Physics-Informed Neural Networks (PINNs) and their variants have gained widespread attention as scientific machine learning (SciML) methods for solving Partial Differential Equations (PDEs), but their computational efficiency and accuracy in low-dimensional problems are still lower than that of classical numerical methods, such as Finite Element Method (FEM). This limitation mainly stems from the generalization error of PINNs, whose point-wise loss functional ignores the interactions between neighboring collocation points. To address this challenge, we propose a new Physics-Graph-Informed Neural Networks (abbreviated as PGINNs) framework that combines deep learning with a conservation-consistent nodal network. Compared to traditional PINNs, PGINNs encode two fundamental forms of graph-topological physical information: (1) global node topology operator represented by the sparse incidence matrix, and (2) local edge feature matrix learned through deep neural networks (DNNs). By reconstructing the loss function as a system of algebraic equations learned by the nodal network representation, PGINNs effectively reduce the generalization error while retaining the mesh-free advantage of PINNs. Theoretical analysis shows that the generalization error convergence rate of PGINNs is improved by an order of magnitude compared with that of PINNs. Numerical experiments for complex boundaries and piecewise homogeneous media validate the high accuracy and computational efficiency of the method. This work combines the advantages of both classical numerical analysis and data-driven PDE solvers, establishing a new direction for high-fidelity SciML.

The integration of two-dimensional graphene with gold nanostructures has significantly advanced surface plasmon resonance (SPR)-based optical biosensors, due to graphene’s exceptional optical, electronic, and surface properties. This review examines recent developments in graphene-based hybrid nanomaterials designed to enhance SPR sensor performance. The synergistic combination of graphene and other functional materials enables superior plasmonic sensitivity, improves biomolecular interaction, and enhances signal transduction. Key focus areas include the fundamental principle of graphene-enhanced SPR, the functional advantages of graphene hybrid platforms, and their recent applications in detecting biomolecules, disease biomarkers, and pathogens. Finally, current limitations and potential future perspectives are discussed, highlighting the transformative potential of these hybrid nanomaterials in next-generation optical biosensing

Recently, physics-informed neural networks (PINNs) and their variants have gained significant popularity as a scientific computing method for solving partial differential equations (PDEs), whereas accuracy is still its main shortcoming. Despite numerous development efforts, there is no literature demonstrating that these methods surpass classic numerical algorithms in solving the forward issue. In this paper, by analyzing the disparities between PINNs and traditional numerical methods based on mesh discretization, we investigate the underlying causes for the in adequate precision of PINNs and introduce a novel approach named global physics-informed neural networks (GPINNs). Inspired by the crucial concept of global nodal association in conventional numerical algorithms, GPINNs leverages the prior field distribution information from pre-trained PINNs to estimate the association weights between arbitrary nodes in space. GPINNs can not only be regarded as a meshless approach but also be demonstrated, both theoretically and in practical circumstances, to have the ability of second-order convergence when trained with equidistant nodes. Overall, GPINNs may be seen as an ideal approach to inheriting the merits of scientific machine learning (SciML) and conventional numerical computing, which also represent the first SciML algorithm to surpass standard numerical methods in terms of accuracy.

The existing GIS disconnector contact state evaluation method is based on the single state characteristic quantity, the reliability of the evaluation result is low, and it is easy to misjudge and erroneous judgment. So a method of GIS disconnector contact state evaluation based on multi-feature fusion was presented. The multi state quantity comprehensive detection experimental platform of 220 k V GIS disconnector was built. And the relationships between the GIS disconnector shell temperature signal, shell vibration signal, partial discharge signal and the contact state of the disconnector were experimentally studied. On this basis, the temperature rise of GIS disconnector shell, the amplitude of shell vibration signal and the discharge amplitude of partial discharge UHF signal were taken as the state characteristic quantities of disconnector, and the contact state evaluation model of disconnector based on support vector machine was established. The test results show that the accuracy of GIS disconnector contact state evaluation method based on multi-feature fusion is the highest, can reach 92.92%.

