Abnormal Flight Research Articles

Flight Safety Risk Prediction for Civil Aircraft Approach and Landing

The approach and landing phase is the critical juncture in flight operations, where flight-state risks must be meticulously identified and mitigated to avert significant safety incidents. This paper proposes a risk prediction model that enhances the Light Gradient Boosting Machine (LightGBM) using the Whale Optimization Algorithm (WOA). First, a six-degree-of-freedom model is used to establish a system of differential equations to describe the aircraft’s flight states. In conjunction with longitudinal and lateral motion risk analyses, parameters prone to typical flight accidents are selected as risk prediction indicators, focusing on abnormal flight states. Subsequently, the LightGBM algorithm is employed to construct a flight safety risk prediction model using WOA to optimize hyperparameters within the LightGBM framework, enhancing the predictive accuracy of the model. Finally, a data set is created using actual flight data from a particular airline, aggregating flight-state parameter data from 910 flights for model training and testing. The proposed model’s performance is compared with similar machine-learning methods, including LightGBM, XGBoost, and Random Forest. The results demonstrate that the WOA–LightGBM model achieves 95.6% accuracy in predicting flight safety risks during the approach and landing phase, demonstrating a significant advantage in predictive precision over other methods.

UAV Anomaly Detection Method Based on Convolutional Autoencoder and Support Vector Data Description with 0/1 Soft-Margin Loss

Unmanned aerial vehicles (UAVs) are becoming more widely used in various industries, raising growing concerns about their safety and reliability. The flight data of UAVs can directly reflect their flight health status; however, the rarity of abnormal flight data and the spatiotemporal characteristics of these data represent a significant challenge for constructing accurate and reliable anomaly detectors. To address this, this study proposes an anomaly detection framework that fully considers the temporal correlations and distribution characteristics of flight data. This framework first combines a one-dimensional convolutional neural network (1DCNN) with an autoencoder (AE) to establish a feature extraction model. This model leverages the feature extraction capabilities of the 1DCNN and the reconstruction capabilities of the AE to thoroughly extract the spatiotemporal features from UAV flight data. Then, to address the challenge of adaptive anomaly detection thresholds, this research proposes a nonlinear model of support vector data description (SVDD) utilizing a 0/1 soft-margin loss, referred to as L0/1-SVDD. This model replaces the traditional hinge loss function in SVDD with a 0/1 loss function, with the goal of enhancing the accuracy and robustness of anomaly detection. Since the 0/1 loss function is a bounded, non-convex, and non-continuous function, this paper proposes the Bregman ADMM algorithm to solve the L0/1-SVDD. Finally, the difference between the reconstructed and the actual value is employed to train the L0/1-SVDD, resulting in a hypersphere classifier that is capable of detecting UAV anomaly data. The experimental results using real flight data show that, compared with methods such as AE, LSTM, and LSTM-AE, the proposed method exhibits superior performance across five evaluation metrics.

Research and application of differential optical absorption two-dimensional detection system for rotorcraft unmanned aerial vehicle

In order to meet the technical requirements for miniaturization, multi-angle, multi-altitude, and fast simultaneous acquisition of atmospheric pollutants, this study develops an integrated, lightweight, and cost-effective airborne differential optical absorption spectroscopy (DOAS) system. This system is designed in order to be used on a rotorcraft unmanned aerial vehicle (UAV) platform for monitoring atmospheric pollutants. The compositions of the hexacopter UAV platform and the airborne DOAS system are detailed in this work. The system includes a multi axis differential optical absorption spectroscopy (MAX-DOAS) spectral acquisition system, a control system, and a flight environment monitoring system. Commands are sent from a computer via serial communication to drive a gimbal, controlling the azimuth angle and elevation angle of the telescope, with a camera recording the light obstruction. The sunlight scattered by the atmosphere is collected by the telescope and transmitted via fiber optics to the spectrometer, which then transmits the data to the control computer. Additionally, the system captures data of altitude, temperature, humidity, and GPS location during flight, and filters out spectral data obtained under abnormal flight conditions. Stability studies indicate that the mean angular deviations for yaw, roll, and pitch are 0.07°, –0.13°, and –0.12° respectively, which meet the requirements for monitoring stability. Comparative experiments with a commercial ground-based DOAS system show that the correlation coefficients between the monitoring data of both systems are both greater than 0.92, confirming the reliability of the airborne system. In field flight experiments, the airborne DOAS system conducts observations at altitudes of 30 m, 60 m, and 90 m, with the elevation angle set at 0° and the azimuth angle measured every 30° from 0° to 360°. The system successfully obtains the concentration distributions of NO<sub>2</sub>, SO<sub>2</sub>, and HCHO at different azimuth angles and altitudes. The results indicate that the concentrations of these three gases decrease with altitude increasing, with higher concentrations observed in the southeast direction, indicating the presence of pollution sources in that direction. Further analysis with considering altitude changes indicates that the rate of decrease in NO<sub>2</sub> concentration and SO<sub>2</sub> concentration slow down with altitude increasing, while the rate of decrease in HCHO remains relatively constant. These findings indicate that this system effectively meets the technical requirements for simultaneous, rapid, multi-angle, and multi-altitude detection of atmospheric pollutants, providing essential support for the detailed monitoring of complex urban micro-environments.

