Purpose High endurance fixed-wing UAVs powered by solar arrays or hydrogen fuel cells carry energy systems that cost more than the airframe itself and a single motor failure can end a multi-hour mission. This study aims to compare twin, trimotor and quad propulsion on measured data to find where the engine-out redundancy of each end. Design/methodology/approach Static thrust stand data for a 6.5 kg, 3.3 m airframe are transferred to flight with advance ratio dependent propeller coefficients. On a 1000 Wh budget, engine-out cases are scored against a two-level criterion of stopped propeller drag, trim drag and rudder authority. Findings All three configurations cruise at 18 m.s-1 on 211–233 W and hold that airspeed after a motor loss with the tested motor and propeller combinations. What separates them is directional control. The quad and trimotor need under 12° of rudder and the trimotor loses its centerline motor with no yaw. The twin needs 73% of the rudder travel and cannot be trimmed below 12.8 m.s-1, marginal against the 12 m.s-1 minimum sustainable flight level. Originality/value For the tested components on this airframe, the twin-engine redundancy assumption inherited from manned practice does not carry over: the fin, sized for cruise stability, sets the redundancy of the twin and a minimum control speed after a motor loss belongs in the design case of such a layout.
Purpose This study aims to address the critical challenges of defect sample scarcity and extreme long-tail distributions in automated aircraft skin inspection, which frequently lead to conventional deep learning models missing critical minority defects. Design/methodology/approach An application-oriented classification pipeline was constructed based on a ResNet-50 backbone pre-trained on the standard ImageNet data set. Class-balanced sampling and weighted cross-entropy loss were used to reduce prediction bias towards majority classes, whilst geometric and colour enhancement strategies were used to increase the diversity of training samples under limited data conditions. Dropout regularisation was incorporated into the task-specific classification head, followed by the use of gradient-weighted class activation mapping to visualise the image regions contributing to the classification decision. The combined effect of these complementary strategies was evaluated through comparison, ablation, visualisation and sensitivity tests under simulated image-noise perturbations. Findings Experimental results on a proprietary aircraft skin data set demonstrate that the model achieves 96.60% accuracy and 92.12% multiclass balanced accuracy. These results indicate improved classification performance under the current proprietary patch-level data set and experimental protocol; however, they do not establish generalisation to independent aircraft fleets, data acquisition systems or full-image inspection scenarios. Originality/value The contribution of this study does not lie in proposing a new backbone architecture or learning algorithm, but in the application-specific configuration and systematic evaluation of complementary class-balancing, data-augmentation, regularisation and interpretability strategies for limited and imbalanced aircraft skin defect data. Through controlled ablation, repeated-run evaluation and minority-class-oriented performance analysis, the study provides empirical evidence on how these established techniques jointly influence classification performance under the present aircraft skin inspection setting.
Purpose Repeated ADRC parameter optimization under wide flight-envelope conditions usually leads to high computational cost when fixed-wing UAVs operate across multiple flight states, while the adaptability of controller parameters between different operating conditions also becomes difficult to maintain. Design/methodology/approach To address this issue, a Kriging-assisted efficient global optimization method is developed for pitch-channel ADRC tuning of fixed-wing UAVs. Several representative operating conditions distributed along the flight envelope are selected to establish corresponding pitch-channel models and the optimized controller parameters are further reconstructed into a continuous gain-scheduling strategy. A nonlinear six-degree-of-freedom simulation model with actuator constraints is subsequently adopted to evaluate the full-trajectory control performance of the proposed method. Findings The obtained results show that the proposed approach reduces the mean number of closed-loop simulation evaluations by approximately 32.7% and 43.5% relative to GA and PSO, respectively, while decreasing the average computational time by 28.4% and 39.8%. Originality/value This work combines Kriging surrogate modeling with ADRC gain scheduling for full-trajectory UAV control and provides an efficient engineering-oriented solution for multi-condition controller optimization and nonlinear flight implementation.
Purpose This study aims to evaluate a maintenance risk-prioritization and barrier-mapping workflow for outdoor-parked general aviation aircraft. Design/methodology/approach Six failure scenarios were identified from 220 anonymized records covering 23 long-parking episodes and assessed by 12 qualified experts. Eleven consistent analytic hierarchy process (AHP) matrices yielded severity, occurrence and detection weights of 0.7572, 0.1224 and 0.1204. Median ratings were evaluated by fuzzy failure mode and effects analysis (FMEA) and mapped to Bow-tie task-card requirements. Robustness was tested across 27 weight combinations; record burden was reported as finding density over 9,225 aircraft-parking days, and rating dispersion and agreement were assessed. Findings Landing-gear/brake corrosion and engine-intake contamination ranked first and second. Three middle scenarios were borderline high because their classifications changed underweight or quartile perturbation; lower-fuselage water ingress remained medium. Conventional risk priority number produced the same ranking (Spearman rho = 1.00), so the integration did not improve ordinal discrimination. Average-measures intraclass correlation coefficient (ICC)(2,12) ranged from 0.954 to 0.973. Research limitations/implications The single-organization, Cessna 172/525, humid-inland case limits generalization. Because the retained index lacks aircraft and parking-episode identifiers, episode-level incidence and cluster-adjusted intervals cannot be estimated. The workflow requires local recalibration and prospective multi-site operational validation. Practical implications The workflow can support inspection and task-card planning after local recalibration; operational benefits require prospective validation. Originality/value The contribution is an auditable applied workflow linking established AHP-fuzzy FMEA methods to explicit Bow-tie-derived maintenance actions, not a new risk-analysis method.
Purpose Multi-phased mission system (MPMS) shows characteristics with multi-phase, uneven-length mission phase and different change trends per run, which bring new challenges on modelling and monitoring for the system. The purpose of this paper is to present a sequential local information increment-based condition monitoring method for the MPMS. Design/methodology/approach Firstly, the sequential local information increment is used to divide uneven-length mission phases. Local information gain indices are deduced through the local information increment, and then the information gain indices of the global mutations can be used to identify the transition point of the mission phase. Here, the sequential movement strategy reveals the order and real-time nature of the method. To ensure the accuracy of the division, a sliding window is used to specify the search range for the information gain index, and an adjusting strategy is used to determine the appropriate phase dividing parameter. Then, considering the uneven-length mission phase and the trends of change in mission phases may be different, similarity analysis is utilized to identify similar mission phases. Finally, several monitoring sub-models are developed for the divided mission phases based on principal component analysis, respectively. Findings Simulation for quick access recorder flight data illustrates that the proposed method can effectively reduce the false alarm rate and missed alarm rate. Originality/value This paper provides a sequential local information increment-based condition monitoring method for the MPMS.