The magnetically controlled current transformer (MCCT) is proposed to harvest stable energy from the magnetic field induced by the transmission-line fluctuating current. Compared with the conventional current transformer (CT), the advance of the proposed MCCT originates from the so-called constant magnetic flux control, which is achieved by connecting an electronic load on the secondary side of MCCT, and adjust the load impedance to make it equivalent to the corresponding resistive-capacitive impedance for magnetization or demagnetization. Furthermore, the working principle and the excitation characteristic of the MCCT system have been analyzed using the equivalent magnetic circuit and equivalent circuit methods. In addition, the details of the electronic load circuit and control system of the MCCT are presented in this paper, and the stability of the system is verified using logarithmic criterion and total harmonic current distortion (THDi) analysis. Finally, the co-simulation and experimental results show that MCCT can produce a stable output power of 13 W in the range of 50–500 A current fluctuations, demonstrating that the proposed magnetic energy harvester is capable of a high conversion efficiency and output stability.

Human beings are facing severe global environmental problems and sustainable development problems. Effective separation technology plays an essential role in solving these challenges. In the past decades, superwettability (e.g., superhydrophobicity and underwater superoleophobicity) has succeeded in achieving oil/water separation. The mixture of oil and water is just the tip of the iceberg of the mixtures that need to be separated, so the wettability-based separation strategy should be extended to treat other kinds of liquid/liquid or liquid/gas mixtures. This review aims at generalizing the approach of the well-developed oil/water separation to separate various multiphase mixtures based on the surface superwettability. Superhydrophobic and even superoleophobic surface microstructures have liquid-repellent properties, making different liquids keep away from them. Inspired by the process of oil/water separation, liquid polymers can be separated from water by using underwater superpolymphobic materials. Meanwhile, the underwater superaerophobic and superaerophilic porous materials are successfully used to collect or remove gas bubbles in a liquid, thus achieving liquid/gas separation. We believe that the diversified wettability-based separation methods can be potentially applied in industrial manufacture, energy use, environmental protection, agricultural production, and so on.

该文提出一种基于动态无功补偿技术的磁平衡式电流互感器设计方法,其应用能够实现对大范围波动电流的精确测量.通过对电流互感器工作原理及其测量误差的分析,提出基于动态无功补偿的磁平衡式电流互感器控制算法,即采用自适应扰动法实现对电流互感器二次侧测量电流的最大电流跟踪.建立了基于Comsol和Simulink联合仿真的磁平衡式电流互感器场-路耦合模型,仿真和实验验证了所提出磁平衡式电流互感器模型的有效性以及控制算法的可行性.

Vacuum arc is a special metal vapor discharge phenomenon, because its discharge medium totally comes from the evaporation and ionization of electrode materials. In the case of low current, the vacuum arc is completely composed of plasma jets emitted from discrete cathode spots on the cathode surface and the current carried by each spot depends on the cathode material. When the arc current exceeds a certain value, a certain number of cathode spot plasma jets will appear. Vacuum arcs play a very important role in some industrial applications such as vacuum circuit breakers, vacuum coatings and electric thrusters. As an important plasma control method, the external axial magnetic field (AMF) has an important influence on the macroscopic morphology and microscopic parameter distribution of the vacuum arc. Various studies of vacuum arc under AMF have been carried out and some progress has been made. However, the existing literature about the simulation research of vacuum arc is mostly concentrated in the case of large current, and less attention is paid to the case of small current. The reason is that the traditional methods, magneto-hydrodynamics or particle-in-cell, are limited by either accuracy or efficiency, and cannot be effectively applied to the low current vacuum arc plasma jet simulations. In this paper, we develop a fully three-dimensional hybrid plasma simulation algorithm to study the single cathode spot vacuum arc plasma jet under AMF. In this model, ions are modelled as particles while electrons are treated as massless fluid, and the self-generated magnetic field is also considered. To simplify the condition, the cathode spot in our model only exists as a plasma jet source, thus the detailed mechanism of producing plasmas is neglected. And the movement of the cathode spot is not considered either. The results show that the single cathode spot plasma jet diffuses into the interelectrode in a cone shape after leaving the cathode spot, and the ion density drops rapidly from cathode to anode. Under the simulation conditions in this paper (I ≤ 150 A), the self-generated magnetic field will not have a significant influence on the plasma jet itself in the case of low current. The external AMF has a compressive effect on the diffusion of the vacuum arc plasma jet. Under the AMF, the radial movement of the ions is suppressed, and the decrease of the ion radial velocity leads to a smaller diffusion radius of the jet. This compression effect of the AMF on the plasma jet is related to both the intensity of the external AMF and the magnitude of the arc current. In the case of a constant arc current magnitude, the compression effect gradually increases as the value of the AMF intensity gradually increases; in the case of a constant value of the external AMF, the compression effect gradually decreases as the current gradually becomes larger.

