179566 research outputs found
Sort by
Two-stage transfer learning for airborne multi-spectral image classifiers
No full textIn this work, we propose a novel training paradigm designed to support transfer learning for more effective classification in multispectral airborne imagery. Current state-of-the-art approaches typically rely on either leveraging solely RGB (red-green-blue) pretraining or applying in-domain transfer learning for multispectral imagery classification. Instead, our approach constructs and trains two separate neural network models (backbones): one specifically for wavelengths with available pretrained data (like visible bands) and another trained from scratch on all-bands available in the dataset. These models are then integrated with a fully-connected layer or multi-layered perceptron, which is trained on the features from both networks. This allows us to exploit the significant benefits of generalizable features learned from RGB datasets and the information provided by the full spectrum of multispectral bands. We employ the BigEarthNet and EuroSAT datasets, encompassing Sentinel-2 satellite imagery in the visual and infrared bands. This approach yields considerable performance gains in comparison with other training strategies across every evaluation metric we utilized for these datasets. The results are also consistent across a variety of backbone architectures, underlining the efficacy of our transfer learning technique in the analysis of multispectral data
Origins and fate of polycyclic aromatic hydrocarbons (PAHs) in sustainable drainage systems (SuDS) in a Scottish urban area: Implications for groundwater systems
No full textIncreasing urbanisation and the effects of climate change have resulted in a decrease in water quality and availability worldwide. At the same time, flooding in urban areas has become one of the most prevalent natural disasters worldwide. Sustainable drainage systems (SuDS) are resilient stormwater management solutions that can help reduce flooding by mimicking natural drainage processes and promoting infiltration. Numerous studies have focused on the benefits SuDS provide. But, to date, studies fail to investigate the risks that detention basins pose to groundwater quality, particularly the potential for infiltration of stormwater pollutants such as polycyclic aromatic hydrocarbons (PAHs). To address this important knowledge gap, this study combines gas chromatography–mass spectrometry (GC–MS) techniques for PAHs characterisation with numerical modelling tools to investigate organic pollutant infiltration patterns. We find that high levels of PAHs originating from urban areas are temporarily stored in SuDS basins and are likely to reach shallow water tables within one year. Multiple factors such as vegetation, precipitation, drainage area size, total organic carbon content in the soil and soil saturation influence the PAHs infiltration rates within the basins. More broadly, this study highlights the need for more research regarding SuDS dynamics to prevent both flooding and groundwater deterioration
Behaviour and design of sheathed light gauge steel panels subjected to in-plane shear and gravity loads
No full textThe in-plane shear behaviour of sheathed light gauge steel wall panels is investigated by tests on 2.4 m square wall panels together with push-out tests, to determine the shear fixing stiffness and resistance. Parameters such as sheathing material, screw geometry, stud arrangement, adhesives, combined bracing with sheathing boards, and profiled sheeting were considered. An X-braced panel was used as a benchmark and to interpret the forces in the bracing obtained from measured strains. Wall panels with single C- and back-to-back C- sections in the middle of the panel have equivalent shear stiffness and resistance. The use of non-winged fixings can significantly enhance the stiffness of the panel due to increased thread engagement with the board material. Adhesives on the board-frame interface reduce damage evolution on the board material, while significantly increasing both stiffness and resistance. The tests also included the combination of X-bracing and sheathing boards to determine how their stiffnesses may be combined. It was also shown that profiled steel sheets serve as a potential alternative to X-bracing, although they require more fixings. The design approach based on elastic theory for the fixing properties obtained from push-out tests is in close agreement with the test results obtained from the representative wall panel tests, where the ratio of the recorded to the predicted panel shear stiffness and resistance have an average of 1.03 and 1.05, respectively, with a variation of up to 8 %.</p
Enhancing bolted joint performance of woven composite laminates using 3D printed interlayers with tailored fibre architectures
