Big data analytics with deep learning approach have attracted increasing attention in transportation engineering, involving operations, maintenance, and safety. In commercial aviation sectors, operational, and maintenance data produced on modern aircraft is increasing exponentially, and predictive analysis of these data is an exciting and promising field in aviation maintenance, which has a potential to revolutionize aerospace maintenance industry. This study illustrates the state-of-the-art applications of deep learning in big data analytics for predictive maintenance and a real-world case study for commercial aircraft. A Long Short-Term Memory network based Auto-Encoders (LSTM-AE) is proposed for complex aircraft system fault detection and classification, which makes use of the raw time-series data from heterogeneous sensors. The proposed method uses nominal time-series samples corresponding to healthy behavior of the system to learn a reconstruction model based on LSTM-AE framework. Then the system health index (HI) and fault feature vectors are derived from the reconstruction error matrix for fault detection and classification. The proposed method is demonstrated on a real-world data set from a commercial aircraft fleet. The typical PCV faults as well as the 390 F sensor and 450 F sensor faults due to sense line air leakage are successfully detected and distinguished based on the extracted features. The case study results show that the computed HI can effectively characterize the health state of the aircraft system and different fault types can be identified with high confidence, which is helpful for line fault troubleshooting.

Anomaly detection and failure prediction of gas turbines is of great importance for ensuring reliable operation. This work presents a novel approach for anomaly detection based on a data-driven performance digital twin of gas turbine engines. The developed digital twin consists of two parts: uncertain performance digital twin (UPDT) and fault detection capability. UPDT is a probabilistic digital representation of the expected performance behavior of real-world gas turbine engines operating under various conditions. Fault detection capability is developed based on detecting UPDT outputs that have low probability under the training distribution. A novel anomaly measure based on the first Wasserstein distance is proposed to characterize the entire flight data, and a threshold can be applied to this measure to detect anomaly flights. The proposed UPDT with uncertainty quantification is trained with the sensor data from an individual physical reality and the outcome of the UPDT is intended to deliver the health assessment and fault detection results to support operation and maintenance decision-making. The proposed method is demonstrated on a real-world dataset from a typical type of commercial turbofan engine and the result shows that the F1 score reaches a maximum of 0.99 with a threshold of 0.45. The case study demonstrated that the proposed novel anomaly detection method can effectively identify the abnormal samples, and it is also possible to isolate anomalous behavior in a single performance signal, which is helpful for further fault diagnosis once an anomaly is detected.

Modern commercial aircraft are usually configured with aircraft condition monitoring system to collect the operating data of subsystems and components, which can be used for airborne system health monitoring and predictive maintenance. This paper presents a baseline model based aircraft auxiliary power unit performance assessment and remaining useful life prediction method using aircraft condition monitoring system reports data, which can facilitate a cost-effective management of auxiliary power units of aircraft fleet. Firstly, the performance baseline model for auxiliary power unit is established using random forest method. Then a health index characterizing the performance degradation of in-service auxiliary power units is obtained based on the performance baseline model. Finally, the performance degradation trend is predicted using Bayesian dynamic linear model. To improve the prediction accuracy, four performance baseline models are established from the data of auxiliary power units under different operating conditions, among which an optimal model is determined. This data-driven baseline model can be used to quantify the performance degradation of auxiliary power units in service, and can be further used to evaluate the remaining useful life of auxiliary power unit using a Bayesian dynamic model. The developed approach is applied on a real data set from 22 auxiliary power units of a commercial aircraft fleet. The results show that the computed health index can effectively characterize the auxiliary power units performance degradation and the remaining useful life relative prediction errors are less than 4% when auxiliary power unit enters the rapid degradation stage. This would allow operators to accurately assess the performance degradation for the auxiliary power units and further proactively plan future maintenance events based on remaining useful life prediction.

This paper addresses the issues of fault diagnosis of the environmental control system of a certain commercial aircraft model of which the environmental control system has a high failure rate in field and causes many unplanned maintenance events. Because of the complexity and reciprocal compensation mechanism of the environmental control system, it is difficult to carry out fault isolation timely once the failure occurred during aircraft turnaround time, which thus may cause flight delay or even cancelation. The original contribution of this work is to propose a Bayesian network–based fault diagnosis method for commercial aircraft environmental control system where a multi-information fusion mechanism is used to incorporate the system first principle, expert experience and condition monitoring data. It incorporates extraction technology of sensor feature parameters and the structural learning of Bayesian network to realize the effective diagnosis of multiple faults. A case study is conducted based on a data set from a commercial aircraft fleet. The results show that the fault isolation ratio of this method is greater than 89%. The proposed Bayesian fault diagnosis network method can be used as a troubleshooting tool for airline maintenance technicians in fault isolation of environmental control system, reducing the time spent on-line troubleshooting and aircraft downtime.

The aerospace industry is striving to reduce the aircraft operating costs while maintaining required safety level. Emerging technologies such as the structural health monitoring to reduce long-term maintenance cost and increase aircraft availability are promoted by the manufacturers. To successfully integrate the structural health monitoring technology into the current maintenance process of modern commercial aviation, a clear definition of the structural-health-monitoring-based maintenance operational concept and the system level requirements is required. This article proposed a structural health monitoring operational concept and the associated maintenance cost modeling and risk assessment methods for the implementation of the structural health monitoring in commercial aviation industry. The developed methodology provides a tool to determine the optimal scheduled structural health monitoring inspection interval and repair decision thresholds for approved scheduled structural health monitoring task. A simulated case study is carried out to demonstrate the structural health monitoring operational concept and how an optimal maintenance strategy can be determined using the proposed methodology. Preliminary results show that the integration of the structural health monitoring into the existing maintenance process can reduce the maintenance cost compared to that of the current practice using the traditional Non-Destructive Evaluation (NDE) techniques while maintaining the risk below an acceptable level.

It is almost impossible to detect the health status of the aircraft hydraulic system via a single variable, because of the complexity and the coupling relationship between components of the system. To serve the purpose, a novel anomaly detection method considering multivariate monitoring data is proposed in this article. The unsupervised auto-encoder model with the long short-term memory layers is used to reconstruct multivariate time series data, and a new comprehensive decision-making index based on two conventional ones is proposed to measure the difference between the observation and the reconstruction. Then, the health threshold of the decision-making index can be calculated by the kernel density estimation. The flight data are divided into several samples, and the anomaly detection of each sample is determined by the specific rule. The healthy status of each flight is determined by voting based on the detection results of all samples included in the flight. The performance of the proposed method is validated on the real continuous monitoring data, and the results confirm that the proposed model overcomes the problems of multistage and multivariate parameters in the anomaly detection of the aircraft system and improves the detection efficiency.