Complex systems such as aircraft engines are continuously monitored by sensors. In predictive aircraft maintenance, the collected sensor measurements are used to estimate the health condition and the Remaining Useful Life (RUL) of such systems. However, a major challenge when developing prognostics is the limited number of run-to-failure data samples. This challenge could be overcome if multiple airlines would share their run-to-failure data samples such that sufficient learning can be achieved. Due to privacy concerns, however, airlines are reluctant to share their data in a centralized setting. In this paper, a collaborative federated learning framework is therefore developed instead. Here, several airlines cooperate to train a collective RUL prognostic machine learning model, without the need to centrally share their data. For this, a decentralized validation procedure is proposed to validate the prognostics model without sharing any data. Moreover, sensor data is often noisy and of low quality. This paper therefore proposes four novel methods to aggregate the parameters of the global prognostic model. These methods enhance the robustness of the FL framework against noisy data. The proposed framework is illustrated for training a collaborative RUL prognostic model for aircraft engines, using the N-CMAPSS dataset. Here, six airlines are considered, that collaborate in the FL framework to train a collective RUL prognostic model for their aircraft's engines. When comparing the proposed FL framework with the case where each airline independently develops their own prognostic model, the results show that FL leads to more accurate RUL prognostics for five out of the six airlines. Moreover, the novel robust aggregation methods render the FL framework robust to noisy data samples.

Taxiing aircraft using electric vehicles is seen as an effective solution to meet aviation targets of climate neutrality. However, making the transition to electric taxiing operations is expected to significantly increase the electricity demand at airports. In this paper we propose a mixed-integer linear program to schedule electric vehicles for aircraft towing and battery charging, while considering a limit for the supply of energy. The objective of the schedule is to maximize emissions savings. For computational tractability, we develop an Adaptive Large Neighbourhood Search which makes use of multiple local search heuristics to identify scheduling solutions. For daily scheduling with a small fleet size, the developed heuristic achieves solutions with an average 4% gap to the best linear programming solution. The results show that charging the vehicles during daytime is essential to maximize saved emissions: removing charging opportunities for a few hours during the day reduces the performance by an average of 6.4%. In addition, it is found that fast charging leads to low vehicle downtime, unless the battery size exceeds 750kWh, when charging rates over 150kW become unnecessary. Overall, our model provides support for infrastructure planning of airports during the transition to aircraft electric taxiing.

Most Remaining Useful Life (RUL) prognostics are obtained using supervised learning models trained with many labelled data samples (i.e., the true RUL is known). In aviation, however, aircraft systems are often preventively replaced before failure. There are thus very few labelled data samples available. We therefore propose a Long Short-Term Memory (LSTM) autoencoder with attention to develop health indicators for an aircraft system instead. This autoencoder is trained with unlabelled data samples (i.e., the true RUL is unknown). Since aircraft fly under various operating conditions (varying altitude, speed, etc.), these conditions are also integrated in the autoencoder. We show that the consideration of the operating conditions leads to robust health indicators and improves significantly the monotonicity, trendability and prognosability of these indicators. These health indicators are further used to predict the RUL of the aircraft system using a similarity-based matching approach. We illustrate our approach for turbofan engines. We show that the consideration of the operating conditions improves the monotonicity of the health indicators by 97%. Also, our approach leads to accurate RUL estimates with a Root Mean Square Error (RMSE) of 2.67 flights only. Moreover, a 19% reduction in the RMSE is obtained using our approach in comparison to existing supervised learning models.

A good weight initialization is crucial to accelerate the convergence of the weights in a neural network. However, training a neural network is still time-consuming, despite recent advances in weight initialization approaches. In this paper, we propose a mathematical framework for the weight initialization in the last layer of a neural network. We first derive analytically a tight constraint on the weights that accelerates the convergence of the weights during the back-propagation algorithm. We then use linear regression and Lagrange multipliers to analytically derive the optimal initial weights and initial bias of the last layer, that minimize the initial training loss given the derived tight constraint. We also show that the restrictive assumption of traditional weight initialization algorithms that the expected value of the weights is zero is redundant for our approach. We first apply our proposed weight initialization approach to a Convolutional Neural Network that predicts the Remaining Useful Life of aircraft engines. The initial training and validation loss are relatively small, the weights do not get stuck in a local optimum, and the convergence of the weights is accelerated. We compare our approach with several benchmark strategies. Compared to the best performing state-of-the-art initialization strategy (Kaiming initialization), our approach needs 34% less epochs to reach the same validation loss. We also apply our approach to ResNets for the CIFAR-100 dataset, combined with transfer learning. Here, the initial accuracy is already at least 53%. This gives a faster weight convergence and a higher test accuracy than the benchmark strategies.

Recent advances in battery technology have opened the possibility for short-haul electric flight. This is particularly attractive for commuter airlines that operate in remote regions such as archipelagos or Nordic fjords where the geography impedes other means of transportation. In this paper we address the question of how to optimize the charging infrastructure (charging power, spare batteries) for an airline when considering a battery swapping system. Our analysis considers the expenditures needed for (i) the significant charging power requirements, (ii) spare aircraft batteries, (iii) the used electricity, and (iv) delay costs, should the infrastructure not be sufficient to accommodate the flight schedule. The main result of this paper is the formulation of this problem as a two-phase recourse model. This is required to account for the variation of the flight schedule throughout a year of operations. With this, both the strategic (infrastructure sizing) and tactical (battery recharge scheduling) planning are addressed The model is applied for Widerøe Airlines, with a network of 7 hub airports and 36 regional airports in Norway. The results show that a total investment of 4412 kW in electricity power supply and 25 spare batteries is needed for the considered network, resulting in a daily investment of €11700. We also quantify the benefits of considering an entire year of operations for our analysis, instead of just one congested day (7% cost reduction) or one average day of operations (31% reduction) at the most congested airport.