No full textThis study investigates the effect of incorporating 3D printed interlayers containing continuous carbon fibres into plain weave CFRP laminates. The impact on stress distribution and the mechanical performance of bolted joints is systematically investigated. Three interlayer design strategies were developed to tailor the fibre distribution within the interlayers using filament-based 3D printing, and the resulting tailored-interlayer/woven laminates were assessed through double-shear testing to characterise the fibre load-transfer mechanisms. A filament-level multiscale finite element model was developed to capture the progressive damage evolution of the laminates. The experimental and numerical results demonstrate that incorporating 3D-printed interlayers can substantially enhance joint performance. Relative to the woven laminate baseline, enhancements were achieved across all interlayer cases. Specifically, improvements of up to 86 % in stiffness, 95 % in initial peak strength, and 59 % in ultimate bearing strength were achieved across the evaluated cases. In addition, substantial enhancements in energy absorption capacity were observed, with the initial fracture energy increasing by as much as 496 %, and the ultimate fracture energy by up to 10 %, depending on the specific architectural conditions. Among the designs, fibre steering guided by failure planes yielded most suppression of damage propagation. Together with micro-CT scans of the final failure morphologies, the simulation results provided insight into the damage progression and showed good agreement with the overall mechanical response observed experimentally. This research highlights the effectiveness of stress-adapted fibre steering in laminates and demonstrates the potential of 3D printing as a tool for locally reinforcing CFRP joints
UNHRD’s humanitarian support in South Asia via multistage stochastic programming
No full textOne of the main tasks of the United Nations Humanitarian Response Depot (UNHRD) relies on allocating relief aid to save people who suffer from disasters. This task is particularly challenging in areas like South Asia, where relief aid efforts are confronted with complex transportation conditions, significant socioeconomic disparities, and the frequent occurrence of disasters, not to mention that financial resources are often scarce. In this paper, we develop a novel Multistage Stochastic Programming model to help UNHRD support critical decisions regarding site selection and relief aid allocation. Differently from the main literature, where these decisions are often made within a two-stage paradigm, our three-stage perspective takes into account in-kind donation campaigns that are triggered depending on the disaster impact and its effects, and is paramount to improving the effectiveness and fairness of the disaster relief operation. Our objective function maximizes the effectiveness of the disaster relief operation, defined as the extent to which it fulfills the needs of the population. Considering that different regions often exhibit distinct coping capacities, the effectiveness measure also factors in a vulnerability score to encourage relief aid allocation to the most in-need populations. The overall results show the importance of in-kind donation to achieve a more equitable relief aid allocation plan and the benefit of targeting more vulnerable regions under severely scarce resources
Thermal response of hybrid steel-timber floor cross-sections exposed to standard fire: experimental and numerical investigations
No full textHybrid systems that combine steel beams with cross-laminated timber (CLT) floor slabs can be vulnerable to fire, given the combustible nature of timber. Specifically, when unprotected, heat from a fire can conduct through steel beams to the CLT panels, which in turn may experience loss of mechanical properties, and possible charring (and combustion) in the connection zone between the CLT panel and the steel beam. Accordingly, this paper aims to establish thermal profiles in hybrid steel-timber floor cross-sections exposed to fire through experimental and numerical investigations. Results from fire tests and numerical validation studies on hybrid cross-sections exposed to a standard fire are presented; a total of six experiments with unprotected, partially protected, and fully protected steel sections were conducted following an ISO 834-1 standard fire exposure. Furthermore, a two-dimensional numerical heat transfer model was developed using SAFIR software, to predict the evolution of temperatures in the hybrid cross-section. The results confirm that passive fire protection of the steelwork using intumescent coatings plays a key role in determining the extent of charring in the connection region between the CLT panel and the steel section. In addition, temperature predictions from the developed numerical model show reasonable agreement with the experimental measurements
Joint Modelling of Line and Point Data on Metric Graphs
