This study presents a systematic review of peer-reviewed academic literature to explore the current landscape of artificial intelligence (AI) applications in sustainable aviation operations. Using a qualitative content analysis approach, four main thematic domains were identified, encompassing emission and fuel efficiency, maintenance reliability, infrastructure sustainability and education- or policy-related applications. In addition to thematic synthesis, the study mapped the annual publication frequency, the AI methods employed and the aviation domains targeted. The results reveal an increasing interest in hybrid and deep learning models, such as long short-term memory (LSTM), convolutional neural networks (CNN) and attention-based architectures, particularly in-flight optimisation and delay prediction tasks. AI-based flight optimisation techniques, such as trajectory prediction and adaptive fuel management, contribute to reducing CO2 emissions through more efficient flight planning and operations. Moreover, predictive maintenance supported by AI-driven digital twin systems has gained prominence due to its potential to reduce downtime and increase safety. The discussion further addresses regulatory challenges, the importance of explainable AI and integration barriers within complex aviation ecosystems. Findings are derived from a focused corpus of 27 peer-reviewed studies, which, although limited in number, offer representative insights into current sectoral trends. This review makes a significant contribution to both academia and industry by offering a comprehensive framework that categorises AI applications and highlights future research directions. Key implications include the need for regulatory harmonisation, real-time decision-support tools, and interdisciplinary approaches that integrate AI with behavioural sciences and sustainability goals.

A transition control methodology using hybrid laminar flow control (HLFC) for drag reduction for civil transport aircraft is described. An HLFC concept with a single suction chamber with varying porosity can be integrated into the leading-edge wing structure, with the wing ice protection system and high lift system. The resultant structural system is less complex than the multi-chamber concept, and with weight saving.

A turbulent type of pressure distribution has been applied to a civil transport retrofit HLFC wing design resulted with a higher aerodynamic performance than a favourable gradient type of pressure distribution. A turbulent type of pressure distribution also has the potential of being able to retain the performance of a conventional aircraft in the case of suction system failure.

An analysis of wind tunnel transition data with linear stability analysis method is described. Detailed measurements of the tunnel environment and fluctuations within the boundary layer are required to provide a better understanding of the physics of transition and data for non-linear methods.

A wind tunnel testing capability developed for measuring the effect of small features on transition in natural laminar flow could be extended to HLFC to investigate the effect of suction on the disturbances due to various degree of surface imperfections. A high Reynolds number wind tunnel simulation technique using surface suction to thin the turbulent boundary layer has been demonstrated theoretically and experimentally for a turbulent flow wing design. A theoretical investigation has shown that this technique could be extended to an aircraft with a HLFC wing design.

Supersonic wind tunnels are an essential tool for high-speed aerodynamics research, supporting studies ranging from fundamental flow analysis to advancements in supersonic transport. Accurately predicting tunnel performance, however, requires precise mathematical modeling. Previous models have primarily focused on plenum pressure predictions, often assuming an adiabatic process and overlooking temperature dynamics. Temperature changes during a test affect velocity and Reynolds number, influencing experimental measurements and underscoring the need to improve temperature prediction capabilities. In this paper, we develop a new model introducing two key corrections: heat addition from the thermal mass of the wind tunnel and real gas effects, particularly the Joule–Thomson effect, allowing us to capture the critical influence of temperature. Additionally, we account for pressure losses within the piping system. Comparative analysis with experimental data shows that our model reduces temperature prediction errors to within 2%, a marked improvement over the base model’s 9–13% error range. Furthermore, pressure predictions are refined, yielding more accurate assessments of plenum, reservoir and valve inlet pressures. These findings underscore the model’s utility in enhancing control system development and its broader value in advancing experimental design and operational precision in supersonic wind tunnel research.