Unstable Approach Detection and Analysis Based on Energy Management and a Deep Neural Network

The study of managing risk in aviation is the key to improving flight safety. Compared to the other flight operation phases, the approach and landing phases are more critical and dangerous. This study aims to detect and analyze unstable approaches in Taiwan through historical flight data. In addition to weather factors such as low visibility and crosswinds, human factors also account for a large part of the risk. From the accidents studied in the stochastic report of the Flight Safety Foundation, nearly 70% of the accidents occurred during the approach and landing phases, which were caused by improper control of aircraft energy. Since the information of the flight data recorder (FDR) is regarded as the airline’s confidential information, this study calculates the aircraft’s energy-related metrics and investigates the influence of non-weather-related factors on unstable approaches through a publicly available source, automatic dependent surveillance-broadcast (ADS-B) flight data. To evaluate the influence of weather- and non-weather-related factors, the outliers of each group classified by weather labels are detected and eliminated from the analysis by applying hierarchical density-based spatial clustering of applications with noise (HDBSCAN), which is utilized for detecting abnormal flights that are spatial anomalies. The deep learning method was adopted to detect and predict unstable arrival flights landing at Taipei Songshan Airport. The accuracy of the prediction for the normalized total energy and trajectory deviation of all flights is 85.15% and 82.11%, respectively. The results show that in different kinds of weather conditions, or not considering the weather, the models have similar good performance. The input features were analyzed after the model was obtained, and the flights detected as abnormal are discussed.

Flexible skin for measurement of boundary layer state and flight attitude identification on UAV

The wall shear stress and pressure are important for analyzing boundary layer flow and evaluating the aerodynamic performance of the aircraft. In this study, a flexible skin consisting of dual layer hot-film sensors and pressure belts was developed to measure the distribution of wall shear stress and pressure on an unmanned aerial vehicle during increase of angle of attack (AOA), tail spin, and decreases of AOA. The sensitivity of the dual layer hot-film sensor is improved by about 150% due to heat conduction reduced. The relative error of pressure belt is less than 1% at 10 °C–65 °C. The boundary layer separation and reattachment time, separation AOA and time-shift of the flow field changes in flight conditions are determined. The separated AOA of the left and right wing boundary layers are 26.41° and 17.58° respectively. There is a delay about 4 s between the separation of the boundary layer and the entry of the tail spin, which can provide early warning to prevent abnormal flight conditions such as post stall and tail spin.

Abnormal Flight Passenger Recovery Algorithm Based on Itinerary Acceptance

Abnormal Flight Passenger Recovery Algorithm Based on Itinerary Acceptance

Retracted: Analysis of Abnormal Flight and Controllers Data Based on DBSCAN Method

Retracted: Analysis of Abnormal Flight and Controllers Data Based on DBSCAN Method

Ghostbusters: Hunting abnormal flights in Europe during COVID-19

The impact of the COVID-19 pandemic is unprecedented in airline history, with irregular flight bans, the inability for accurate demand estimation, several turns in the epidemiological evolution, and a wide range of downstream effects on all aviation stakeholders. While most airlines have increasingly entered a recovery stage, compared to the utmost disruption around April 2020, the airline business is far from back-to-normality. Throughout the past two years, recurrent statements have been made regarding the existence of so-called ghost flights, where airlines operate nearly empty aircraft on markets with insufficient demand, partially with the aim to avoid losing precious airport slots. This study investigates the extent of such abnormal market service during the COVID-19 pandemic through an explorative, data-driven analysis, based on actual load factor data of European airlines for the years 2017 to 2021. We break down the observed deviations by airlines, markets, and airports. We find that low-cost carriers are most-likely to have performed abnormal flights during the pandemic; and that abnormal flights have mostly occurred on frequently-served markets. In addition, we show that airline responses, in terms of departure and yield changes, are largely heterogeneous across the 24 airlines in this study. Our study is the first one to shed light on the important issue of load factor deviations, and we hope that our findings can contribute to a better understanding of the existence of abnormal flights during the pandemic, as well as deriving appropriate policies for dealing with the ubiquitous threat and impact of ghost flights in the future.

