Search

Due to the differences in the feature distribution between classes, when the model learns in a continuous data stream, it will encounter catastrophic forgetting. The incremental learning methods have shown great potential to solve this problem. However, most existing methods based on task-incremental learning are difficult to adapt to characteristics of remote sensing scenes with few differences in appearance but large differences in features, which is not conducive to artificially distinguish task-ID. Thus, we propose a class-incremental learning network for small objects enhancing of semantic segmentation in aerial imagery. Specifically, considering the superior accurate of the binary classifier, we propose a Twin-Auxiliary model that adds an auxiliary binary classification task. Then, for expansion and contraction at the edge and small object confusion problems, we introduce a diversity distillation loss, using the results of binary-classifier to constrain the multi-class segmentation results, and strengthen the attention to the locations of the segmentation results that have changed. Finally, we design a conflict reduction mechanism for multi-head classifier to achieve single-head prediction for class-incremental learning. Experiments demonstrate that our method has good performance on the Vaihingen and Potsdam data sets by ISPRS, outperforming state-of-the-art incremental learning methods. The code will be available soon.

The fatigue tests of Chinese RPV steel (A508-3) were carried out under the simulated primary coolant environment of AP1000 which is a kind of 3rd generation nuclear power plant. The effects of temperature, strain rate and dissolved oxygen (DO) on fatigue life were analyzed. During the fatigue test, the material exhibited a cyclic hardening followed by cyclic softening behavior, with no obvious cyclic stability region. In logarithmic coordinates, the S–N curve showed a bilinear relationship. As the strain amplitude increased from 0.2 to 0.6%, the fatigue life approximately decreased from 105 to 3 × 102. The environmental correction factor (Fen), which is defined as the ratio of life in air at room temperature to that in the primary coolant environment, decreases as the temperature decreases from 321 to 150 °C. The Fen also decreases as the DO decreases from 0.6 to 0.03 ppm while it increases as the strain rate decreases from 1 to 0.0004%/s. Compared with the NUREG6909 model, the presented model is more applicable for fatigue life prediction of Chinese A508-3 steel.

A universal numerical model based on the particle size distribution (PSD) approach has been developed for the simulation of precipitation kinetics in multicomponent alloys during isothermal ageing. Nucleation was implemented utilizing the classical nucleation theory (CNT). Growth and coarsening were modeled by a single growth kinetics equation, which is constructed based on the interfacial diffusion flux balance and the capillarity effect. Only partial off-diagonal terms in the diffusion matrix (diffusion of individual components in the matrix) were taken into account in the calculations to minimize the computational cost while coupling with CALPHAD to extract thermodynamics equilibrium around the interface. A new feature of the model is the incorporation of a more realistic spatial site distribution via a Voronoi construction in the characteristic cell, for the purpose of modifying the diffusion distance. Computational predictions of the precipitate dimensions and the precipitation kinetics were compared with the atom probe tomography (APT) measurements on ternary Ni-Al-Cr alloys isothermally aged at 873 K. It is found that the temporal evolution of the dimensions and composition of the precipitates is well captured, as is the dependence on changes in the alloy composition. The new modification with Voronoi construction demonstrates that the overall precipitation kinetics depends on the density and the spatial site distribution of precipitates. The ability to handle sophisticated alloy chemistries by quantitative equations, the compositional sensitivity of microstructural characteristics emerging from the simulation results, and the ability to visualize the spatial distribution of precipitates make the work very promising for multicomponent alloy design and optimization.

A resonant state at 3.21-0.04+0.12MeV, located just above the one-neutron separation threshold, was observed for the first time in Be12 from the Be11(d,p)Be12 one-neutron transfer reaction in inverse kinematics. This state is assigned a spin-parity of 0- according to the systematics of the level scheme of the N=8 isotones and decay-width analysis. Gamow coupled-channel and Gamow shell-model calculations show the importance of the continuum coupling, which dramatically influences the excitation energy and ordering of low-lying states. Various exotic structures associated with cross-shell intruding configurations in Be12 and in its isotonic nucleus Li11 are comparably discussed.

A new ion cyclotron range of frequency (ICRF) antenna, instead of a B-port antenna, is installed at the N-port to improve the power coupling in the 2021 experimental advanced superconducting tokamak (EAST) experimental campaign. In all the experiments, a frequency of 37 MHz is used, corresponding to hydrogen minority heating of deuterium plasma at B 0 = 2.5 T. The maximum coupling resistance is up to 10 Ω, which is close to two times the resistance of the original two antennas in EAST. The plasma stored energy is increased by 30 kJ MW−1, and significant core ion and electron heating are observed. For long pulse operation, we have achieved 1.5 MW/41 s and 1.1 MW/61 s with the new antenna in the high poloidal beta plasma discharge. ICRF power up to 1.8 MW is routinely coupled to the plasma for pulse lengths up to 21 s.

