132571 research outputs found
Sort by
Condition-Based Maintenance scheduling of an aircraft fleet under partial observability: A Deep Reinforcement Learning approach
No full textIn the Condition-Based Maintenance (CBM) context, the definition of optimal maintenance plans for an aircraft fleet depends on an efficient integration of : (i) the probabilistic predictions of the health condition of the components and (ii) the stochastic arrival of the corrective maintenance tasks, together with consideration of the preventive maintenance tasks as defined in the Maintenance Planning Document (MPD). To this end, in this paper, we present a two-stage dynamic scheduling framework to solve the aircraft fleet maintenance scheduling problem under a CBM strategy in a disruptive environment. In the first stage of the framework, we address the uncertainty in the predicted health state of the monitored components by planning the optimal maintenance policy based upon the belief state-space of the health of the components. The decision-making process is formulated as a Partially Observable Markov Decision Process (POMDP) and is solved using the Partially Observable Monte Carlo Planning (POMCP) algorithm, considering the aircraft maintenance scheduling problem requirements. In the second stage, a Deep Q-Network (DQN) is developed, that integrates the defined maintenance policy of the monitored components within the scheduling of the aircraft fleet's preventive and corrective maintenance tasks. Our model, through a rolling horizon approach, continuously creates and adjusts the maintenance schedule, reacting to new updated task information, where the availability of maintenance resources constraints the execution of each task. The proposed framework was tested on a case study from a large airline and the performance was evaluated against the current state practice of the airline. The results show that our model can schedule 96.4% of monitored components on-time. As a consequence of this, a 46.2% maintenance cost reduction is achieved for the considered monitored components relative to a corrective maintenance approach.Air Transport & Operation
L<sup>p</sup>-estimates for the square root of elliptic systems with mixed boundary conditions II
No full textWe show Lp-estimates for square roots of second order complex elliptic systems L in divergence form on open sets in Rd subject to mixed boundary conditions. The underlying set is supposed to be locally uniform near the Neumann boundary part, and the Dirichlet boundary part is Ahlfors–David regular. The lower endpoint for the interval where such estimates are available is characterized by p-boundedness properties of the semigroup generated by −L, and the upper endpoint by extrapolation properties of the Lax–Milgram isomorphism. Also, we show that the extrapolation range is relatively open in (1,∞).Analysi
Finlet rails for the reduction of the trailing-edge noise
No full textThe present study focuses on the application of finlet rails as a passive technique of flow control to mitigate trailing-edge noise. Finlet rails are small cylinders whose axes are aligned along the streamwise direction, transversally positioned with respect to the trailing edge. In the first part of this study, the effects of finlet geometry on the aeroacoustic emission of a NACA 633−018 airfoil are investigated using an array of microphones. It is observed that reducing the transversal spacing of finlet rails leads to increasing the maximum noise reduction, found to be of 4 decibels at relatively low frequencies. An optimum for the height of the finlets was determined, equivalent to 1.6δ∗, where δ∗ is the displacement thickness of the boundary layer. With the aim of unveiling the underlying physical mechanism for finlet rails, PIV at high spatial resolution is applied around the surface treatment. It is found that the turbulence energy is lifted-up and moved away from the scattering edge, which attenuates the wall-pressure fluctuations. The observed attenuation of the wall-pressure fluctuations occurs at the energy-containing scales, which is an important difference with finlet fences. In the region underneath the finlet rails, the transversal size of the energetic structures diminishes when the surface treatment is applied. The combination of the lift-up of the turbulence structures, that reduces the wall-pressure fluctuations, with the smaller turbulence scales is responsible for the noise reduction observed for finlet rails.Wind Energ
Fault detection for LTI systems using data-driven dissipativity analysis
No full textMotivated by the physical exchange of energy and its dissipation in electro-mechanical systems, we propose a new fault detection method based on data-driven dissipativity analysis. We first identify a dissipativity inequality using one or multiple shots of data obtained from a linear time-invariant system. This dissipativity inequality's storage and supply rate functions assume generic quadratic difference forms encompassing all LTI systems. By analysing the norm of the identified dissipative inequality as the residual function, we can detect the occurrence of faults in real-time without the need to model each fault the system is subjected to. Through academic examples, we demonstrate how we can identify supply rate and storage functions from persistently exciting data shots. We present a practical example of detecting faults on a two-degree-of-freedom planar manipulator with zero missed fault detection rate, which is compared to a standard PCA-based fault detection algorithm.Team Tamas Keviczk