Fitts’ law on the flight deck: evaluating touchscreens for aircraft tasks in actual flight scenarios

This research investigated the effects of an abnormal flight environment using touch-based navigation displays (TNDs). Fitts’ law was used to compare the performance of TNDs with control display units (CDUs) and mode control panel (MCPs) under three different flight scenarios (normal, turbulence and startled). A within-subjects design involving 15 male participants was used. Data were collected in respect to accuracy, movement time, subjective feelings, choices and comments. The results showed that under abnormal conditions, TNDs showed worse operation performance and stability than CDUs and MCPs; however, it was easy to learn from TNDs, and they provided a good user experience. Moreover, this research demonstrated the application of Fitts’ law to describe pilot behaviours in interactive flight devices, particularly for tasks involving real flight operations. TND designs for aviation could be developed based on these findings to improve flight crew performance when using new technology. Practitioner summary: This research built a Fitts’ law model to evaluate the performance of aircraft cockpit touchscreens under normal, turbulence and startled scenarios. We compared the different touchscreens (TNDs) with other traditional interactive devices, such as CDUs and MCPs. The results have implications for the design of aircraft cockpit touchscreens and define the task scenario. Furthermore, the results contribute to the development of scenes utilising Fitts’ law.

Analysis of Abnormal Flight and Controllers Data Based on DBSCAN Method

In order to quickly and timely analyze the airborne and controllers data of domestic ARJ21 (Advanced Regional Jet for 21st Century) aircraft and find out the existing flight problems and potential safety hazards, this paper proposes a DBSCAN (density-based spatial clustering of applications with noise) clustering analysis method for aircraft airborne and controllers data outlier detection to evaluate and monitor the flight status. The flight QAR (quick access recorder) data stored in the aircraft rapid data recorder are input to the DBSCAN clustering analysis algorithm to detect the flight parameters that are different from the normal range. Compared with the traditional airborne data analysis method, this method can realize the real-time analysis and prediction of data, improve the efficiency of data analysis, and find the potential security risks according to the analysis results and deal with them in time. In this paper, 1,102 ARJ21 aircraft operation data are used to test. The results show that the DBSCAN clustering anomaly data detection method based on density is fast and accurate in detecting the continuous parameters recorded in the flight process, and the display results are easy to analyze, which can predict the potential safety problems in time. The outliers detected by this method can provide support for the controller to detect the outliers and related flight risks in daily flights.

Ghostbusters: Hunting abnormal Flights in Europe during COVID-19

Ghostbusters: Hunting abnormal Flights in Europe during COVID-19

Digital Forensic Methodology for Detection of Abnormal Flight of Drones

When a drone accident has occurred, it is difficult to decide whether it is due to a crime, malfunction, mistake, or external force. Although the cause of the accident is elucidated through analysis of artifacts or flight data, there are many limitations. In this study, we present a method for detecting an abnormal flight using the motor current values and controller direction values of a drone. The experimental result revealed that, in the case of a normal flight, the current values of four motors were similar in hovering state and the current value of rear motors were increased when the drone was flying forwards. In the case of an abnormal flight, when the drone moved rightwards due to external force in hovering state, the current values of the two motors on the right side were increased greatly. After a period of time following the movement to the right side, the current values of all the motors converged to 0. In the future, motor current values and controller direction values may be used to determine whether an abnormal flight in a drone accident has occurred because of external force by wind, birds, persons, or the like.

Modeling multi-loop intelligent pilot control behavior for aircraft-pilot couplings analysis