The I-mode is a promising operation mode for fusion in the future, featuring high-temperature and low-density confinement, but the reason why the temperature and density are decoupled remains an important aspect to be explored. The experimental results from the experimental advanced superconducting tokamak (EAST) showed that the weakly coherent mode (WCM) is directly related to sustaining the I-mode and that the peak amplitude of the WCM is proportional to the temperature in the pedestal. Simulating the experimental data from EAST with the six-field model of BOUT++, we find a density perturbation close to the frequency of the WCM observed in the experiment. By testing all the physical terms in this model, we find that the density perturbation and particle transport are directly related to the drift Alfvén wave (DAW) mode. We also use the SymPIC program (Xiao et al Plasma Sci. Technol. 20 110501; Phys. Plasmas 22 112504; Plasma Sci. Technol. 23 055102) to simulate the same experimental data and find that the frequency range of the WCM is close to both experimental and BOUT++ results. Therefore, the WCM of the I-mode can be considered to be driven by the DAW, which helps improve the transport of the I-mode.

The compatibility of efficient divertor detachment with high-performance core plasma is vital to the development of magnetically controlled fusion energy. The joint research on the EAST and DIII-D tokamaks demonstrates successful integration of divertor detachment with excellent core plasma confinement quality, a milestone towards solving the critical plasma–wall-interaction (PWI) issue and core-edge integration for ITER and future reactors. In EAST, actively controlled partial detachment with T et,div ∼ 5 eV around the strike point and H 98 > 1 in different H-mode scenarios including the high β P H-mode scenario have been achieved with ITER-like tungsten divertor, by optimizing the detachment access condition and performing detailed experiments for core-edge integration. For active long-pulse detachment feedback control, a 30 s H-mode operation with detachment-control duration being 25 s has been successfully achieved in EAST. DIII-D has achieved actively controlled fully detached divertor with low plasma electron temperature (T et,div ⩽ 5 eV across the entire divertor target) and low particle flux (degree of detachment, DoD > 3), simultaneously with very high core performance (β N ∼3, β P > 2 and H 98 ∼ 1.5) in the high βP scenario being developed for ITER and future reactors. The high-β P high confinement scenario is characterized by an internal transport barrier (ITB) at large radius and a weak edge transport barrier (ETB, or pedestal), which are synergistically self-organized. Both the high-β P scenario and impurity seeding facilitate divertor detachment. The detachment access leads to the reduction of ETB, which facilitates the development of an even stronger ITB at large radius in the high β P scenario. Thus, this strong large radius ITB enables the core confinement improvement during detachment. These significant joint DIII-D and EAST advances on the compatibility of high confinement core and detached divertor show a great potential for achieving a high-performance core plasma suitable for long-pulse operation of fusion reactors with controllable steady-state PWIs.

Given the low price and relatively high transition temperature (39 K) of MgB2 conductor, MgB2-based superconductors are a potential candidate for the lower field fusion coils, such as poloidal field (PF) coils, correction coils and feeders. However, to date, the application of MgB2 is limited to demonstrators in a low magnetic field of up to 5 T and at temperatures of up to 10 to 20 K, relying on cryogen-free, helium gas or liquid hydrogen cooling, which significantly reduce the cost of cryogenic systems. To demonstrate the feasibility and performance verification of large size MgB2 PF conductors based on ITER and CFETR requirements, a 4th-stage subsize MgB2 cable-in-conduit conductor (CICC) cable sample is made at the Institute of Plasma Physics, Chinese Academy of Sciences. The CICC contains 96 in situ MgB2 superconducting wires, manufactured by Western Superconducting Technology Ltd. and 48 copper wires. The critical current of the sub-size cables and MgB2 witness wires are examined with different background magnetic fields at 4.2 K. In addition, the AC loss is measured utilizing magnetization and calorimetric methods. To further clarify the influence of electromagnetic force on the AC loss performance, the cable sample is pressed transversely at room temperature and then inserted into a dipole magnet for AC loss measurement at 4.2 K. The critical current at 4.2 K of the subsize MgB2 CICC cable shows 20% degradation compared to the witness wires at 2 T background magnetic field. However, no further critical current degradation is visible during ramping up and down the magnetic field. The coupling loss time constant for 1 T background magnetic field amounts to 480 ms. No significant effect of the applied transverse stress on the coupling loss is observed between 0 and 10 MPa.

