Multiscale Elements Research Articles

Improved YOLOv5 Network for Detection of Peach Blossom Quantity

In agricultural production, rapid and accurate detection of peach blossom bloom plays a crucial role in yield prediction, and is the foundation for automatic thinning. The currently available manual operation-based detection and counting methods are extremely time-consuming and labor-intensive, and are prone to human error. In response to the above issues, this paper proposes a natural environment peach blossom detection model based on the YOLOv5 model. First, a cascaded network is used to add an output layer specifically for small target detection on the basis of the original three output layers. Second, a combined context extraction module (CAM) and feature refinement module (FSM) are added. Finally, the network clusters and statistically analyzes the range of multi-scale channel elements using the K-means++ algorithm, obtaining candidate box sizes that are suitable for the dataset. A novel bounding box regression loss function (SIoU) is used to fuse the directional information between the real box and the predicted box to improve detection accuracy. The experimental results show that, compared with the original YOLOv5s model, our model has correspondingly improved AP values for identifying three different peach blossom shapes, namely, bud, flower, and falling flower, by 7.8%, 10.1%, and 3.4%, respectively, while the final mAP value for peach blossom recognition increases by 7.1%. Good results are achieved in the detection of peach blossom flowering volume. The proposed model provides an effective method for obtaining more intuitive and accurate data sources during the process of peach yield prediction, and lays a theoretical foundation for the development of thinning robots.

Smart Manipulation of Complex Optical Elements via Contact-adaptive Dry Adhesives.

The implementation of complex, high-precision optical devices or systems, which have vital applications in the aerospace, medical, and military fields, requires the ability to reliably manipulate and assemble optical elements. However, this is a challenging task as these optical elements require contamination-free and damage-free manipulation and come in a variety of sizes and shapes. Here, a smart, contact-adaptive adhesive based on magnetic actuation is developed to address this challenge. Specifically, the surface bio-inspired adhesives made of fluororubber facilitate contamination-free and damage-free adhesion. The stiffness modulation of packaged magnetorheological grease based on the magnetorheological effect endows the smart adhesive with a high conformability to the optical elements in the soft state, a high grip force in the stiff state, and the ability to quickly release the optical elements in the recovered soft state. The smart adhesive provides a versatile solution for reliably and quickly manipulating and assembling multiscale optical elements with planar or complex 3D shapes without causing surface contamination or damage. These extraordinary capabilities are demonstrated by the manipulation and assembly of various optical elements, such as convex/concave/ball lenses and extremely complex-shaped light guide plates. The proposed smart adhesive is a promising candidate for conventional optical element manipulation technologies.

Surrounding-aware representation prediction in Birds-Eye-View using transformers.

Birds-Eye-View (BEV) maps provide an accurate representation of sensory cues present in the surroundings, including dynamic and static elements. Generating a semantic representation of BEV maps can be a challenging task since it relies on object detection and image segmentation. Recent studies have developed Convolutional Neural networks (CNNs) to tackle the underlying challenge. However, current CNN-based models encounter a bottleneck in perceiving subtle nuances of information due to their limited capacity, which constrains the efficiency and accuracy of representation prediction, especially for multi-scale and multi-class elements. To address this issue, we propose novel neural networks for BEV semantic representation prediction that are built upon Transformers without convolution layers in a significantly different way from existing pure CNNs and hybrid architectures that merge CNNs and Transformers. Given a sequence of image frames as input, the proposed neural networks can directly output the BEV maps with per-class probabilities in end-to-end forecasting. The core innovations of the current study contain (1) a new pixel generation method powered by Transformers, (2) a novel algorithm for image-to-BEV transformation, and (3) a novel network for image feature extraction using attention mechanisms. We evaluate the proposed Models performance on two challenging benchmarks, the NuScenes dataset and the Argoverse 3D dataset, and compare it with state-of-the-art methods. Results show that the proposed model outperforms CNNs, achieving a relative improvement of 2.4 and 5.2% on the NuScenes and Argoverse 3D datasets, respectively.