Aircraft-pilot couplings (APCs) are one of the key factors that affect the flight safety of the modern aircraft. Among the several types of APCs, the nonlinear pilot induced oscillations caused by aircraft flight control system anomalies are extremely severe. In this paper, a multi-loop intelligent pilot model is presented for the analysis of the nonlinear pilot induced oscillations investigations. The multi-loop control tasks are introduced to define a typical multi-loop pilot modeling problem. The model is capable of describing the pilot's behavior during a multi-loop control task and analyzing the nonlinear pilot-induced oscillations when the flight control system is changed in different forms. Fuzzy control logic is designed to reflect the human pilot's behavior of making judgments. To be specific, the human pilot can change his/her control behavior adaptively when the aircraft flight control system falls into anomalies. Six cases of the abnormal flight control system are considered, and the control effects of the human pilot model are analyzed. The experimental results demonstrate that the human pilot can adjust the control strategy in different cases, which reflects the human pilot's adaptation to the changes of aircraft dynamics characteristics. Furthermore, it indicates that the human pilot model can obtain good control performance in tracking the command signal and can complete the task well facing the failures of the control system. Besides, the validity of the human pilot model is assessed by the scalogram-based PIO metric. The proposed pilot model can be applied in evaluating the aircraft loss-of-control events in the form of adverse aircraft-pilot couplings.

Flight Situation Recognition Under Different Weather Conditions

The weather conditions affect the approach and departure routings of aircraft. The detection of flight status of aircraft faces two major issues under adverse weather conditions. They are how to select an approximate and effective feature combination from different flight parameters and how to identify the status of an aircraft via a reasonable flight parameter combination. This article presents a solution to the flight status recognition problem of an aircraft. The time-domain and wavelet-domain features are extracted from a complete flight dataset and a typical flight dataset, respectively. ARMA coefficients entropy is also extracted to represent dynamic behavior of flight data. A new deep sparse learning network with an optimized Gaussian process classifier is proposed to detect the aircraft status. Experiments are executed via a practical flight dataset under different weather. The feasibility of the selection scheme of flight parameters and the proposed deep learning method to detect abnormal flight situation are verified by competitive experimental results, which presents widespread conclusions on feature selection methods and abnormality flight status detection.

Development of interaction model for air traffic controllers and pilots using discrete mathematics methods

The purpose of the article is to create a database of errors and to develop an algorithm for a situational decision-making model taking into account availability of potential errors of air traffic controllers and pilots. Air traffic controllers and pilots typical errors were compiled and analyzed, arrays of specialists errors were created, binary error relations based on methods of discrete mathematics were also compiled in this article. This decision is caused by the need to formalize the interaction of specialists, since each error of the air traffic controller can be compared with a certain set of pilot errors and vice versa. In case of further in-depth analysis, it is possible to expand the database by adding additional errors arrays of the adjacent point controller, aerodrome service, planning service, etc. The goal is formed after analyzing the features of simulator training in higher educational institutions. The peculiarity is the absence of hazardous factors during the simulator training. This training takes place according to the ideal model. Undoubtedly, this approach is aimed at developing the correct algorithm of actions in normal or abnormal flight conditions, but thus the trainee can’t work out the decision-making skills if there is an error in the ideal algorithm. At the same time, existing specialists face unintended errors every working day, so having experience in this field plays an important role in minimizing the impact of the human factor on flight safety. In our case, it is proposed to include such a dangerous factor as an unintentional error in the joint training program for air traffic controllers and pilots, which will improve the training quality of specialists.

Helicopter Cockpit Audio Data Analysis to Infer Flight State Information

In recent years, the National Transportation Safety Board has highlighted the importance of analyzing flight data as one of the effective methods to improve the safety and efficiency of helicopter operations. Since cockpit audio data contain various sounds from engines, alarms, crew conversations, and other sources within a cockpit, analyzing cockpit audio data can help identify the causes of incidents and accidents. Among the various types of the sounds in cockpit audio data, this paper focuses on cockpit alarm and engine sounds as an object of analysis. This paper proposes cockpit audio analysis algorithms, which can detect types and occurrence times of alarm sounds for an abnormal flight and estimate engine-related flight parameters such as an engine torque. This is achieved by the following: for alarm sound analysis, finding the highest correlation with the short time Fourier transform, and the Cumulative Sum Control Chart (CUSUM) using a database of the characteristic features of the alarm; and for engine sound analysis, using data mining and statistical modeling techniques to identify specific frequencies associated with engine operations. The proposed algorithm is successfully applied to a set of simulated audio data, which were generated by the X-plane flight simulator, and real audio data, which were recorded by GoPro cameras in Sikorsky S-76 helicopters to demonstrate its desired performance.

