Adapting the target dataset for a pre-trained model is still challenging. These adaptation problems result from a lack of adequate transfer of traits from the source dataset; this often leads to poor model performance resulting in trial and error in selecting the best performing pre-trained model. This paper introduces the conflation of source domain low-level textural features extracted using the first layer of the pretrained model. The extracted features are compared to the conflated low-level features of the target dataset to select a higher quality target dataset for improved pre-trained model performance and adaptation. From comparing the various probability distance metrics, Kullback-Leibler is adopted to compare the samples from both domains. We experiment on three publicly available datasets and two ImageNet pre-trained models used in past studies for results comparisons. This proposed approach method yields two categories of the target samples with those with lower Kullback-Leibler values giving better accuracy, precision and recall. The samples with the lower Kullback-Leibler values give a higher margin accuracy rate of 6.21% to 7.27%, thereby leading to better model adaptation for target transfer learning datasets and tasks
Fluoroelastomer has received increasing attention for energetic materials application due to its high fluorine contents. Different contents of poly(VDF-ter-HFP- ter- TFE) terpolymer are added into Al/MnO2 nanothermite. The peak exothermic temperature of thermite reaction for Al/MnO2 system is about 554 oC with 1070 Jg-1 heat release. After adding terpolymer, it mainly exists in the gap among Al nanoparticles and MnO2 nanorods, and can react with Al and MnO2 at the range of 350 oC to 540 oC before the occurrence of thermite reaction. 10wt% terpolymer has relatively little effect on the thermite reaction, and for the samples with higher terpolymer content, more nanothermite components reacts with terpolymer at early stage. Ignition and combustion performance show terpolymer can reduce ignition energy threshold by up to 9.82% and increase combustion duration time at least several times. The potential reasons for above results are analyzed. This work can shed light on application of fluoroelastomer in energetic-materials.
Application of deep learning (DL) for automatic condition assessment of bridge infrastructure has been on the rise in the last few years. From the published literature, it is evident that lot of research efforts has been put in identifying the surface defects such as cracks, potholes, spalling etc. using deep learning. However, a concrete bridge deck health is jeopardized by the presence of subsurface defects substantially, however, the task of defect detection using deep learning has not received the proper attention. The goal of this survey paper is to provide a critical review of existing technical knowledge for DL application on NDE data for bridge deck evaluation. The authors reviewed prominent NDE techniques for subsurface defect detection of bridge decks and explored the various DL models proposed to identify these defects. First a brief overview of the working principle of NDE techniques and DL architectures is provided, and then the information about proposed DL models and their efficacy is highlighted. Based on the existing knowledge gaps, various challenges and future prospects associated with application of DL in bridge subsurface inspection are discussed.
The usage of the gas sensor has been increasing very rapidly in the industry and in daily life for various potential applications. In recent years, metal oxide semiconductors (MOS) become the primary choice for designing highly sensitive, stable, and low-cost real-life applications-based gas sensors due to their inherent physical and chemical properties. Researchers have proposed numerous sensing mechanism to explain the functionality of MOS based gas sensors. In this review, we have comprehensively covered different sensing mechanisms used for MOS. We have also discussed different parameters affecting the sensitivity and selectivity of the gas sensors. Moreover, the different techniques used to enhance the gas sensing response of MOS based sensors are also extensively covered. And finally, we give our prospective on recent opportunities and challenges on future applications of MOS based gas sensors.
Pipeline flow visualization of cemented tailings backfill slurry (CTBS) improves the safety and stability of transportation. High turbidity and low resolution make it difficult for conventional methods to monitor the particle distribution state of CTBS in a short period of time. Particle tracking technology (PTT) is used to simulate and investigate the flow characteristics of CTBS pipeline, combine with theoretical analysis to construct a CTBS pipeline visualization model, elaborate the particle distribution state when CTBS flows in the pipeline, and explore the effects of pipe diameter (PD), flow velocity (FV) and tailings gradation (TG) on the particle distribution. The results show that particle tracking technology is better applied to investigate the particle transport distribution characteristics of CTBS tailings. Three concepts of particle accumulated gravity Ga, static friction angle θ and diameter dividing line are defined, and the transport pipe is divided into light wear zone, medium wear zone and heavy wear zone. The increase in pipe diameter increases the content of fine particles at the pipe wall and the thickness of the lubrication layer becomes larger, which improves the safety and stability of CTBS transport. The increased flow velocity reduces the settling phenomenon of large size particles and improves the transport efficiency, which increases the pipeline transport resistance. The wider the range of tailings gradation and the smaller the ratio of the number of large size tailings to small size tailings, the more suitable the tailings are for pipeline transportation as a backfill aggregate.