The effects of defects and building directions in additive manufactured AlSi10Mg on its high-cycle fatigue and very-high-cycle fatigue performance are studied based on crystal plasticity finite element (CPFE). Among the three models provided in this work, the results of model 2 (Np2) are in good agreement with the experimental ones. It is found the fatigue life of the sample with defects are much lower than that without defects. 0° specimens have a better fatigue performance than 90° ones. This work is beneficial in determining the fatigue lifetime and helps to improve the fatigue behavior of AlSi10Mg.

A new 4.6-GHz lower hybrid (LH) current drive (CD) (LHCD) launcher has been successfully developed in the Experimental Advanced Superconducting Tokamak (EAST) to achieve long-pulse high-performance plasma. It is capable of coupling up to 6 MW of LH power into the plasma with a parallel index N// from 1.79 to 2.23. Before manufacturing the launcher, key component mock-ups were fabricated and tested to validate the radio-frequency (RF) design and the process feasibility. Test results show good agreement with the design value. So far, up to 3.5 MW of net LHCD power was injected into the plasma, and long-pulse operation capability has been demonstrated: 1.26 MW and 100.4 s, suggesting that the new launcher can effectively couple the microwave energy into plasma, to drive current and extend the high-performance plasma. In this paper, some of the RF design, construction, testing, and recent experimental results of the new launcher are presented.

The EPED1 model and self-consistent core-pedestal coupling in integrated modeling are used to design the pedestal structure of the China Fusion Engineering Testing Reactor (CFETR) steady-state scenario. The key parameters, such as β p and q 95, are based on the grassy edge-localized-mode (ELM) experimental database. In this work, we use the BOUT++ six-field two-fluid code to simulate the onset of the ELM in the CFETR steady-state scenario. The ELM size is around 0.2% in nonlinear simulations, which is in the experimental range of the grassy ELM discharges, 0.1%–1% observed in multiple tokamak devices. Linear and nonlinear simulations show that the dominant high-n ballooning modes peak around n = 40. Compared to type-I ELM crashing dynamics, grassy ELM crashing has a smaller initial crash and is then followed by three phases of turbulence spreading, which are dominated by multi-modes, a high-n mode of n = 45 and low-n mode of n = 5, respectively. In contras to type-I ELM, the perturbation of the high-n mode has a narrow width around ψ = 0.95, and magnetic island formation and reconnection occur only beyond ψ = 0.95, leading to a small initial crash. Mode–mode interaction in the multi-mode coexistence stage stops the growth of individual modes and reduces the transport of particles and heat, and these are the two reasons why the ELM size is small. In–out asymmetry of transient heat flux with a ratio of E out/E in = 3.5 is found during grassy ELM crash. The rise and delay times of the heat flux match the calculation from the free-streaming model. To evaluate the erosion of the divertor target, the energy fluence at the outer divertor target is calculated, which is 0.029 MJ m−2, 5.5 times smaller than the tungsten melting limit 0.16 MJ m−2. The calculated energy fluency still follows the experimental scaling law from type-I ELM experiments. The fluctuation eddies in the toroidal direction show a filament structure at the outer mid-plane. Parallel heat flux patterns with a toroidal mode number n = 10 are found at the outer divertor with an amplitude of 680 MW m−2.

The martensitic transformation (MT) and its kinetics in Ni-Mn-In-Mg shape memory alloys (SMAs) were calculated for the first time. The differences in the phase transition processes of alloys with various Mg contents were analyzed using two different models, and their microstructures were investigated using differential scanning calorimetry (DSC) and scanning electron microscopy (SEM). As the Mg content increased, the grain size decreased, and the MT temperature and activation energy increased significantly due to the grain refinement of the second phase. The Johnson-Mehl-Avrami-Kolmogorov (JMAK) equation was used to simulate the MT time-temperature-transformation (TTT) curves of the alloys, which exhibited the same trends as the DSC test results. To study the dependence of the activation energy on the heating rate, DSC studies were conducted using nine different heating rates to study the kinetics of the thermally induced transformation process. It was clarified that the heating rate could adjust the phase transition temperature and significantly affect the activation energy, indicating that the heating rate can be used as a method for adjusting the phase change parameters. Moreover, for the same alloy system, using the same heating rate is a useful means of comparing activation energies. This study provides a reference for designing and calculating the kinetics of Ni-Mn-In memory alloys.