针对干式空心电抗器早期匝间短路故障,通过分析电抗器匝间电弧性短路过程中的状态参数变化规律,提出了基于时变参数模型的电抗器故障特征提取及诊断方法.首先建立了基于Mayr电弧模型的电抗器匝间短路故障的时变电路模型,并利用傅里叶变换分析了故障电抗器等效电感的频谱特性,发现其等效电感不再恒定而是存在周期性振荡.然后利用时变电路状态方程求解方法,分析在工频激励下故障电抗器输出电流的转移解、零状态响应及频谱特性,证明在匝间电弧的周期性短路过程中,故障电抗器输出电流中将激发出特定次数的谐波分量,分析电流谐波特性可对干式空心电抗器匝间短路故障进行诊断.通过对一起电抗器故障的电流互感器(CT)和电压互感器(PT)录波数据进行求解和分析,验证了本文提出的故障特征提取及诊断方法的可行性和有效性.

In this exploratory study, selective tungsten (W) deposition is used before the copper (Cu) metallization steps with the aim to solely fill the via with W. The W deposition is tested on the bottom metal cobalt (Co) or ruthenium (Ru) and shows an excellent selectivity towards the dielectric SiO 2 and dense low-k material 3.0. The via resistance shows up to a 40% reduction for the W-Cu hybrid system compared to a Cu dual damascene (DD) filled via. The material compatibility is tested in a thermal storage study and shows no performance degradation of the bottom barrierless W vias. This feasibility study using a middle of line (MOL) metal shows that a W-Cu hybrid system can be an option for further extension of Cu interconnects while suppressing the via resistance.

Fully kinetic simulations of arc discharge in vacuum interrupters (VIs) often suffer from prohibitive computational costs. In this article, a hybrid kinetic-magnetohydrodynamic (MHD) transient simulation of vacuum arc using a field-circuit coupling method is proposed for a vacuum circuit breaker. Compared with the fully kinetic simulation method, the advance of the proposed method originates from the so-called hybrid modeling of ion kinetics and electron MHD, where ions and electrons in a VI are separately treated as macroparticles and massless fluids. The simulation results of the arc are consistent with the arc morphology in the discharge plasma experiment. Furthermore, the proposed field-circuit coupling transient model is used to study the number and the decay time of plasma between the electrodes during the interruption process. Finally, we discussed the dielectric-recovery characteristics and the breakdown probability under different voltage frequencies, voltage levels, and electrode gaps.

该文分析了基于电磁感应加热技术的大型轴承拆卸过程,并结合工程实际,建立了涉及电磁场、温度场和应力应变场的轴承多物理场耦合模型.为了提出一种高精度的轴承电磁热弹耦合模型,根据各个物理场之间的耦合关系确定了序贯耦合方法,并基于数值仿真技术求得了大型轴承涡流场数值解;采用分离变量法和特征值法求得了温度场解析解;采用位移解法求得了热应力应变解析解.与数值仿真和实验结果对比,该文验证了轴承电磁热弹耦合半解析模型的可行生和有效性生.

Plasma transport during continuous extinction processes from high-current arcs to full extinction is an important issue in the study of vacuum arcs. However, to describe this transition, the traditional methods, magneto hydrodynamics or particle-in-cell, are limited by either accuracy or efficiency. In this paper, we developed a fully 3D hybrid plasma simulation algorithm to study the extinction process of multiple cathode spots in a vacuum environment. Cathode spots are modelled as independent plasma sources rather than treating the whole cathode surface as an indivisible source. For a better balance between efficiency and accuracy, ions are modelled as particles and electrons are treated as a massless fluid. Using this model, the effects of cut-off time and the locations of cathode spots on the extinction process are studied. The results show that during an extinction process, the total number of ions between the electrodes gradually decreases and the mean velocity of the ions gradually increases. When the current approaches zero, two regions can be observed in the gap: the separated jet region, and the common plasma channel region. The last three surviving cathode spots have significant impacts on the ion distribution at current zero, whose values depend on the distances to them. The closer together the three cathode spots are, the higher the value that the local ion density will reach. In addition, at current zero, the maximum ion density appears near the last extinguished cathode spot.