Compensating torque ripples in a coarse pointing mechanism for free-space optical communication: A Gaussian process repetitive control approach
No full textActuators that require commutation algorithms, such as the switched reluctance motor (SRM) considered in this paper and employed in the coarse pointing assembly (CPA) for free-space optical communication, often have torque-ripple disturbances that are periodic in the commutation-angle domain that deteriorate the positioning performance. The aim of this paper is to model the torque ripple as a Gaussian Process (GP) in the commutation-angle domain and consequently compensate for it at arbitrary velocity. The approach employs repetitive control (RC) at a constant velocity. A spatial GP with a periodic kernel is trained using data that is obtained from the RC step resulting in a static non-linear function for compensation at arbitrary velocity. Stability conditions are provided for both steps. The approach is successfully applied to a CPA prototype to improve the tracking performance for laser communication, where the torque ripple is compensated at arbitrary velocity.Team Jan-Willem van Wingerde
A directional coupler for partial discharge measurements in gas-insulated substations
No full textThis paper introduces a directional coupler for partial discharge (PD) measurements in gas-insulated substations (GIS). The sensor comprises a combination of magnetic and electric couplers, effectively segregating forward and backward pulses to enhance PD charge estimation and defect location. The sensor's design was supported with finite element method simulations and measurements conducted in a transverse electromagnetic test bench. Comparative analyses were performed against independent magnetic and electric couplers. The charge estimation and the directional coupler's directivity were evaluated in both the test bench and a full-scale GIS with different PD defects. Initially, the combined magnetic and the electric couplers exhibited undesired interactions, prompting corrective measures. Subsequent adjustments included changes to the electric coupler material and modifications to the magnetic coupler construction. The resulting high-voltage directional coupler performed better than the separated couplers in a GIS with discontinuities. This partial discharge sensor emerges as a candidate for future SF6-free alternative GIS.High Voltage Technology Grou
Quantifying and modelling the effect of external and internal vegetation water dynamics on radar data
No full textAgriculture plays a critical role in the economy and environment worldwide, and the provision of real-time, reliable information on large-scale agricultural activity is essential for precision agriculture and global economic prosperity. In this context, remote sensing, especially through Synthetic Aperture Radar (SAR), can play an important role by offering accurate estimation of crop biophysical parameters such as Leaf Area Index (LAI), crop height, dry biomass, and Vegetation Water Content (VWC). Unlike traditional high-resolution optical imagery, which is often undermined by cloud cover, SAR microwave remote sensing overcomes these limitations by generating and transmitting longer wavelengths (300MHz – 10 GHz) that penetrate clouds and aerosols and allowing data acquisition both day and night. However, SAR data is influenced by various factors such as sensor characteristics such as frequency, polarization, and incidence angle, as well as target characteristics like the size and shape distribution of crop constituents, and more importantly, the water content of crop constituents. In the path towards precision farming and more sustainable and efficient farming using SAR data, understanding the role of these factors, particularly the dynamics of external and internal vegetation water content on radar backscatter, is vital.To date, however, the potentially confounding effects of both internal and, particularly, external water dynamics in vegetation on radar backscatter have not been adequately addressed. Existing studies have indeed illustrated the effects of SCW on radar backscatter, but the degree to which it influences different frequencies and polarizations, and the subsequent impact on crop bio-geophysical parameters remains unclear. Therefore, the main goal of this thesis is to expand our knowledge of the relationship between radar backscatter, vegetation dynamics, and surface canopy water (SCW) in agricultural monitoring. In this thesis we utilized statistical analysis and radiative transfer modeling in combination with fully polarimetric L-band data from a truck-mounted scatterometer and C-band data from Sentinel-1, along with extensive field data… Mathematical Geodesy and Positionin
Limiting the Performance Loss of an RTG Container Terminal with a Highly Utilized Storage Yard