Recovering building information model from 2D drawings for mechanical, electrical and plumbing systems of ageing buildings

Ageing buildings have become a significant concern for many cities, exacerbated by inadequate management and maintenance of Mechanical, Electrical, and Plumbing (MEP) systems. Building Information Modelling (BIM) enables efficient MEP Operation and Maintenance (O&M) through the digital representation of system information; however, many ageing buildings were constructed without BIM, and manual reconstruction is costly and inefficient due to the sheer number of such structures. Although some studies have proposed methods for automatically recovering BIM from 2D drawings, few are suitable for MEP systems due to the multiscale and irregular shapes of MEP components. To fill this gap, an automatic approach is proposed for recovering MEP model from 2D drawings with three modules: 1) semantic extraction by combining image cropping with Cascade Mask R-CNN to detect and segment multiscale, irregular MEP components; 2) geometric extraction by semantic-assisted image processing to extract contours and skeletons of irregular parts; and 3) Industry Foundation Class (IFC)-based BIM reconstruction via the open-source pythonOCC and IfcOpenShell. The performance was tested on two MEP systems with 335 and 282 multiscale and irregular elements, and the results show that the method recovered BIMs for the two MEP systems in 2.85 s and 0.79 s, with semantic extraction accuracy exceeding 0.9 and geometric error below 5%. This paper contributes to the existing body of knowledge by providing a semantic and geometric-based approach for recovering multiscale and irregular components from 2D drawings. Future studies could further improve the approach by integrating elevation drawings, reconstructing abstract symbols, and aligning text-geometry.

Modeling the resuspension of small inertial particles in turbulent flow over a fractal-like multiscale rough surface

A generalized dynamic resuspension model for particles rolling on a surface with fractal-like multiscale roughness elements is introduced and compared with experimental measurements. Second, nonGaussian stochastic models for the flow velocity seen by a particle are compared with a Gaussian stochastic model. Third, particle resuspension is predicted using large-eddy simulations with a stochastic model for the subgrid-scale flow velocity. The present results confirm the importance of including the multiscale asperity model, nonGaussian flow velocity model, and particle subgrid-scale model to accurately predict resuspension dynamics.

Multi-Scale Massive Points Fast Clustering Based on Hierarchical Density Spanning Tree

Spatial clustering is dependent on spatial scales. With the widespread use of web maps, a fast clustering method for multi-scale spatial elements has become a new requirement. Therefore, to cluster and display elements rapidly at different spatial scales, we propose a method called Multi-Scale Massive Points Fast Clustering based on Hierarchical Density Spanning Tree. This study refers to the basic principle of Clustering by Fast Search and Find of Density Peaks aggregation algorithm and introduces the concept of a hierarchical density-based spanning tree, combining the spatial scale with the tree links of elements to propose the corresponding pruning strategy, and finally realizes the fast multi-scale clustering of elements. The first experiment proved the time efficiency of the method in obtaining clustering results by the distance-scale adjustment of parameters. Accurate clustering results were also achieved. The second experiment demonstrated the feasibility of the method at the aggregation point element and showed its visual effect. This provides a further explanation for the application of tree-link structures.

Efficient field-circuit co-simulation method for GaN-based high power microwave devices

Due to the development of the third-generation semiconductors representative of gallium nitride (GaN), the microwave power devices are developing towards higher power, higher efficiency and high integration. However, the electromagnetic field effects are more significant inside the device. As a result, circuit-level based simulation techniques can no longer satisfy the accuracy requirements of device design. Therefore it is necessary to urgently establish the field-circuit co-simulation techniques to couple the active GaN devices with passive electromagnetic structures. In this work, we propose a high-precision discontinuous Galerkin time-domain method to analyze the performances of GAN-based high-power microwave devices. The extracted large-signal compact model of the GaN HEMT is incorporated into the electromagnetic field equations. A local time-stepping technique is adopted to remove the constraints of nonlinear compact models and multiscale elements on the stability conditions of the global algorithm. The comparisons among numerical simulations, experimental results, and software calculations demonstrate the excellent accuracy and efficiency of the proposed method, which can provide a theoretical analysis and design tool for the high reliability design of advanced high-power microwave devices.