To reveal the influence laws of casing abradable coating wear on compressor aerodynamic performance, a numerical simulation study was conducted on the performance of Rotor 37 under high-speed scraping conditions with different hardness coating wear morphologies. The results show that compressor isentropic efficiency and outlet mass flow are sensitive to scraping-induced morphology changes. The medium-hardness coating causes the most significant performance degradation, with maximum reductions reaching 1.96% and 1.39%, respectively, while the pressure ratio shows little variation. Scraping grooves aggravate mixing losses between leading-edge (LE) leakage flow and mid-chord (MID) leakage flow. The tip leakage flow pushes suction-side separation vortices toward the main flow path. Under medium-hardness coating scraping conditions, the maximum entropy generation region affects up to 9.1% blade height and induces tip leakage vortex-shockwave interactions, resulting in substantial tip losses. The compressor aerodynamic performance is less affected by wear zone roughness, with the maximum isentropic efficiency reduction being only 0.31%. When wear zone roughness increases, near-wall turbulence fluctuations intensify and separation zones expand, causing flow structure changes in tip leakage paths and blade wake regions, which shifts the compressor aerodynamic characteristics toward lower flow rates. The study demonstrates that coating hardness alters leakage flow structures through wear morphology depth: medium-hardness coatings with the deepest wear grooves exhibit maximum performance deterioration, while high-hardness coatings show better wear resistance and performance maintenance.

It is well known that the amount of damage caused by lightning strikes to protected composite airframe structures depends on the paint characteristics, often applied on the surface of composite structures to protect from environmental effects and to personalise a product. In this work, physically based models of the mechanical loads induced by lightning strikes are employed in the generation of the mechanical overpressure fields due to a simulated lightning strike, while accounting for the paint thickness. These fields are then implemented into a three-dimensional finite element framework and combined with a damage model to predict the effect of paint thickness on the mechanical damage in composite structures subjected to this type of events. These models accurately predict the increase of damage extent with the increase of paint thickness, which is corroborated by experimental observations from industry and by the experimentally observed trends reported in literature.

Unmanned aerial vehicles (UAVs) tracking one moving target in an oilfield environment with dynamic obstacles has been studied. Firstly, the UAV tracking one target is modeled mathematically, including the oilfield environment, the UAV kinematic model, the moving target model and so on. Considering the oilfield environment with dynamic obstacles, the chance-constraint method is incorporated into the distributed model predictive control (DMPC) to realise the target tracking and avoid the obstacles in the environment. Secondly, to obtain high-quality control-input sequences, the improved salp swarm algorithm (ISSA) is designed as the solver for the chance-constraint DMPC, integrating the sine-tent-cosine chaotic mapping in the initialisation, the differential mutation operation in the leader-position updating and the nonlinear adaptive inertia weights in the follower-position updating. Finally, the simulation results show that the proposed method has great performance in tracking the moving target in an oilfield environment with dynamic obstacles.

This research is focused on the formulation, implementation and verification of novel transition control strategy for an originally designed tail-sitter VTOL hybrid UAV. A differential thrust transition vehicle was designed to achieve wide-envelope flight control by using four actuators. Attitude-tracking and altitude-hold controllers were developed to provide adequate control actuation for smooth transition between different flight modes. The control system was formulated for flight dynamic model of the aircraft, and implemented in MATLAB/Simulink environment for flight simulations. Tilt-angle-based feedback control was employed to calculate attitude errors. Altitude errors were obtained by direct comparison of commanded signal with the vehicle’s response. To maintain the desired altitude and attitude, required total thrust and differential thrust were calculated from the output of the altitude and attitude controllers, respectively. High-fidelity aerodynamic analysis was performed over the entire flight envelope to obtain the aerodynamic coefficients of the aircraft. Motor dynamics, propeller effects and airframe aerodynamics were incorporated in the flight dynamics model of the vehicle. Flight simulations verified effectiveness of the flight control system in tracking attitude and altitude-hold commands. The flight modes transition manoeuvre was successfully accomplished with the designed transition control system. The control technique for takeoff, hover and landing was successfully validated on manufactured prototype.

The privatisation of unmanned aerial vehicles (UAVS) used for situational awareness purposes to ensure their own situational awareness based on parameters gives direction to progress and provides a basic framework for future studies. In this context, a unique communication system architecture was proposed for obtaining state-of-the-art mini-UAV data and evaluations were carried out on the basis of data flow and parameters. Within the scope of the evaluation this study postulates a trailblazing approach as a means of optimising flight data pattern recognition by integrating Cross Industry Standard Process for Data Mining (CRISP-DM) and long short-term memory (LSTM)-based predictive depiction. By leveraging the structured framework of CRISP-DM and the sequential learning capabilities of LSTM, this research aims to enhance the accuracy of mini unmanned aerial vehicle systems (UAVS) flight scenario predictive reliability. This framework is applied to navigation phase, where the overall flight trajectory can be seen, and accurate forecasting and pattern recognition are critical for optimising operational efficiency. The findings have displayed the ability to perform high-accuracy predictions of flight parameters within a structured process. The experimental results demonstrate that key flight parameters can be predicted with near–comma-level numerical accuracy, indicating a high level of estimation precision. Through facilitating immediate data transfer and organised navigation phase evaluation, this research provides a methodical strategy for managing flight data, simultaneously contributing notably to UAV decision support systems and self-qualification engineering.