The accumulation and dynamic stock of soil organic carbon (SOC) in agricultural land are highly sensitive to changes in climate. However, there is uncertainty about how the global SOC cycle will respond to future climate change. In order to assess changes in SOC storage, basing on the historical weather data and field observations (1951–2018), the study used the DeNitrification-DeComposition (DNDC) model to simulate future SOC storage in an arid oasis under three representational concentration paths (RCPs): RCP2.6, low-end path; RCP4.5, middle stable path; and RCP8.5, high-end path) for the period from 2018 to 2086. Farmland soil in the Yanqi Basin in northern China was divided into grid units using ArcGIS, and then kriging interpolation was used to extract the mean value of the 200 grid units to calculate regional changes in SOC storage. The model indicated that more than 92.75 % of the measured results accurately simulated the SOC storage in Yanqi Basin cropland. The measured SOC storage in the Yanqi Basin was 1.32 Tg C in 2018. The range of SOC density was 653.27–5527.04 kg C/hm2 under the historical climate scenario, and 1863.99–4581.16 kg C/hm2, 2372.74–6910.15 kg C/hm2, and 623.656–5431.95 kg C/hm2 under the RCP 2.6, RCP 4.5, and RCP 8.5 scenarios, respectively. The simulation results show that, compared with the SOC stock in 2018, the SOC stocks in the oasis area in the future will show a rapid decline under the historical climate scenario (7.6 % decrease) as well as in the RCP 2.6 (3% decrease), RCP 4.5 (9.17 % decrease), and RCP 8.5 (16 % decrease) scenarios. These results indicate that organic carbon outputs and density will decrease faster under high emissions and more slowly under low emissions. This study provides supporting data and a scientific basis for increasing soil organic carbon in cropland and for the management of oasis soil to mitigate the effects of future climate change.

Three kinds of methods based on extrusion and 3D printing and different acidic solutions (formic acid (FA), acetic acid (AA), glycolic acid (GA) and lactic acid (LA)) were applied for manufacturing the CS ducts. The tensile properties and preliminary cytotoxicity were measured for selecting the optimal ratio of CS slurry. The 3D printability of CS slurry was also studied. The tensile strength, Young’s modulus, and fracture strain were tested for evaluating the degree of mechanical matching to soft-tissue. The optimal solvent to CS was 30 wt.% GA solution. The CS slurry possessing shear-thinning properties was suitable for 3D printing. The tensile strength, Young’s modulus, and fracture strain of the CS rods were 10.98 ± 0.61 MPa, 12.38 ± 1.19 MPa, and 146.03 ± 15.05 %, correspondingly. The CS ducts manufactured by 3D printing had an excellent mechanical matching to soft-tissue, outstanding biocompatibility and have great potential for soft-tissue restorations.

In order to achieve steady state and high performance plasma in EAST, a new lower hybrid current drive system at a frequency of 4.6 GHz has been built. The system is composed of 24 continuous wave (CW) klystron amplifiers to generate 6MW/CW microwave, 24 standard rectangle waveguide transmission lines with water-cooling plate, a multi-junction grill composed of 576 active (in groups of 8) and 84 passive sub-waveguides arranged in 12 rows and 6 columns, and four sets of high voltage power supplies. The power value and the spectrum of the launched microwave from the antenna can be controlled by the low-power microwave circuits in front of the klystrons. The new LHCD system has been applied to the experiments on EAST tokmak since 2014, and the obtained results suggest that it is effective to couple the wave into plasma and drive plasma current.

Using synchrotron radiography, we observed the advancing L1/L interface and its interaction with L2 droplets during directional solidification of the Al-10 wt% Bi immiscible alloy. Five forces act upon L2 droplets in front of the L1/L interface. The repulsive force that pushes the droplets away from the interface is the largest force when the droplet radius (r) is smaller than 20.5 µm, while Marangoni force is the largest one when 20.5 µm

The martensite transformation (MT) behavior and magneto-caloric effect (MCE) in polycrystalline Ni44Mn36In14Co6 alloy were tuned by doping the fifth element, 2 at.% Cu. The process of heat-induced MT was observed by in-situ scanning electron microscopy and the magnetic-field-induced reversible MT was confirmed by magnetic properties test near room temperature. A large magnetization change (ΔM, 111emu/g), relatively high sensitivity to magnetic fields (3.8 K/T), a large MCE (33.7J kg−1 K−1) and the two-way magnetic-field-induced transformation were attained around the MT. The combination properties in the Ni42Mn36In14Co6Cu2 alloy are instructive for the development of high-performance magnetic refrigeration and intelligent actuator materials.