A graph-based top-down visual attention model for lockwire detection via multiscale top-hat transformation

As a locking device, lockwire is widely used in industrial fields that produces high levels of vibration. Lockwire detection is vital for ensuring mechanical stability. However, the existing methods are inapplicable to detect all types of small-size lockwires in complex backgrounds. In this paper, we construct a graph-based top-down visual attention model via the multiscale top-hat transformation to pick out lockwires from various complex backgrounds. Since lockwires typically exhibit separated bright spot-like structures at different scales in images, we firstly construct multiscale anisotropic Gaussian structuring elements to obtain the top-hat feature map. Based on a novel connectivity function, an undirected graph is then constructed. Afterwards, we propose an improved shortest path algorithm to remove prominent complex background components and extract lockwire candidates by minimizing the redefined cost function. Taking full advantages of imaging characteristics of lockwires, we design three saliency metrics to strengthen the saliency of lockwires while weakening backgrounds. Finally, a global top-down saliency map is produced to detect lockwires from complex backgrounds by combining three saliency maps. The experimental results show that our proposed method achieves superior performance compared to state-of-the-art methods.

ASFF-YOLOv5: Multielement Detection Method for Road Traffic in UAV Images Based on Multiscale Feature Fusion

Road traffic elements are important components of roads and the main elements of structuring basic traffic geographic information databases. However, the following problems still exist in the detection and recognition of road traffic elements: dense elements, poor detection effect of multi-scale objects, and small objects being easily affected by occlusion factors. Therefore, an adaptive spatial feature fusion (ASFF) YOLOv5 network (ASFF-YOLOv5) was proposed for the automatic recognition and detection of multiple multiscale road traffic elements. First, the K-means++ algorithm was used to make clustering statistics on the range of multiscale road traffic elements, and the size of the candidate box suitable for the dataset was obtained. Then, a spatial pyramid pooling fast (SPPF) structure was used to improve the classification accuracy and speed while achieving richer feature information extraction. An ASFF strategy based on a receptive field block (RFB) was proposed to improve the feature scale invariance and enhance the detection effect of small objects. Finally, the experimental effect was evaluated by calculating the mean average precision (mAP). Experimental results showed that the mAP value of the proposed method was 93.1%, which is 19.2% higher than that of the original YOLOv5 model.

Hierarchical Urban Land Mappings and Their Distribution with Physical Medium Environments Using Time Series of Land Resource Images in Beijing, China (1981–2021)

Rapid urban expansion and structural changes are taking place in China’s capital city, Beijing, but without an update of urban land features in a timely manner our understanding of the new urban heterogeneity is restricted, as land-background data is indispensable for bio-geophysical and bio-geochemical processes. In this plain region, the investigations of multi-scale urban land mappings and physical medium environmental elements such as slope, aspect, and water resource services are still lacking, although Beijing can provide an exemplary case for urban development and natural environments in plains considering the strategic function of China’s capital city. To elucidate these issues, a remote-sensing methodology of hierarchical urban land mapping was established to obtain the urban land, covering structure and its sub-pixel component with an overall accuracy of over 90.60%. During 1981–2021, intense and sustained urban land expansion increased from 467.13 km2 to 2581.05 km2 in Beijing, along with a total growth rate of 452.53%. For intra-urban land structures, a sharp growth rate of over 650.00% (i.e., +1649.54 km2) occurred in terms of impervious surface area (ISA), but a greening city was still evidently observed, with a vegetation-coverage rate of 8.43% and 28.42% in old and newly expanded urban regions, respectively, with a more integrative urban ecological landscape (Shannon’s Diversity Index (SHDI) = −0.164, Patch Density (PD) = −8.305). We also observed a lower rate of ISA (0.637 vs. 0.659) and a higher rate of vegetation cover (0.284 vs. 0.211) in new compared to old urban regions, displaying a higher quality of life during urban expansion. Furthermore, the dominant aspect of low, medium, and high density ISA was captured with the north–south orientation, considering the sunlight conditions and traditional house construction customs in North China, Over 92.00% of the ISA was distributed in flat environment regions with a slope of less than 15°. When the water-resource service radius shifted from 0.5 km to 0.5–1 km and 1–2 km, high density vegetation displayed a dependence on water resources. Our results provide a new survey of the evolution of hierarchical urban land mapping during 1981–2021 and reveals the relationship with physical medium environments, providing an important reference for relevant research.