This study introduces a mission-centric design optimisation framework for unmanned aerial vehicles (UAVs) to enhance mission performance across diverse operational scenarios. The proposed framework integrates multidisciplinary design optimisation with a wargaming-based simulation environment and leverages deep neural network-based surrogate models to balance key performance metrics, such as aerodynamic efficiency, radar cross section, structural weight and payload capacity. By incorporating automated task assignment, path planning and a probabilistic combat model, the framework evaluates UAV configurations in multi-domain, multi-asset scenarios. The algorithm identifies optimal solutions that maximise mission success while managing trade-offs among survivability, lethality and cost. Simulation results illustrate the framework’s functionality through representative mission scenarios, highlighting how design variables can influence operational effectiveness relative to baseline configurations. Furthermore, the modular design approach enables rapid UAV reconfiguration for evolving mission needs, offering scalable and adaptable solutions. These findings highlight the importance of integrating mission simulation tools with advanced optimisation techniques to address challenges in dynamic, high-threat environments, providing a robust methodology for UAV and fleet design.

This study presents a systematic approach to optimising drone airframe geometry with a specific focus on enhancing flight dynamics for racing applications. The proposed method integrates an elitist multi-objective evolutionary algorithm with mathematical procedures of varying natures, encapsulated within an algorithmic system of a simulation platform. The framework combines implicit numerical methods with a custom implementation to seamlessly interact across multiple simulation environments. This approach ensures the mathematical model closely reflects real-world conditions, providing reliable optimisation outcomes. The preliminary findings suggest that optimised airframe geometries lead to notable advancements in thrust efficiency and overall performance. Symmetrical and non-symmetrical designs exhibit distinct benefits, highlighting the potential for significant improvements in trajectory precision and flight efficiency. These results emphasise the practical impact of shape design and dynamic optimisation in drone airframes, offering valuable insights for developing high-performance racing drones.

Maintenance procedures are critically important for preserving the structural integrity, maintaining the functionality and ensuring the operational safety of aircraft. Traditional inspection techniques used in aircraft are often costly, time-consuming and prone to human mistake. Today, the opportunities provided by digitalisation and automation in aircraft maintenance and inspection processes are paving the way for innovative approaches. In this context, the use of inspection systems supported by image processing technologies has the potential to bring about a significant transformation in aircraft maintenance. Visual inspection methods integrated with unmanned aerial vehicles (UAVs) enable the rapid, accurate and repeatable detection of defects such as corrosion and cracks on the external surfaces of aircraft. This study focuses on the automatic detection and classification of defects on the external surfaces of aircraft, based on tests and analyses carried out by artificial intelligence algorithms using high-resolution data. The model developed in this study was implemented in Python in the Google Colab environment and supported by AI algorithms trained on visual data. The main objective is to investigate the feasibility of UAV-based systems for aircraft visual inspection and to provide concrete evidence of their practical applicability. In this regard, the UAV platform selected for image acquisition is intended to comprehensively scan the target areas and capture images with sufficient resolution for processing by artificial intelligence algorithms. A review of the literature reveals that UAV- and AI-based integrated approaches have been explored in only a limited number of studies related to aircraft maintenance. In this context, the present study proposes a system that enables the rapid and accurate detection of structural defects such as corrosion and cracks on the external surfaces of aircraft.

Urban air mobility (UAM) is not a new concept, but it is anticipated that UAM operations will increase. With such an increase in UAM operations, South Africa should prepare itself to handle this phenomenon. There is a limited scope of UAM operations in South Africa, and developing a conceptual framework for UAM would assist in identifying the key factors pertinent to regulating UAM operations. A systematic literature review (SLR) methodology was used in developing a conceptual framework. The method entailed searching the literature, screening for inclusion and exclusion and using thematic analysis to analyse and synthesise the findings. The conceptual framework identified key factors such as legislation, air traffic management, infrastructure and security. The key factors were analysed within the South African context, taking into consideration the socio-economic factors that will impact UAM operations. It was determined that South Africa’s policy outlook supports increased UAM operations, but the current South African energy crisis will impact the use of electric aircraft.