典型敷设环境下超高压(extra-high voltage,EHV)交联聚乙烯(crosslinking polyethylene,XLPE)绝缘海底电缆系统的热特性研究对线路、性能优化和电网安全均有重要意义.首先采用有限元法分析了典型敷设环境下海底电缆的热场分布,然后以2条500 kV XLPE海底电缆为试验线路,在敷设环境为直埋、隧道内、穿管内和空气中,持续运行1年,得到了4种典型环境下海底电缆系统沿线的温度分布.研究结果表明:环境温度约为0～10℃时,穿管段的温度最高;环境温度约为10℃时,穿管段或空气段的温度最高且相近;环境温度约为10～40℃时,空气段的温度最高;穿管段和空气段是海底电缆系统实时温度测控的关键点和工程应用的薄弱点.该研究可为海底电缆系统的优化设计和运行维护提供借鉴.

根据矩形截面导线绕制的干式空心电抗器结构特点,该文建立以层等电阻电压,包封等温升和包封等高为约束条件的电抗器优化模型,针对该模型提出等式约束规划问题降维算法,将有等式约束优化问题转换成降维的无等式约束优化问题,使得优化设计变量由原来的2m+n+2个减少到5个.此外,综合分析电抗器原材料成本最小化和运行成本最小化这两个相互冲突的目标函数,提出干式空心电抗器多目标Pareto最优算法.结果表明,优化设计的矩形截面导线绕制电抗器具有结构紧凑、无环流、损耗小及散热效率高等优点.

Discontinuous Galerkin time-domain (DGTD) methods using explicit time integration schemes often suffer from unreasonably small time-step size when applied to simulations involving electrically small structures. In this paper, an exponential time integration (ETI) scheme based on Krylov subspace is proposed for DGTD. The proposed scheme is unconditionally stable and exhibits good accuracy for large time-step sizes. When compared with classic ETI schemes, in the proposed scheme, the dense matrix exponentials are absorbed in matrix–vector products, hence the prohibitive memory costs are avoided. Moreover, a preconditioner combining the shift-and-invert technique and restricted additive Schwarz method is also developed to accelerate the proposed scheme.

The basic behaviors of vacuum arcs are directly affected by the plasma jets ejected from cathode spots. It is of great interest to investigate the features of the plasma jets, especially under magnetic fields. However, for a single jet, the traditional methods, magnetohydrodynamics (MHD) or Particle-in-Cell (PIC), are limited by either accuracy or efficiency. In this paper, we have developed a 3-D parallel hybrid plasma simulation code, which treats ions as particles and electrons as massless fluids, studying the movement of particles and propagation of electromagnetic fields in a self-consistent way. By this approach, the behaviors of a vacuum plasma jet from a single cathode spot under axial magnetic field (AMF) are simulated. It is shown that the spreading angle of the plasma jet decreases gradually when the strength of AMF becomes larger. Moreover, the ion density of the axis near the anode increases. The simulation results are in reasonable agreement with the experimental results.

The tunable localized surface plasmons in novel antenna of Au nanosphere dimer coated by graphene is studied theoretically. We demonstrate the electronic tuning of graphene based Au nanosphere antenna via modifying the Femi level of graphene for realizing active tunable localized surface plasmons. It is found that localized electronic field shows an evident increasing, as the graphene layers increase. The resuts are explained as the more evidently enhanced resonance of localized surface plasmons for multilayer graphene than monolayer graphene nanoantenna when the incident light matches to the resonance wavelength of the Au-graphene hybrid system. In addition, it is revealed there is observable blue-shift for the resonance wavelength when the graphene layers get increased. The study provides basic understanding for tuning graphene based on Au nanosphere antenna for a wide range of applications such as single-molecule fluorescence, SERS and photothermal therapy.

An adaptive cell method is proposed in order to simulate physical fields on a coarse grid with high accuracy. Compared with the conventional cell method, the advance of the proposed adaptive method originates from the so-called node weight correction procedure by using a patch-recovery technique, which corrects the local constitutive relations and in turn mitigates the error source that results from the discretization of constitutive equations. Furthermore, the proposed method has been extended to general curvilinear grids, in which the construction of the local constitutive matrix is derived by using differential geometry. Finally, we have demonstrated by theoretical analysis and numerical results that the proposed strategies can achieve a high-order convergence as well as high-efficiency.