical reliability of the fatigue analysis, fatigue tests for each experimental condition were repeated three times (n = 3). Compared to untreated specimens, roller burnished specimens exhibited up to a 1018% improvement in fatigue life, while the average fatigue life improvement was calculated as 383%. In addition, the surface roughness decreased from 3.2 µm to 0.12 µm, whereas the microhardness increased from 288 HV to 418 HV after the burnishing process. According to the ANOVA results, burnishing force was identified as the most influential parameter affecting fatigue life, with a contribution ratio of 93.53%. The obtained results demonstrate that RB is an effective surface enhancement method for improving the fatigue performance and surface integrity of Inconel 718 components used in demanding engineering applications.

The phenomenon of ground vortex on engines operating near the ground has been studied for several decades, but the phenomenon of sand ingestion coupled with ground vortex has been less noticed. In this paper, an Eulerian–Lagrangian particle tracking scheme is developed to study the coupling ingestion characteristics of sand and ground vortex into a nacelle inlet near the ground. The influences of freestream air speed, side-wind speed, an aerodynamic interaction plane (AIP) Mach number, ground clearance, and particle properties are comprehensively discussed. It is found that the existence of ground vorticity is the precondition of particle ingestion into the inlet. Finally, a mechanism of particle ingestion coupled with ground vortex is revealed, which can be divided into three stages, namely accumulation, uplift and ingestion. Both the ground vortex and stream tube near the ground are found responsible for the ingestion of particles into the inlet.

A mass- and momentum-conserved secondary injection model (MMC-SIM) is presented for thrust vector control (TVC) analysis in axisymmetric nozzles. TVC modelling provides fast thrust vector predictions for control design, preliminary nozzle sizing, and integration into flight simulation frameworks. Existing low-order models rely on simplifying assumptions regarding the secondary jet exit state, which limits reproducibility and can compromise predictive accuracy. MMC-SIM addresses this limitation by reformulating the equivalent obstruction height problem. Mass conservation and momentum conservation provide two independent constraints that are solved simultaneously to determine both the effective injection height and the secondary jet exit state, eliminating the need to prescribe downstream jet conditions a priori. These constraints are embedded within a blunt-body framework to predict boundary layer separation, wall pressure distribution and thrust vector characteristics. Computational fluid dynamics (CFD) simulations are used to assess the modelling assumptions and examine the flow structures. MMC-SIM shows strong agreement with experimental results, yielding average and maximum lateral force prediction errors of 3.3% and 4.8%, respectively. Compared to current leading models, MMC-SIM substantially improves the prediction of the pressure-driven lateral force while providing a transparent and reproducible modelling framework suitable for engineering design applications.

This study proposes an efficient aerodynamic prediction and optimisation framework integrating blade element momentum theory (BEMT), a Bayesian-optimised aerodynamic residual network (BO-ARN), and a genetic algorithm (GA) for high-altitude long-endurance solar UAV propellers under stratospheric low-Reynolds-number conditions. An aerodynamic database was generated using Latin hypercube sampling and XFoil. A fully connected residual network was developed to learn nonlinear mappings from Reynolds number, Mach number and angle-of-attack to lift and drag coefficients, while tree-structured Parzen estimator Bayesian optimisation selected network depth, width, learning rate, weight decay and batch size. The trained BO-ARN surrogate was embedded in BEMT to replace repeated XFoil evaluations, and a GA was then used to optimise the spanwise chord and twist distributions under motor power and torque constraints. Compared with a Bayesian-optimised multilayer perceptron, BO-ARN reduced the RMSEs of lift and drag coefficient prediction by 25.65% and 37.80% on the training set and by 4.06% and 7.71% on the test set, respectively. For one propeller operating condition, BO-ARN-supported BEMT reduced the calculation time from 43.322 s to 0.411 s. After optimisation, cruise thrust increased from 17.89 N to 19.49 N, corresponding to an 8.94% improvement, while system propulsion efficiency increased from 68.75% to 70.59%, with the motor constraints satisfied. These results demonstrate that the proposed framework provides an accurate and computationally efficient approach for propeller-motor matching and aerodynamic shape optimisation of high-altitude UAV propulsion systems.