Large eddy simulation (LES) is used to simulate flame acceleration (FA) and deflagration to detonation transition (DDT) of methane–air mixtures in a small-scale 3D channel. The simulation results show that, in the early stages, the flame velocity increases exponentially because of the expansion of combustion products and the wrinkle of flame surface. In the next stage, the interaction between flame and pressure wave makes flame accelerate continuously, and the acceleration rate of the flame velocity decreases first and then increases. As the pressure of the leading shock increases, the boundary layer is heated by the preheating area in front of the flame surface which causes the ultrafast flame propagates in the boundary layer. The ultrafast flame generates oblique shock waves continuously moving to the center of the channel and colliding with each other, which promote the occurrence of local explosion and the coupling of flame surface and leading shock wave.
Aircraft cabins have high-performance ventilation systems, yet typically hold large numbers of people in close proximity for long periods. The current study estimated airborne virus exposure and infection reductions for vacant middle seats and masking in aircraft. Tracer particle data reported by U.S. Transportation Command (TRANSCOM) and CFD simulations reported by Boeing were used, along with NIOSH data, to build nonlinear regression models with particle exposure and distance from particle source as variables. These models that estimate exposure at given distances from the viral source were applied to evaluate exposure reductions when middle seats are vacant compared to full occupancy. Reductions averaged 54% for the seat row where an infectious passenger is located and 36% for a 24-row cabin containing one infectious passenger, with middle seats vacant. Analysis of the TRANSCOM data showed that universal masking (surgical masks) reduced exposures by 62% and showed masking and physical distancing provide further reductions when practiced together. For a notional scenario involving 10 infectious passengers, compared with no intervention, masking, distancing, and both would prevent 6.2, 3.8 and 7.6 secondary infections, respectively, using the Wells-Riley equation. These results suggest distancing and masking reduce SARS CoV-2 exposure risk when an infectious passenger is present.
Numerical method was widely used in bridge seismic performance analysis, but the diversity and complexity structures increased trouble in the modelling process. This paper presented a systematic, universal and flexible bridge numerical modelling method, and the seismic performance of a concrete filled steel tubes arch bridge was studied. The geometric/attribute model involved structural/stiffness equivalence, and the bridge parametric modeling strategy combined commercial computer-aided design language of software MATLAB and ANSYS. Data exchange in different fields and rapid reconstruction of numerical model were realized. Finally, the vibration mode, deformation and internal force of the bridge under seismic load was studied. The results indicated that the proposed parametric model formed a complete data transfer process, and the bridge model rapidly modified and reconstructed. The numerical calculation efficiency was greatly improved because of the structure element was employed instead of solid element. The displacement and internal force of arch rib were mainly controlled by lateral load under horizontal uni-input, and the combined horizontal and vertical load should be considered in seismic design. The leaning angle and width-span ratio of transverse braces are considered to improve the seismic performance of bridges.
Healthcare interoperability unfolds the way for personalized healthcare services at a reduced cost. Furthermore, a decentralized system holds the promise to prevent compromises such as cyber-attacks due to data breaches. Hence, there is a need for a framework that seamlessly integrates and shares data across the system stakeholders. We propose SENSIBLE, a blockchain-powered, knowledge-driven data-sharing framework that gives patients complete control of their medical history and can extract rich information hidden in it using knowledge graphs (KGs). By incorporating both blockchain and KGs, we can provide a platform for secure data sharing amongst stakeholders by maintaining data privacy and integrity through data authentication and robust data integration. We present a Proof-of-Concept of the SENSIBLE network with Ethereum to share dynamic knowledge across stakeholders. Dynamic knowledge generation on the blockchain provides a two-fold advantage of cooperation and communication amongst the stakeholders in the healthcare ecosystem. This leads to operational ease through sharing relevant portions of complex information while also ensuring the isolation of sensitive medical data.