Flight Recovery Method of Regional Multiairport Based on Risk Control Model

In the regional multiairport system, the contradiction between the limited operating resources and the large flight flow is serious, and the flight delays can easily lead to the occurrence of unsafe events. This paper investigates the abnormal flight recovery method in regional multiairport system based on risk control. The focus is to reschedule arrival-departure flights in real time with minimized delay time and risk probability. In this study, the risk about terminal area control and scene operation was considered in the analysis of the risk control model (RCM), which includes six key risk points: airspace control, flight conflict, ground service, apron support, ground control, and taxiing conflict. The mathematical model on flight recovery was constructed to solve minimized delay time and risk probability with MSINS (multistart algorithm with intelligent neighborhood selection). The data of a typical regional multiairport system in China were selected for experimental verification in order to compare the RCM with the traditional recovery model (TRM). The experimental results show that first, there are some hidden dangers in the traditional recovery methods of flight delay. Flight conflict and apron support are the risk points that need to be controlled most in the multiairport system. Secondly, for the effective solution with the shortest delay time, the RCM can reduce the overall operation risk of the system, but the flight delay time is a little longer. For the effective solution with the lowest risk probability, RCM can reduce the risk of system operation and the delay time of flights at the same time. Therefore, RCM can improve the security level of the system during abnormal flight recovery and ensure or even improve the recovery efficiency.

Anomaly Location Method for QAR Data Based on Principal Component Analysis Hierarchical Clustering

The existing QAR (Quick Access Recorder) data anomaly detection algorithm can detect abnormal flights, but it cannot effectively locate the abnormal QAR data of abnormal flights. In order to address the issue, QAR data of abnormal flights are smoothed and pre-processed based on k-medoids algorithm. Pre-processed QAR data are transformed into a one-dimensional time series represented by angle cosine, and a subsequence feature matrix is generated from the angle cosine sequence by sliding window mechanism. Then, the dimension of the matrix is reduced based on principal component analysis, and the top-down hierarchical clustering is performed on row vectors of the reduced matrix according to the amount of information within each column. The anomaly nodes are detected according to the number of vectors contained in the clustering tree nodes. The abnormal data segments of QAR data of abnormal flights are located by the positions of vectors generated in the cosine angle sequence in the abnormal nodes. The experimental results on real flight data sets show that the proposed method can locate not only the known exceedance events, but also the abnormal fragments besides the monitoring items.

Clustering With Outlier Removal

Cluster analysis and outlier detection are two continuously rising topics in data mining area, which in fact connect to each other deeply. Cluster structure is vulnerable to outliers; inversely, outliers are the points belonging to none of any clusters. Unfortunately, most existing studies do not notice the coupled relationship between these two tasks and handle them separately. In this article, we consider the joint cluster analysis and outlier detection problem, and propose the Clustering with Outlier Removal (COR) algorithm. Specifically, the original space is transformed into a binary space via generating basic partitions. We employ Holoentropy to measure the compactness of each cluster without involving several outlier candidates. To provide a neat and efficient solution, an auxiliary binary matrix is introduced so that COR completely and efficiently solves the challenging problem via a unified K-means— with theoretical supports. Extensive experimental results on numerous data sets in various domains demonstrate the effectiveness and efficiency of COR significantly over state-of-the-art methods in terms of cluster validity and outlier detection. Some key factors including the basic partition number and generation strategy in COR with an application on abnormal flight trajectory detection are further analyzed for practical use.

Responding to an Unexpected In-Flight Event: Physiological Arousal, Information Processing, and Performance.

The study was designed to investigate whether a simulated unexpected abnormal flight event can lead to startle and explore differences in behavioral responses between expected and unexpected abnormal flight events. Recent research suggests startle (an autonomic response to an acute stimulus) following unexpected abnormal flight events can impact pilot performance and can increase the probability of a negative outcome following the event. Information processing, physiological measures, and performance differences between responses to expected and unexpected flight events were compared. General aviation (GA) pilots flew a series of flights in a fixed-base flight simulator including two experimental flights which included an unexpected and an expected, engine failure. During the flights, heart rate, eye tracking, and flight data were recorded. During the unexpected engine failure, pilots showed greater increases in heart rate and pupil dilation. Significant differences in scanning were evident with fewer areas scanned following the unexpected event. During the unexpected engine failure, performance was impaired when compared to the expected events. However, poor performance was not associated with higher levels of arousal. The study provides an empirical demonstration of impaired pilot response to unexpected events with associated symptoms consistent with the induction of startle. The present research builds on Landman et al.'s conceptual model of startle and surprise. Standardized training protocols may not adequately prepare pilots to deal with the unexpected effects of startle in real-world encounters. Introducing greater variety into training events may be useful. The effects of startle in disrupting well-trained responses may also occur in areas other than aviation where critical events may occur unexpectedly or present in an unfamiliar manner.