The nonlinear behaviour and parameter coupling inherent in gas turbine engines pose persistent challenges in the development of accurate thrust prediction models for unmanned aerial vehicle (UAV) propulsion systems. This study investigates the application of ensemble regression methods to address these complexities and enhance predictive accuracy using experimental data obtained under controlled test conditions. The evaluated models include bootstrap aggregating (bagging), adaptive boosting (AdaBoost), light gradient boosting machine (LightGBM), CatBoost, stacked generalisation (stacking) and a voting regressor ensemble. Each model was trained and evaluated using experimentally measured thermodynamic and operational parameters across a range of engine operating conditions. Cross-validation results revealed that stacking achieved the lowest prediction errors with a root mean square error (RMSE) of 0.0635 and a mean absolute error (MAE) of 0.0442, while AdaBoost demonstrated a very high cross-validated coefficient of determination (${R^2}$) of 0.9998. CatBoost achieved ${R^2}$ values above 0.99, and voting produced ${R^2}$ values close to 0.99. LightGBM exhibited comparatively higher prediction errors and greater variability across folds. These results indicate the potential effectiveness of stacking and adaptive boosting approaches for thrust estimation in UAV gas turbine propulsion systems, while also demonstrating that ensemble performance depends strongly on algorithm selection and dataset characteristics.

X-wing unmanned aerial vehicles (UAVs) possess the unique capability to perform both bank-to-turn (BTT) and skid-to-turn (STT) manoeuvres by combining lift-vector tilting with sideslip-induced lateral force generation, thereby offering superior manoeuvrability compared with conventional fixed-wing platforms. In this study, a high-fidelity numerical investigation of X-wing manoeuvring performance is conducted using a six-degree-of-freedom (6-DOF) flight-dynamics model integrated with validated aerodynamic coefficients generated from USAF Digital DATCOM. The coupled effects of front-wing and rear X-tail anhedral–dihedral angles are systematically examined under both calm and crosswind conditions. The results indicate that increasing the combined angles significantly enhances lateral control authority and reduces the turning radius during STT manoeuvres, albeit at the expense of increased trajectory oscillations and heightened sensitivity to wind disturbances, whereas BTT manoeuvres exhibit comparatively smoother and more robust behaviour in crosswind environments. To further isolate the underlying physics, a dedicated roll- and yaw-control verification framework is developed to distinguish intrinsic aerodynamic characteristics from controller-induced effects, revealing distinct mechanisms for crosswind compensation. In addition, target interception simulations demonstrate that larger anhedral–dihedral angles improve interception performance, with the 90° configuration reducing interception time by approximately 16% and the minimum target separation distance by nearly 94%. Overall, this study elucidates the coupled aerodynamic–control mechanisms governing X-wing manoeuvrability and provides quantitative guidelines for the design of high-agility, wind-resilient UAV systems.

This paper addresses the challenge of obtaining accurate finite element model updating (FEMU) results for complex flexible aircraft structures without relying on prohibitively large numbers of direct finite element evaluations. An adaptive two-stage framework is proposed by integrating Bayesian regularisation neural networks (BR-ANN) with trust-region resampling. In the global stage, a coarse-grained surrogate model is constructed via Latin hypercube sampling (LHS) and Bayesian inference, enabling the non-dominated sorting genetic algorithm II (NSGA-II) to identify the Pareto-front region containing promising candidate solutions. Subsequently, a trust-region strategy performs high-fidelity resampling within the local neighbourhood of the knee-point solution, establishing a refined surrogate model that reduces local surrogate bias and improves parameter convergence. Validation using the GARTEUR SM-AG19 benchmark aircraft demonstrates that the proposed method effectively mitigates the smoothing effect inherent in conventional global surrogate models while maintaining computational efficiency. The average modal frequency error was reduced from 4.56% to 2.39%, and the method supports physically interpretable updating of material and equivalent beam parameters, providing a reliable baseline for subsequent aeroelastic predictions.

In this research article, integrated and multi-objective flight performance enhancement for a turbojet powered unmanned aerial vehicle (TJP-UAV) is considered. For this purpose, its autonomous flight control system (AFCS) and its (fuel mass/total mass) ratio are redesigned in order to enhance its trajectory tracking performance of the flight controller and also its general flight performance (i.e. maximum range, maximum endurance, and ceiling altitude). A TJP-UAV is manufactured in Erciyes University and Iskenderun Technical University drone laboratories and called as ERU-ISTE-TJP-UAV in this research article. Its AFCS parameters (i.e. gains of the relevant PID controllers) and its (fuel mass/total mass) ratio are stochastically revised in an integrated and multi-objective way for maximization of both of trajectory tracking performance of the flight controller and general flight performance by using a certain stochastic optimization tool. The optimized parameters are then used to create a simulation environment of autonomously flying TJP-UAV. A novel framework for application of integrated and multi-objective redesign concept in order to determine optimum values of AFCS parameters and fuel mass of a TJP-UAV is developed in this article. Another considerable contribution of this research article is implementation of the c-SPSA optimization for previously mentioned aim. These chief contributions result in much less energy consumption and satisfy green environmental demands. Additionally, 29.44% improvement in total cost index, 36.96% improvement in autonomous flight cost index and 24.78% improvement in general flight cost index are obtained for this optimized TJP-UAV compared to the default control and propulsion system design.