This paper aims to investigate the synergistic effects in moisturizing activity and antioxidant activity between the promising unconventional polysaccharide and flavonoids from stems of Trollius chinensis Bunge (TCS). The results showed that the mixture with the mass ratio (w/w) of 7:3 (flavonoids to polysaccharides) appeared better moisture retention (73.08 ± 2.4%) and scavenging effects on·OH radicals (85.46±0.52 %). Meanwhile, the mixture with the mass ratio (w/w) of 3:7 (flavonoids to polysaccharides) unveiled better scavenging effects on DPPH radicals (44.10±0.81%) and reducing capacity. The results confirmed that the polysaccharides and flavonoids from TCS have good synergistic effects in moisturizing activity and antioxidant activity, and have the potential to be used in the food industry as edible films or edible packaging materials.
Cracks in concrete structures can serve as pathways for aggressive chemical substances that can lead to a progressive deterioration of the cement stone as well as of the reinforcement, affecting the load capacity, service life and useability of concrete structures. However, concrete and reinforced concrete exhibit an intrinsic ability to heal cracks, defined as autogenous self-healing. This effect includes the precipitation of calcium carbonate in the presence of water and CO2 and is accompanied by continued hydration, swelling and mechanical blocking of the crack pathway. Experiments led to the inclusion of crack healing by autogenous self-healing in Eurocode 1992-3 for water retaining concrete structures. However, despite code restrictions, autogenous self-healing of concrete shows limited effectiveness in practice. This indicates the need for further research to provide engineers with reliable design rules. Therefore, this study aims for giving a broad literature review on the state-of-the-art knowledge on autogenous self-healing, the boundary conditions, consensus and controversy of processes and factors influencing the efficiency of autogenous self-healing. Regarding the transferability of laboratory results to real concrete constructions, materials, crack initiation techniques, experimental concepts and methods for assessing the effectiveness of autogenous self-healing are discussed and recommendations for future experiments are set.
This study reports on performance enhancement in a solar cell introducing Sb2S3 as hole transport layer (HTL)along WS2 as buffer layer. We have investigated photovoltaic (PV) characteristics by utilizing SCAPS-1D. A comparative observation on PV performances of conventional CZTS/CdS with proposed CZTS/WS2 and Ni/Sb2S3/CZTS/WS2 /FTO/Al solar cells is presented. It is revealed that “spike like” band structure at CZTS/WS2 interface having smaller conduction band offset makes it potential alternative to commonly used CdS buffer. This study also evaluates that Sb2S3 as an HTL proposed at back of CZTS enhances performances by reducing carrier recombination at back interface with appropriate band alignment. The impacts of thickness, carrier concentration of different layers, and bulk defect density in CZTS as well as defects at interfaces on performances are analyzed. The influences of temperature, work function as well as cell resistances are also explored. Optimum absorber thickness of 1.0 µm along doping of 10 17 cm-3 is selected. A maximum efficiency of 30.63% is achieved for the optimized CZTS cell. Therefore, these results suggest that Sb2S3 as HTLand WS2 as buffer layer can be employed effectively to develop highly efficient and low-cost CZTS solar cells.
Square loop and square slot are the simplest and commonly used elements of frequency selective surfaces (FSSs) providing bandstop and bandpass responses, respectively. Despite it is already known that employment of a double square slot structure presents a tunable stopband between two passbands, it is not clear which parameters of the structure are crucial for tunning the inbetween stopband. We propose an analysis of a double square slot FSS considering its element as a parallel connection of square slot and square loop predicting two geometrical parameters of the constituent square loop are crucial for tunning the inbetween stopband. The stopband is tunable by changing geometrical parameters of the constituent square loop and does not depend on geometries of the constituent square slot. The validity of this analytical prediction is verified by full wave EM simulation results.
