Discovering and mitigating potential risks in advance is essential for preventing aviation accidents on routine flights. Although anomaly detection-based explanation techniques have successfully uncovered potential risks for proactive flight safety management, explaining group-scale precursors using these methods is challenging due to the assumption that risky flights are significantly fewer in number than normal flights, as well as the reliance on non-domain knowledge for hyperparameter adjustment. To characterize the group-scale precursors more accurately, we propose a novel technique called Catalyst Mass-Based Clustering Analysis (CMCA), which employs a composite entropy-energy dissipation index during approach to evaluate the energy management performance. On this basis, an optimization objective is constructed to identify clusters exhibiting significant energy management differences during the approach phase. We successfully identify group-scale precursors with energy management issues by applying CMCA to a combination of minority-labeled and majority-unlabeled flights. Comparative experiments show that these precursors have energy levels that deviate from normal flights by 5.83% and 10.93%, respectively, 1000 ft above touchdown, demonstrating the effectiveness of our method. The analysis suggests that poor energy management awareness on the part of pilots could be responsible for these group-scale precursors. Notably, the results obtained using CMCA are comprehensible for Subject Matter Experts, making the method a valuable tool for proactive flight safety management.

High-value aviation rotable parts require accurate demand forecasting because each failure simultaneously creates a replacement demand, a repair workload, and a potential aircraft on-ground (AOG) risk. This study develops a continuous-time Markov chain (CTMC) queuing framework for forecasting demand and optimizing target stock levels under realistic maintenance, repair, and overhaul (MRO) capacity constraints. Historical service-engineering records from B737 mechanical engine control (MEC) units are first benchmarked and filtered to retain inherent random failures during the useful-life phase. A Kolmogorov-Smirnov goodness-of-fit test is then used to verify the exponential time-between-failure assumption required by the Markov model. Based on this validated failure-pattern classification, the study derives steady-state probability models for both an ideal $M/M/\infty$ repair system and a finite-capacity $M/M/c$ repair system. The main contribution is not the isolated use of Weibull or exponential reliability models, Markov chains, or inventory optimization, which are established methods, but their auditable integration into an airline rotable-parts workflow that links failure-pattern screening, finite repair capacity, service-level constraints, and engineering validation. The empirical results show that repair bottlenecks shift probability mass toward higher numbers of failed units, create a fat-tailed backlog distribution, and double the required target stock from 6 to 12 MEC units under a 95% service-level requirement. Sensitivity experiments further show that repair turnaround time, fleet size, and MRO channel capacity jointly determine the inventory-capacity cost trade-off. The proposed framework provides an interpretable decision tool for airlines to align spare-part procurement with actual repair-system performance.

The positioning of civil aviation aircraft relative to a geographic reference point on Earth in a Cartesian frame is significant to detect the deviations from the desired path, especially for high-altitude airports or special airports based on performance-based navigation (PBN). To obtain these critical deviations during aircraft approach and landing, it is fundamental to estimate the continuous flight variables and discrete flight modes simultaneously with enough accuracy. With the coordinate conversion between the North, East, and Down (NED) frame and the geographic coordinate system based on World Geodetic System 1984 (WGS-84) considered, this study proposed a non-linear stochastic hybrid estimation algorithm with adaptive square-root unscented particle filtering (ASR-UPF) to estimate the true path. The probabilities of mode transition, represented by the normal cumulative density function of continuous states, determine whether to proceed with mode transitions. In addition, the adaptive update characterized by tracking variable noise and the importance sampling distributions based on the results of square-root unscented Kalman filtering (SR-UKF), as a comparative study of continuous system filtering, were used. The experiments illustrated the ASR-UPF is able to reduce the state estimation error more effectively, and more promptly track the error caused by incorrect mode estimation with adaptability compared to the SR-UKF. A further test with real flight data indicates that the proposed method gives the refined estimation of position and azimuth in NED frame.

Low-altitude micro downbursts pose a severe threat to aviation safety, yet conventional control approaches and standard deep reinforcement learning (DRL) often fail due to explicit modeling difficulties and sparse reward constraints. To address these challenges, this study proposes an explainable, data-driven framework integrating active-reward proximal policy optimization (AR-PPO). A bilevel optimization architecture driven by meta-gradients is developed to dynamically discover optimal reward functions without human intervention. Furthermore, a policy distillation pipeline utilizing wavelet-multivariate singular spectrum analysis (W-MSSA) and classification and regression trees (CART) is proposed to translate high-frequency continuous neural outputs into discrete, pilot-readable rules. Simulation results on a B737-800 model demonstrate that AR-PPO effectively overcomes the “stall trap” by autonomously learning to trade altitude for airspeed, outperforming static-reward baselines and empirical human pilots in extreme, zero-shot windshear encounters (22.0 m/s downdraft). Ultimately, the proposed framework successfully distills black-box AI strategies into verifiable, physics-informed standard operating procedures (SOPs), providing a highly transparent and robust solution for autonomous windshear escape and future competency-based flight training.

Threshold-based flight data monitoring (FDM) identifies parameter exceedances but provides limited visibility into gradual operational deviation within nominal operating bounds. This study presents a three-layer framework for continuous operational deviation quantification from quick access recorder (QAR) data. A standard operating procedure (SOP)-constrained Gaussian mixture model–hidden Markov model (GMM-HMM) infers 27 procedural states from flight parameters, aligning latent state representation with operational procedure. From these state sequences, maximum entropy inverse reinforcement learning (IRL) recovers an expert value function that estimates execution quality at each time step. Split conformal prediction calibrates detection thresholds with distribution-free coverage guarantees, while counterfactual analysis traces identified deviations to their procedural origin. Learned from real flight data, the recovered reward function assigns 1.6 times greater weight to lateral and speed stability than to vertical stability. Split conformal calibration yields an empirical false alarm rate of 4.2% under a nominal 5% significance level. Counterfactual analysis identifies vertical speed management as the dominant source of procedural deviation. These results support the proposed framework as a candidate continuous safety performance indicator that surfaces gradual operational deviation that threshold-based exceedance monitoring would not flag.