No full textThis research addresses the critical role of the Yard Utilization Rate (YUR) in the performance of container terminals, focusing on Rubber-Tyred Gantry Crane (RTG) container terminals. Existing literature reveals a gap in studies specifically targeting high YURs in RTG container terminals. The primary research question is, ”How to reduce Quay Crane (QC) productivity loss in RTG container terminals with highly utilized storage yards?” A Root Cause Analysis (RCA) identifies key factors affecting the loading and unloading processes of QCs, with a focus on the root causes "Too few TTs deployed", "Congestion", and "YUR too high". Design alternatives are proposed to address these root causes, and three design alternatives are tested through simulation: varying the number of Terminal Trucks (TTs), sacrificing driving lanes for additional storage space, and implementing an alternative shuffle policy with a different bay distribution. Results show a significant influence of the number of TTs on QC productivity, with an optimum at 8 TTs per QC. Sacrificing driving lanes for extra storage space proves counterproductive due to increased congestion. However, the alternative shuffle policy Multi-bay with Sets Shuffle Policy with a Clustered Bay Distribution shows promise, offering a potential solution to mitigate QC productivity loss in highly utilized RTG container terminals, with an average increase of 0.4 boxes per hour. In conclusion, the study successfully addresses the identified knowledge gap, providing valuable insights and proposing a practical solution for reducing QC productivity loss in RTG terminals with highly utilized storage yards. The suggested approach of Clustered Bay Distribution with a Multi-bay with Sets Shuffle Policy proves effective in maintaining terminal performance during periods of an increased YUR.Marine Technology | Transport Engineering and Logistic
An Experimental Assessment of the Stability of Graph Contrastive Learning
No full textThe Deep Neural Network (DNN) has become a widely popular machine learning architecture thanks to its ability to learn complex behaviors from data. Standard learning strategies for DNNs however rely on the availability of large, labeled datasets. Self-Supervised Learning (SSL) is a style of learning that allows models to also use unlabeled data for training, which is typically much more abundant.SSL is being applied many different data domains such as images and natural language. One such a domain is the domain of graph data. A graph is a data structure describing a network of nodes connected by edges. Graphs are a natural way of presenting many forms of data such as molecules, social networks, and 3D meshes.The style of SSL that has found the most success on graphs is Contrastive Learning (CL). In CL, an encoder is trained to produce semantically rich representations from unlabeled input data by smartly separating task-relevant information in the input from task-irrelevant information. The encoder backbone most commonly used for Graph Contrastive Learning (GCL) is the Graph Convolutional Neural Network (GCNN).While GCNNs are the state of the art on many graph data tasks, they suffer from underfitting when made too deep. This is especially a problem for GCL as it prevents encoder complexity to scale with the large availability of unlabeled data.In this thesis, we investigate this underfitting behavior through the lens of GCNN stability. Stability refers to a model's ability to continue producing consistent outputs, even when its inputs are perturbed slightly. Theoretical work has shown that stability guarantees for GCNNs weaken when their complexity is increased. We confirm experimentally that, in many cases, GCNNs indeed grow less stable when made more complex. This a relevant finding given that learning stable representations is a prerequisite to CL. Additionally, we show in our experiments that, even when trained using CL, stability discrepancies between different GCNN architectures do not disappear. This, in turn, suggests that GCNN architectures with poorer stability may also produce poorer representations. We confirm experimentally that, on at least one dataset, poor stability as a result of architectural complexity can indeed be correlated to a degradation in representation quality. With this result we provide an additional explanation as to why deeper GCNNs are often found to perform worse in GCL settings. These insights can, in turn, motivate the design of model architectures for GCL that do not suffer from this trade-off between complexity and representation quality.Computer Scienc
Infrared temperature measurements on fast moving targets: A novel calibration approach
No full textIn this study, an infrared system is developed for accurate measurements of surface temperature and heat transfer on fast moving targets. The system was designed for the Oxford Turbine Research Facility, a world-leading experimental facility delivering highly engine representative, scalable heat transfer results for aerospace research. Infrared thermography is employed to acquire temperature maps of high-pressure turbine blades, allowing assessment of surface thermal conditions including heat transfer coefficient, adiabatic wall temperature, Nusselt number, cooling effectiveness, and metal effectiveness. Achieving accurate infrared thermography measurements in rotating turbomachinery experimental conditions is arduous due to reflections from the surroundings, low emissivity of metallic parts, and motion blur resulting from high speed. To overcome these challenges, calibration procedures were developed against a traceable standard using a bespoke steady experimental facility. A method to determine the reflected temperature from surroundings was also validated. Correction for all measurement disturbances is demonstrated to within the accuracy of the primary measurement thermocouple. Finally, the developed calibration method was validated on a fast-moving rotating geometry demonstrating accurate correction for all measurement disturbances, without the need for an in situ calibration. A detailed uncertainty analysis for each calibration step is also presented.Fluid Mechanic
Data sources:
Access Repository Dashboard
Do you manage Open Research Online? Become a CORE Member to access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard! 👇