The selection of layers in the transfer learning fine-tuning process ensures a pre-trained model’s accuracy and adaptation in a new target domain. However, the selection process is still manual and without clearly defined criteria. If the wrong layers in a neural network are selected and used, it could lead to poor accuracy and model generalisation in the target domain. This paper introduces the use of Kullback-Leibler divergence on the weight correlations of the model’s convolutional neural network layers. The approach identifies the positive and negative weights in the ImageNet initial weights selecting the best-suited layers of the network depending on the correlation divergence. We experiment on four publicly available datasets and four ImageNet pre-trained models that have been used in past studies for results comparisons. This proposed approach method yields better accuracies than the standard fine-tuning baselines with a margin accuracy rate of 10.8% to 24%, thereby leading to better model adaptation for target transfer learning tasks.
A method to reduce crosstalk using TL-shaped defect microstrip structure (DMS) is proposed to solve the far-end crosstalk between microstrip lines. This method optimizes the ratio of the capacitive coupling and the inductive coupling between the coupled microstrip lines by etching the TL-shaped DMS on the microstrip line and reduces the strength of the electromagnetic (EM) coupling, which can achieve crosstalk suppression. The equivalent circuit model, S-parameters and full-wave EM simulations are used to analyze the crosstalk between the microstrip lines etched with and without the TL-shaped DMS. High Frequency Structure Simulator (HFSS) software simulation and samples test results show that the TL-shaped DMS can effectively reduce the far-end crosstalk while guaranteeing the transmission ability of microstrip line to the signal. The maximum far-end crosstalk can be reduced by 42dB in the frequency range of 0–8 GHz and the test results of the samples are in good agreement with the simulation results.
Expansive soil is a kind of special soil with soil particles mainly composed of hydrophilic minerals, and has significant deformation characteristics of water absorption expansion and water loss shrinkage. This characteristic often brings great harm to engineering construction. Relying on the foundation engineering of the bridge on the expansive soil slope of the Jiangxi-Huaihe River, this paper studies the effects of slope excavation and rainfall on the horizontal deformation of the bridge foundation by using temperature simulation and humidity fields and numerical simulation methods. The research results show that during the excavation process, the bridge foundation will produce a horizontal displacement pointing to the top of the slope, and the rainfall after excavation will cause the horizontal displacement of the bridge foundation to decrease and gradually develop in the opposite direction. During the excavation process, the rebound of slope excavation is the main factor affecting the horizontal displacement of the bridge foundation. Under short-term rainfall conditions, the hygroscopic expansion of the slope soil becomes the main influencing factor. Under extreme rainfall conditions, the main cause of the horizontal deformation of the bridge foundation is The influencing factors are transformed into the reduction of the soil weight of the slope and the strength of the soil material.
Time series classification (TSC) is an important and challenging problem in data mining. Time series data sets are an important basis for this research and are widely used in baseline verification of various algorithm models. Aiming at the problem that there are few domestic data sets and the current TSC data set is relatively old, a new data set for TSC task is established based on the average price data of concrete in major cities in China, which provides new data support for the research of TSC algorithm. We made use of the data center of Oriental Fortune to disclose the sample data of the average price of concrete from 2013-10-23 to 2021-01-20, created 730 autoregression-based series data sets by using sliding windows of different lengths, and then selected the appropriate sliding window length through machine learning model verification, finally, convolutional neural network (CNN) and long and short memory (LSTM) network, which are good at processing temporal features, are used to verify the data and prove the validity of the dataset. The dataset is freely available at https://gitee.com/lq2012/tsc-dataset.
This paper presents a thorough methodology for the voltage assessment of pilot offshore wind project in Shandong province, China. The results presented in this paper can be considered as the milestone of the offshore wind research in Shandong since there is still no grid codes for offshore wind power plant grid connection in the local electrical power grid. It mainly consists of three parts. In the first part, a detailed model of the offshore wind farm created in the DIgSILENT PowerFacotry simulation platform is presented, including wind turbines, power converters, transformers, submarine cables and relevant control schemes. In the second part, nonlinear time-domain simulations were performed to analyse the wind farm’s active power, reactive power, and voltage conditions under different wind scenarios. Based on the simulations results, a dynamic reactive power compensation system was proposed, and the consequence of the reactive power compensation was also demonstrated using nonlinear time-domain simulations.