Multi-scale numerical simulation analysis of CFST column-composite beam frame under a column-loss scenario

Experimental studies are costly for human and financial resources, while numerical simulation studies can further amplify the conclusions based on experimental results. The conventional numerical model based on brick elements and beam-shell elements is computationally inefficient and inaccurate, respectively. It is advisable to develop a simplified simulation method that ensures computing efficiency while not losing computational accuracy. Consequently, this paper innovatively proposed a numerical simulation method that integrates multi-scale elements modelling and simplified non-essential components, significantly increasing the computing efficiency and accuracy. Subsequently, the concrete-filled steel tubular (CFST) column-composite beam frame models based on the previous tested specimens were numerically developed and studied, considering different simulation methods, column-loss scenarios, and number of storeys. The numerical results identified the cantilevered deep beam effect in a multi-storey composite frame under an edge column-loss scenario, and the simply supported deep beam effect under a penultimate column-loss or a middle column-loss scenario, both reasonably represent the internal force pattern. Furthermore, to enhance the capacity of composite frames to resist progressive collapse, two improvement alternatives were proposed, i.e., fully bolted connection and pre-stressed strand device.

ADRD and COVID-19: A multi-scale framework for elucidating and addressing mechanisms and outcomes.

BackgroundThe COVID‐19 pandemic has disproportionately impacted vulnerable populations. The elderly, including individuals with Alzheimer's Disease and Related Dementias (ADRD) are at an increased risk due to a variety of factors ranging from the molecular, to the physiological and social domains. Here we describe the development of a neural network‐inspired, expanded Adverse Outcome Pathways (AOP) approach to examine the relations, outcomes and potential interventions across scales at the intersections of these acute and long‐term health crises.MethodUsing an Adverse Outcome Pathway (AOP) approach we compared key events and their relationships in COVID‐19 and ADRD. However, whereas current AOP approaches are often limited and centralized around the molecular domain we expand these to encompass a range of domains across scales such as the neuropsychological and social factors. Inspired by neural network modeling approaches we also extend the traditional AOP framework to explicitly include: inhibitory and excitatory mechanisms, causal recurrence, strength of interactions and advantageous pathways.ResultBy including an analysis that spans molecular pathways, system level vulnerabilities, psychological, environmental and social factors we are able to assess commonalities of events (e.g., macrophage activation, systemic inflammation, environmental exposures and disproportionately affected racial/ethnic communities). In addition, by utilizing higher order and systems perspectives, we were able to better identify mechanistic differences (e.g., vulnerabilities of specific brain regions, timescale of bio‐cognitive consequences, sex‐gender differences), as well as chart interactions resulting in comorbidities and resilience factors.ConclusionIn recent years the global ADRD research field has strongly shifted to a conceptual framework that recognizes the essential role of interactions between levels of biosocial organization. The COVID‐19 pandemic has also revealed the crucial biopsychosocial connections of disease vulnerability, resistance and management. AOPs are powerful tools in clarifying key linkages in these multifactorial diseases. However, AOPs have also showed their limitation by biasing molecular events as the prime initiating events. Our hybrid approach integrates multiscale elements to more effectively develop a mechanistic framework and identify pathways to health. Taken together, this new approach can help bridge the divide across bio‐molecular events and higher level factors such as health disparities in COVID and the ADRD spectrum.

Scanning Doppler lidar measurements of drag force on a solitary tree

Abstract

Программный комплекс РАДУГА-Т для моделирования полей нейтронов в ядерно-энергетических установках

Design and operation of nuclear power stations (NPS) are followed by simulation of neutron propagation in these stations. It is necessary to consider borders of multi-scale structural elements consisting of different materials. It is desirable to use parallel computer because of big size of NPSs. To keep such conditions, algorithms and codes for solving the integro-differential transport equation on unstructured grids are being developed. In the paper such algorithms included into the code RADUGA T are outlined. Grids, grid schemes, iterative methods to solve grid equations are presented. Calculation parallelization methods for hybrid computers are considered, MPI and OpenMP techniques are used. Methods of building, improvement, decomposition and visuzlisation of spatial grids are considered. Software implementation is described. The algorithms of the code RADUGA T are compared with methods of other codes. Computation parallelization efficiency study results are presented. The problem of neutron multiplication factor calculation in a light-water reactor model is solved. The multi-processor computer MVS-10P of the Joint SuperComputernal Center is used. Acceleration of each algorithm being used and summary acceleration are given.