Following the Paris Accords, the aviation industry aims to become climate neutral by 2050. In this line, electric vehicles that tow aircraft during taxiing are a promising emerging technology to reduce emissions at airports. This paper proposes an end-to-end optimization framework for electric towing vehicles (ETVs) dispatchment at large airports. We integrate the routing of the ETVs in the taxiway system where minimum separation distances are ensured at all times, with the assignment of these ETVs to aircraft towing tasks and scheduling ETV battery recharging. For ETV recharging, we consider a preemptive charging policy where the charging times depend on the residual state-of-charge of the battery. We illustrate our model for one day of operations at a large European airport. The results show that the 913 arriving and departing flights can be towed with 38 ETVs, with battery charging distributed throughout the day. The fleet size is shown to increase approximately linear with the number of flights in the schedule. We also propose a greedy dispatchment of the ETVs, which is shown to achieve an optimality gap of 6% with respect to the number of required vehicles and with 22% with respect to the maximum delay during towing. We also show that both algorithms can be leveraged to account for flight delays using a rolling horizon approach, and that over 95% of the flights can be reallocated if delays occur. Overall, we propose a roadmap for ETV management at large airports, considering realistic ETV specifications (battery capabilities, kinematic properties) and requirements for aircraft collision avoidance during towing.

The increasing availability of condition-monitoring data for components/systems has incentivized the development of data-driven Remaining Useful Life (RUL) prognostics in the past years. However, most studies focus on point RUL prognostics, with limited insights into the uncertainty associated with these estimates. This limits the applicability of such RUL prognostics to maintenance planning, which is per definition a stochastic problem. In this paper, we therefore develop probabilistic RUL prognostics using Convolutional Neural Networks. These prognostics are further integrated into maintenance planning, both for single and multiple components. We illustrate our approach for aircraft turbofan engines. The results show that the optimal replacement time for the engines is close to the lower bound of the 99% confidence interval of the RUL estimates. We also show that our proposed maintenance approach leads to a cost reduction of 53% compared to a traditional Time-based maintenance strategy. Moreover, compared with the ideal case when the true RUL is known in advance (perfect RUL prognostics), our approach leads to a limited number of failures. Overall, this paper proposes an end-to-end framework for data-driven predictive maintenance for multiple components, and showcases the potential benefits of data-driven predictive maintenance on cost and reliability.

Aircraft maintenance design aims to identify strategies that render the aircraft reliable for flight in a cost-efficient manner. These are often conflicting objectives. Moreover, existing studies on maintenance design often limit themselves to only one type of maintenance strategy, overlooking other potentially dominating designs. We propose a framework for aircraft maintenance design with explicit reliability and cost-efficiency objectives. We explore the design space of a variety of maintenance strategies ranging from traditional time-based maintenance to predictive maintenance. To explore this design space, we propose an adaptive algorithm using Gaussian process learning and a novel adaptive sampling method. Gaussian process learning models rapidly pre-evaluate new maintenance designs, while adaptive sampling selects for further exploration only those designs that are expected to improve the available Pareto front of maintenance designs. This framework is illustrated for the maintenance of multi-component aircraft systems with k-out-of-n redundancy. The results show that novel predictive maintenance designs based on Remaining-Useful-Life prognostics dominate other maintenance designs, especially in the knee region of the obtained Pareto front, where the most beneficial balance between conflicting objectives is achieved. Our proposed exploration algorithm also outperforms other state-of-the-art exploration algorithms with respect to the quality of the Pareto front obtained.

The increasing availability of condition monitoring data for aircraft components has incentivized the development of Remaining Useful Life (RUL) prognostics in the past years. However, only few studies consider the integration of such prognostics into maintenance planning. In this paper we propose a dynamic, predictive maintenance scheduling framework for a fleet of aircraft taking into account imperfect RUL prognostics. These prognostics are periodically updated. Based on the evolution of the prognostics over time, alarms are triggered. The scheduling of maintenance tasks is initiated only after these alarms are triggered. Alarms ensure that maintenance tasks are not rescheduled multiple times. A maintenance task is scheduled using a safety factor, to account for potential errors in the RUL prognostics and thus avoid component failures. We illustrate our approach for a fleet of 20 aircraft, each equipped with 2 turbofan engines. A Convolution Neural Network is proposed to obtain RUL prognostics. An integer linear program is used to schedule aircraft for maintenance. With our alarm-based maintenance framework, the costs with engine failures account for only 7.4% of the total maintenance costs. In general, we provide a roadmap to integrate imperfect RUL prognostics into the maintenance planning of a fleet of vehicles.

Reducing the length of departure queues at runway entry points is one of the most important requirements for reducing aircraft traffic congestion and fuel consumption at airports. This study designs an aircraft departure model at a runway using a time-varying fluid queue. The proposed model enables us to determine the aircraft waiting time in the departure queue and to evaluate effective control approaches for assigning suitable holds at gates rather than runway entry points. As a case study, this study modeled the departure queue at runway 05 of Tokyo International Airport for an entire day of operations. Using actual traffic data of departures at the airport, the model estimates that aircraft spend a total of 2.5 h departure waiting time in a day at runway 05. Considering the stochastic nature of actual departure traffic, the relevance of the proposed model is discussed using validation criteria. The model estimation shows a reasonable, expected order of magnitude compared with the departure queue recorded in the actual traffic data. Furthermore, ecological and economic benefits are quantitatively evaluated assuming a reduction in the departure queue length. Our results show that about one kiloton of fuel oil per year is wasted due to aircraft waiting to depart from a single departure runway.