No full textMetric graphs are useful tools for describing spatial domains like road and river networks, where spatial dependence act along the network. We take advantage of recent developments for such Gaussian Random Fields (GRFs), and consider joint spatial modelling of observations with different spatial supports. Motivated by an application to traffic state modelling in Trondheim, Norway, we consider line-referenced data, which can be described by an integral of the GRF along a line segment on the metric graph, and point-referenced data. Through a simulation study inspired by the application, we investigate the number of replicates that are needed to estimate parameters and to predict unobserved locations. The former is assessed using bias and variability, and the latter is assessed through root mean square error (RMSE), continuous rank probability scores (CRPSs), and coverage. Joint modelling is contrasted with a simplified approach that treat line-referenced observations as point-referenced observations. The results suggest joint modelling leads to strong improvements. The application to Trondheim, Norway, combines point-referenced induction loop data and line-referenced public transportation data. To ensure positive speeds, we use a non-linear link function, which requires integrals of non-linear combinations of the linear predictor. This is made computationally feasible by a combination of the R packages inlabru and MetricGraph, and new code for processing geographical line data to work with existing graph representations and fmesher methods for dealing with line support in inlabru on objects from MetricGraph. We fit the model to two datasets where we expect different spatial dependency and compare the results
High Content Drug Screening & Analysis Dataset: A comprehensive pharmacological survey across heterogeneous patient-derived GBM stem cell models
No full text384w microplate format of high throughput, high content drug screening data from image based readouts. Screening was conducted against 6 glioma stem cell lines using multple drug libraries using approved and pre-clinical compounds. The dataset includes quantified image analysis from CellProfiler pipeline outputs: Raw data and processed data (feature re-scaling/transformation, removal of redundant features, normalisation to dmso), including metadata and related plate maps/informatio
Blurred boundaries:Community treatment orders as instruments of racial surveillance
Get PDFThe Mental Health Act 1983 (MHA) authorizes the compulsory detention and treatment of people with mental disorders who are perceived to pose a risk to themselves or to others. Since its enactment, there have been concerns that the coercive powers of the Act have been disproportionately used for Black people with mental disorders. This disproportionate impact of the MHA on Black people is most clearly seen in the excessive use of Community Treatment Orders (CTOs). Although there is limited evidence on the effectiveness of CTOs, they continue to be used increasingly in the care of Black people, with latest data showing that Black people are over seven times more likely to be issued CTOs than White people (National Health Service Digital, 2024). This is particularly concerning as CTOs are commonly perceived as intrusive and have been described as a form of racial surveillance. In this paper, we explore the idea of CTOs as instruments of racial surveillance and argue that their continued use in the care of Black people with mental disorders is an extension of the intrusive powers of the State and might constitute a form of racial injustice.</p
Effect of Deliberate Physical Damage and Repair on the Icephobicity of Oil‐Infused Elastomer Coatings
No full textOil-infused elastomers have shown promise as anti-icing coatings, but their softness has led to durability concerns. The effect of damage is investigated to study the resilience of the icephobicity of silicone elastomer coatings with and without oil infusion. Physical damage is applied to specimens by abrading with grit paper or cutting with a scalpel. Surface characterization reveals morphological changes in the coatings due to damage and de-icing on the surfaces, along with changes in the static water contact angle. Abrasion of the surfaces does not overwhelmingly or universally worsen the ice adhesion strength. Some damage even lowers the ice adhesion strength, possibly due to Cassie–Baxter wetting. However, it decreases the average freezing time. Cutting causes accelerated deterioration to ice adhesion strength and worsens freezing time. Though damage to the oil-infused coatings is greater, changes in icephobicity are similar to coatings without oil infusion. Re-coating is an effective method of repairing even severely damaged surfaces and recovering icephobicity. We show oil-infused elastomers have durable icephobicity and are effective anti-icing coatings
Access Repository Dashboard
Do you manage Edinburgh Research Explorer? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!