AIRSENSE-TO-ACT: A Concept Paper for COVID-19 Countermeasures Based on Artificial Intelligence Algorithms and Multi-Source Data Processing

The aim of this concept paper is the description of a new tool to support institutions in the implementation of targeted countermeasures, based on quantitative and multi-scale elements, for the fight and prevention of emergencies, such as the current COVID-19 pandemic. The tool is a cloud-based centralized system; a multi-user platform that relies on artificial intelligence (AI) algorithms for the processing of heterogeneous data, which can produce as an output the level of risk. The model includes a specific neural network which is first trained to learn the correlations between selected inputs, related to the case of interest: environmental variables (chemical–physical, such as meteorological), human activity (such as traffic and crowding), level of pollution (in particular the concentration of particulate matter) and epidemiological variables related to the evolution of the contagion. The tool realized in the first phase of the project will serve later both as a decision support system (DSS) with predictive capacity, when fed by the actual measured data, and as a simulation bench performing the tuning of certain input values, to identify which of them led to a decrease in the degree of risk. In this way, we aimed to design different scenarios to compare different restrictive strategies and the actual expected benefits, to adopt measures sized to the actual needs, adapted to the specific areas of analysis and useful for safeguarding human health; and we compared the economic and social impacts of the choices. Although ours is a concept paper, some preliminary analyses have been shown, and two different case studies are presented, whose results have highlighted a correlation between NO2, mobility and COVID-19 data. However, given the complexity of the virus diffusion mechanism, linked to air pollutants but also to many other factors, these preliminary studies confirmed the need, on the one hand, to carry out more in-depth analyses, and on the other, to use AI algorithms to capture the hidden relationships among the huge amounts of data to process.

Realizing ultralow thermal conductivity in Cu3SbSe4 via all-scale phonon scattering by co-constructing multiscale heterostructure and IIIB element doping

This article attempts to realize ultralow lattice thermal conductivity in Cu3SbSe4 by substitution Sb site by IIIB elements (Sc/Y), the facial microwave reaction were chosen to fabricate the above samples. The results show that group IIIB elements combined with multiscale heterostructure can effectively suppress the lattice thermal conductivity in Cu3SbSe4 system, resulting in a typical low value of 0.491 W/m/K and 0.481 W/m/K in Cu3Sb0.92Sc0.08Se4 and Cu3Sb0.92Y0.08Se4 at 623 K, respectively. Eventually, a peaking zT value of 0.47 and 0.46 can be obtained for each doping sample. This research highlights a novel strategy by co-constructing multiscale heterostructure and IIIB elements doping for realizing multiscale phonon scattering in Cu3SbSe4 materials.

Optimizing the SERS Performance of 3D Substrates through Tunable 3D Plasmonic Coupling toward Label-Free Liver Cancer Cell Classification

Three-dimensional (3D) plasmonic nanostructures are emerging as excellent surface-enhanced Raman spectroscopy (SERS) substrates for chemical and biomedical applications. However, the correlation of 3D (including both in-plane and out-of-plane) plasmonic coupling with the SERS properties to deepen the understanding of 3D SERS substrates remains a challenge. Here, we perform correlation studies of 3D plasmonic coupling and SERS properties of the 3D hierarchical SERS substrates by tuning the multiscale structural elements. The effects of zero-dimensional (0D; the size of the building blocks), one-dimensional (1D; the thickness of the 3D substrates), and two-dimensional (2D; the composition of individual monolayers) structural elements on 3D plasmonic coupling are studied by performing UV-vis-near-infrared (NIR) spectroscopy and measuring SERS performance. It shows that both the extinction spectra and SERS enhancement are tuned at the 3D structural level. It is demonstrated that the plasmonic resonance wavelength (PRW) stemming from the 3D plasmonic coupling correlates with the SERS averaged surface enhancement factor (ASEF) and is improved by more than tenfold at the optimum 3D nanostructure. The optimized substrate is used to quantitatively analyze two small biological molecules. Moreover, as a proof-of-concept study, the substrate is first applied to differentiate between living liver normal and cancer cells with a high prediction accuracy through the spectral features of the cell membranes and the metabolites secreted outside the cells. We expect that the tuning of plasmonic coupling at the 3D level can open up new routes to design high-performance SERS substrates for wide applications.