We have studied the giant elastocaloric effect of directionally solidified Ni–Mn–In magnetic shape memory alloys (MSMAs) near room temperature. The textured microstructure significantly enhances the ductility of Ni–Mn-based MSMAs, which allows obtaining the martensitic transition characteristics of polycrystalline samples in a uniaxial stress field. A significant cooling level of about 4 K is measured when the stress is released. The experimentally measured temperature changes are found to be in good agreement with the theoretically calculated values.

In the last years several V-SLAM(Visual Simultaneous Localization and Mapping) approaches have appeared showing impressive reconstructions of the world. However these maps are built with far more than the required information. This limitation comes from the whole process of each key-frame. In this paper we present for the first time a mapping method based on the LOOK UP TABLE(LUT) for visual SLAM that can improve the mapping effectively. As this method relies on extracting features in each cell divided from image, it can get the pose of camera that is more representative of the whole key-frame. The tracking direction of key-frames is obtained by counting the number of parallax directions of feature points. LUT stored all mapping needs the number of cell corresponding to the tracking direction which can reduce the redundant information in the key-frame, and is more efficient to mapping. The result shows that a better map with less noise is build using less than one-third of the time. We believe that the capacity of LUT efficiently building maps makes it a good choice for the community to investigate in the scene reconstruction problems.

The foreland basin of piedmont belt is one of the oil prospects and a current hotspot for exploration in western China. Complex surface and structures put forward higher requirements for surface correction and migration imaging. Usually, static correction only solves the vertical component of travel time, and conventional time migration can't accommodate the lateral changes in velocity. Based on the Walkaway VSP data of one well in western China, we studied the Walkaway VSP velocity tomography modeling, wave field continuation and prestack depth migration imaging of complex surface in this paper. The core idea includes updating the velocity model, extending the wave field, realizing the near-surface correction, and completing the f-x prestack depth migration. Know from the test of model data and measured data, the velocity modeling from reflection travel time tomography can provide an accurate velocity model for prestack depth migration, and the one-way wave field continuation and f-x prestack depth migration migration imaging, can effectively solve the wave field correction problem of complex surface. The results show that this method can be applied to prestack depth migration imaging of complex surface and lateral varied velocity medium with universal applicability.

EAST, with full superconducting magnetic coils, has been designed and constructed to address the scientific and engineering issues under steady state operation. It has passive stabilizers with the function of protecting the vacuum vessel, heating systems and diagnostic components from the plasma particles and heat loads. As the vertical plasma motion is unstable in tokamak with elongated D-shape plasma, the passive stabilizers and in-vessel active feedback coils are used for vertical plasma position control in EAST. To accommodate with the new stage for high performance plasma and enhance the control of vertical stabilization in EAST, the project of passive stabilization loop (PSL), connect all the independent passive stabilizers together as a loop, has been accomplished. The PSL consists of 32 passive stabilizers and the current bridges. All the supporting structures of PSL are insulated from the vacuum vessel. The separate passive stabilizers are electrically connected together forming the upper and lower toroidal loops, and then the two loops are electrically connected by the copper current bridges. The current bridges are designed coaxial and symmetrical in order to counterpoise the electromagnetic forces. There are insulation layer and external resistor between the outer and inner bridges to avoid breakdown. What’s more, a stainless steel shell with graphite tiles is covered on the bridges to protect them from the plasma. The total resistance of PSL is reduced to 150 micro ohms by silver plating on the contact surface. As indicated, the PSL was fabricated and assembled successfully and met the design requirement for the plasma operation. With the PSL, the control of the vertical stabilization is improved.

Experiments with liquid lithium limiters (LLLs) have been successfully performed in HT-7 since 2009 and the effects of different limiter surface structures on the ejection of Li droplets have been studied and compared. The experiments have demonstrated that strong interaction between the plasma and the liquid surface can cause intense Li efflux in the form of ejected Li droplets – which can, in turn, lead to plasma disruptions. The details of the LLL plasma-facing surface were observed to be extremely important in determining performance. Five different LLLs were evaluated in this work: two types of static free-surface limiters and three types of flowing liquid Li (FLLL) structures. It has been demonstrated that a FLLL with a slowly flowing thin liquid Li film on vertical flow plate which was pre-treated with evaporated Li was much less susceptible to Li droplet ejection than any of the other structures tested in this work. It was further observed that the plasmas run against this type of limiter were reproducibly well-behaved. These results provide technical references for the design of FLLLs in future tokamaks so as to avoid strong Li ejection and to decrease disruptive plasmas.