Altitude adjustment for wing-in-ground (WIG) crafts during obstacle avoidance or mission execution necessitates a rapid response followed by precise stabilisation at the target altitude. Conventional approaches typically treat the minimisation of response time and the minimisation of deviation from the target altitude as two separate objectives. These are assigned weights, and their performance is balanced through iterative trial-and-error adjustment of these weights, ultimately finding a target trajectory after multiple optimisation runs. This paper proposes a novel reward function that yields, via a single deep reinforcement learning pass, a trajectory minimising both the response time and deviation for WIGs’ altitude changes. Formally, this function bounds the reward value based on the deviation magnitude and incorporates triangular numbers to assess the response time. Its underlying principle ensures that the reward gain from reducing the response time outweighs the reward loss incurred by the potential increase in accumulated deviation resulting from that same reduction. The validity of this reward function is verified through mathematical derivation. A numerical example involving a WIG descending to a lower altitude is presented, using proximal policy optimisation. Comparative results against traditional multi-objective functions requiring extensive parameter tuning demonstrate that the proposed reward function directly produces the desired trajectory. This reward function provides a valuable reference for trajectory planning in second-order motion systems aiming to achieve minimised rise time and subsequent minimised integrated absolute error.

Electric vertical take-off and landing (eVTOL) unmanned aerial vehicles (UAVs) have limited flight time and range due to energy and power constraints in their power system components. This paper proposes a novel energy-aware trajectory generation method for rotary-wing UAVs powered by hydrogen that optimises fuel cell energy consumption by considering the dynamic behaviour of the propulsion system. The proposed method incorporates multi-physical effects of the fuel cell, electronic speed controller, motor, UAV dynamics and propeller. After validating the fuel cell propulsion model based on literature, the energy-aware trajectory optimisation problem is formulated and solved using the $ CasADi$ and $ OptimTraj$ frameworks. The study examines energy-optimal trajectory planning for two fuel cell-powered topologies: semi-active (Setup A) and direct (Setup B). The key characteristics of the fuel cell, voltage drop ($ {E}_{d}$) and voltage of the fuel cell capacitor voltage ($ {V}_{d}$), are considered to represent the dynamic behaviour of two different fuel cells during a mission starting at 40 m in the x direction and 20 m in the z direction. The simulation results show that Setup B has lower transient current fluctuations and a more stable power profile than Setup A, as evidenced by the decreased derivative of fuel cell current and hydrogen consumption rate. Energy-aware trajectory generation has been shown to reduce cumulative energy consumption by optimising vertical and forward velocity profiles, increasing the endurance of rotary-wing UAVs. To summarise, this study presents a novel methodology for energy-aware trajectory optimisation in fuel cell-powered UAVs by incorporating propulsion system dynamics into flight planning, which has been largely overlooked in the existing literature on hydrogen-powered rotary-wing UAVs.

The two-dimensional inlet, a critical component of scramjet engines, encounters substantial design challenges, particularly when operating over a wide-speed range. To accelerate the design process, a one-dimensional model-based methodology is proposed, encompassing three main stages: initial geometric design, boundary layer modification design and variable geometry design. Among these three stages, the boundary layer modification design ensures a flow coefficient of 1.0 while accounting for boundary layer effects under design point conditions. And the variable geometry design can enhance the startability of the inlet under off-design point conditions. Moreover, a newly proposed uniformisation model is applied at the inlet exit. The results indicate that the performance parameters calculated using the one-dimensional model are in good agreement with the numerical simulation results, although minor discrepancies exist for some parameters, thereby demonstrating the reliability of the design method. Additionally, the designed inlets exhibit superior performance and a lower self-starting Mach number. Overall, the one-dimensional model method enables the rapid design of two-dimensional inlets, offering a valuable tool for reducing the inlet design cycle.