Deep convolutional neural network for accurate segmentation and quantification of white matter hyperintensities

White matter hyperintensities (WMHs) appear as regions of abnormally signal intensity on magnetic resonance imaging (MRI) images, that can be identified in MRI images of elderly people and ischemic stroke patients. However, manual segmentation and quantification of images with WMHs is laborious and time-consuming. Moreover, ischemic stroke lesion and WMHs appear as similar signals in MRI images, making it difficult to accurately segment the WMHs. Analysis of WMH-containing images is important for clinical diagnosis, and thus several segmentation methods have been proposed. However, these methods cannot accurately differentiate WMH and ischemic stroke lesions. We propose a deep convolutional neural network, M2DCNN, that can accurately segment WMHs and distinguish them from ischemic stroke lesions. M2DCNN consists of two subnets that rely on a set of novel multi-scale features and a novel architecture (inclusion of dense and dilated blocks). Our model is trained and evaluated on two public segmentation challenges with multi-modality MRI images. Empirical tests demonstrate that M2DCNN outperforms current segmentation methods. We empirically demonstrate that M2DCNN effectively separates WMHs from stroke lesions. Finally, ablation experiments reveal that both multi-scale features and architectural elements in our method contribute to the improved predictive performance.

Mixed multiscale three-node triangular elements for incompressible elasticity

Purpose This study aims to introduce the hybrid finite element (FE) – meshfree method and multiscale variational principle into the traditional mixed FE formulation, leading to a stable mixed formulation for incompressible linear elasticity which circumvents the need to satisfy inf-sup condition. Design/methodology/approach Using the hybrid FE–meshfree method, the displacement and pressure are interpolated conveniently with the same order so that a continuous pressure field can be obtained with low-order elements. The multiscale variational principle is then introduced into the Galerkin form to obtain stable and convergent results. Findings The present method is capable of overcoming volume locking and does not exhibit unphysical oscillations near the incompressible limit. Moreover, there are no extra unknowns introduced in the present method because the fine-scale unknowns are eliminated using the static condensation technique, and there is no need to evaluate any user-defined stability parameter as the classical stabilization methods do. The shape functions constructed in the present model possess continuous derivatives at nodes, which gives a continuous and more precise stress field with no need of an additional smooth process. The shape functions in the present model also possess the Kronecker delta property, so that it is convenient to impose essential boundary conditions. Originality/value The proposed model can be implemented easily. Its convergence rates and accuracy in displacement, energy and pressure are even comparable to those of second-order mixed elements.

A Simplified Computational Strategy Focused on Resin Damage to Study Matrix Cracking of The Cross-Ply Laminates Under Uniaxial Tension Load.

Transverse cracking is probably the first and most dominant mode of damage in composite materials. In this paper, transverse cracking of cross-ply [02/90n]s (n = 2,3,4) laminates under uniaxial tension load was studied by means of experimental and numerical methods. In the numerical simulations, a simplified computational strategy only focusing on the damage of the resin was proposed and the mechanical response of the cracking cross-ply laminates was studied by finite element analysis of multi-scale representative volume elements (RVEs). In the RVEs, the longitudinal 0° plies were represented by macro-scale, homogeneous, orthotropic elastic solids while the 90° plies were modeled by the discrete fibers and the surrounding matrix resin in micro-scale. Based on researching the critical longitudinal mechanical strain which initiates the cracks, the in-situ transverse ply strength and the stiffness degradation of the transverse plies, the simplified computational strategy proposed was proven correct. In addition, the crack initiation is sensitive to residual stress. Higher process-induced residual stress levels are dangerous to laminates, leading to early crack initiation.