Predictive maintenance in safety-critical systems like turbofan engines increasingly relies on machine learning (ML) models to estimate remaining useful life (RUL), but the ‘black box’ nature of these models hinders their adoption and trustworthiness. While traditional ex-ante prognostic metrics (e.g. monotonicity, trendability) are used to pre-screen sensor data, a systematic comparison against the post-hoc explanations of what a model actually learns is lacking. We explore the application of SHapley Additive exPlanations (SHAP) from explainable artificial intelligence (XAI) to investigate feature importance in engine failure prediction using the second dataset of the Commercial Modular Aero-Propulsion System Simulation (CMAPSS). The preprocessing pipeline includes z-score normalisation of sensor data and the calculation of a health index (HI) to quantify system degradation. A power-law fit is applied to the HI to capture the underlying trends of engine wear and failure progression. We use the normalisation data to calculate prognostic feature selection metrics: monotonicity, trendability and prognosability. Then, we train two machine learning models – random forest (RF) regressor and gradient boosting (GB) method – directly from the raw data to predict the RUL based on the actual sensor readings. The SHAP values generated for both models are analysed to identify the features with the most significant impact on RUL predictions. By comparing the SHAP value distributions across models and prognostic predictors, we highlight feature robustness and their relative influence on engine degradation and failure prediction. This work provides insights into the interpretability of machine learning models in prognostics and enhances the understanding of sensor contributions to engine health monitoring. The results demonstrate the effectiveness of SHAP in elucidating feature importance, supporting the development of more transparent and reliable prognostic systems.

This study presents a surrogate-model-assisted Quasi-Newton optimisation framework for simultaneously improving the aerodynamic performance and radar stealth characteristics of an unmanned aerial vehicle (UAV). High-fidelity computational fluid dynamics (CFD) and computational electromagnetics (CEM) simulations are integrated through surrogate models generated via a face-centred central composite design within a design of experiments framework. Quadratic polynomial response surface equations are constructed for key aerodynamic and radar cross-section (RCS) metrics, enabling analytical gradient evaluation. A gradient-based quasi-Newton method with Broyden–Fletcher–Goldfarb–Shanno Hessian updates is employed to minimise a scalarised objective function combining normalised maximum lift coefficient, overall RCS and frontal RCS. Constraints are imposed on the lift-to-drag ratio ($L/D \geq 10$$L/D\ge 10$) and static longitudinal stability (${C_{m0}} \geq 0$$C_{m0}\ge 0$). Analytical derivatives from the response surface equations (RSEs) eliminate the need for direct numerical differentiation of CFD/CEM outputs, reducing computational cost and eliminating simulation noise. An interior-point sequential quadratic programming strategy is used to ensure satisfaction of nonlinear constraints during the optimisation process. The optimised UAV design demonstrates a $12{\rm{\% }}$$12\mathrm{\%}$ increase in maximum lift coefficient and a $30{\rm{\% }}$$30\mathrm{\%}$ reduction in both overall and frontal RCS compared to the baseline configuration. The results are confirmed through high-fidelity CFD and RCS simulations and are further validated experimentally in an anechoic chamber, with close agreement across all measured frequencies. The proposed methodology provides an efficient and experimentally verified approach for integrated aerodynamic and stealth optimisation in UAV design.

Aerodynamic modeling of tiltrotor aircraft is inherently limited by incomplete aerofoil data for blades constructed from multiple aerofoil sections. This data deficiency leads fundamental errors into the blade element theory (BET) inputs, resulting in inaccurate induced velocity and unsteady load predictions that compromise flight control design. To overcome this limitation, a corrective induced inflow factor k is introduced into an augmented Pitt–Peters dynamic inflow model. For fixed tilt angles, k is efficiently identified via an optimisation framework. The fmincon algorithm attains the optimal k with a computational speed 5–6 times faster than the genetic algorithm, while maintaining thrust coefficient errors within 1.26%. For continuous tilting, a direction-aware, time-varying quadratic model for k(t) is proposed, which uses the Fréchet distance metric to accurately reconstruct the aerodynamic hysteresis loop and eliminate spurious load predictions during rotation reversal. Validated through wind tunnel tests, the framework reduces the maximum thrust prediction error in dynamic tilting from 34.10 to 10.70%. The corrected aerodynamic model supports real-time simulation and provides a reliable foundation for advanced tiltrotor control system design.