This paper investigates the feasibility of novel data-driven deep learning (DL) models to predict solution Maxwell's equations for low-frequency electromagnetic (EM) devices. With ground truth (empirical evidence) data being generated from a finite-element analysis solver, convolutional neural network is trained in supervised manner learn mapping magnetic field distribution topologies different complexities geometry, material, and excitation, including simple coil, transformer, permanent...

Deep learning has achieved remarkable success in diverse applications; however, its use solving partial differential equations (PDEs) emerged only recently. Here, we present a feasibility study of applying physics-informed deep methods for PDEs related to the physical laws electromagnetics. The methodology uses automatic differentiation, and loss function is formulated based on underlying PDE boundary conditions. method shown using three electromagnetic problems varying complexity results...

In this article, a new method for predicting efficiency maps of electric motor drives is proposed using deep learning (DL). Since many operating points need to be simulated finite-element (FE) analysis estimate the map single drive topology with certain geometry dimensions and materials, incorporating whole into design optimization process an overwhelmingly time-consuming task may impossible, depending on availability computational resources. Therefore, two DL network architectures are...

In this article, a method for topology optimization (TO) of synchronous reluctance motor (SynRM) is proposed using deep reinforcement learning (RL). Due to the need simulating large number finite-element models in traditional TO task, incorporating study involving different problem formulation (such as varying design domain) can be an overwhelming task. A neural network (NN)-based agent trained RL able extend knowledge from one other similar tasks. The applicability such performed...

Accurately modeling magnetic hysteresis plays a crucial role in developing precise digital twins for low-frequency electromagnetic systems. However, large 3-D analysis systems, the evaluation of performance at hundreds thousands points components containing steels is challenging. It utmost importance that any system can assess within shortest possible timeframe. The utilization neural networks (NNs) offers potential to achieve this objective. This article provides comprehensive review...

Abstract The diversified ocular disorders in which cystoid macular edema (CME) occurs, are strongly associated with the vision loss. Optical coherence tomography (OCT) scans that allow screening of retina, contain artifacts including blur‐edges, speckle noise, and so forth, create difficulty identifying retinal fluid. In this work, major image preprocessing techniques such as minimum filtering, block‐matching 3D Richardson–Lucy deconvolution method applied to minimize noise other degradation...

Designing an electrical machine which is part of a system, such as electric vehicle (EV) powertrain requires considerable performance information early in the process. Often, this expensive to acquire using numerical models and existing analytical approaches are insufficient. In paper, surrogate model based around Machine Learning (ML) approach proposed for fast estimation device during exploration design space find initial candidate design.

This The effective representation of material properties is fundamental to the simulation electromagnetic devices such as electrical machines, actuators, sensors, transformers, etc. However, actual operating point a dependent both on position within device and excitation. Every in an machine can be different part magnetization curve. To determine performance parameters efficiency machine, hysteretic behavior crucial, used impact code. In this paper, use deep learning methods proposed reduce...

This article proposes a new topology optimization (TO) method which transforms the material distribution problem into movement sequence search for controller moving in design space. enforces connectivity between cells discretized domain that contains same material. removes need any filtering or smoothing of optimal result to obtain manufacturable design. The validity this is proved by employing genetic algorithm on C-core electromagnetic actuator and synchronous reluctance motor problem....

Purpose The purpose of this paper is to develop surrogate models, using deep learning (DL), that can facilitate the application EM analysis software. In current status quo, electrical systems be found in an ever-increasing range products are part everyone’s daily live. With advances technology, industries such as automotive, communications and medical devices have been disrupted with new electronic systems. innovation development increasing complexity over time has supported by increased use...

Electrophoregrams generated after the electrophoresis experiment can be evaluated by eye, using simple measurement tools. A basic strip or spot densitometer and a ruler provide data that are adequate for some purposes. More detailed analysis requires more sophisticated We may need to know molecular weights proportional purities abundances of various compounds distributed across lane. In these sorts quantitative applications, image offers many advantages.In this paper we present computerized...

This paper presents the study of dynamic electrical resistance electrodeposition cell during growth metallic dendrites showing fractal character. The electric circular is measured in real time using a computer based data acquisition system. system constructed capable measuring voltage and current through under program control at pre-decided intervals. allows for measurement cell. on standard analogue to digital controller ADC interfaced printer port.

This paper explores methods to extend a trained deep neural network for predicting efficiency maps work on different motor drive topologies. procedure reduces the computation cost associated with training networks by transferring knowledge over similar tasks handled networks. Two types of synchronous AC machines, including flat-type interior and surface-mounted permanent magnet are tested their entire torque-speed profiles validate applicability proposed methodology. The obtained results...

Over the past couple of decades, dependence on magnetic materials has increased tremendously. Since every kind material behaves differently, it is very difficult to create a universal model these behaviors. While there exist models and simulation tools that can represent behavior materials, may take days or even an entire week compute when embedded in analysis system, which highly inefficient. With good machine learning (ML) model, computation time be reduced by significant amount with...

The best individual stellar parameters of the close visual binary system "HIP 57894" using synthetic photometric solution based on Al-Wardat's complex method are presented. match between and observed photometry is presented entire spectral energy distributions which constructed by utilizing Atlas9 model atmospheres two special subroutines method. From solution, we determine masses radii as: MA = 1.22 ± 0.18 M⊙, RA 1.328 0.04 R⊙ MB 0.99 0.14 RB 0.975 0.03 for primary secondary components...

The study of the viscous fingering in Hele Shaw cell and evolution pattern was presented growth velocity determined. fingers were recorded using a digital web camera. movie frames separated selected patterns analyzed. box counting technique applied to Richardson plot used for characterization shapes terms structure texture. thus obtained different analyzed structural textural analysis presented. scale invariance at length scales discussed.

Progress in modern industry requires the improvement of "classical" materials as well development new "advanced" materials.In great majority these have a polycrystalline structure, i.e. they consist number crystallites, properties which are direction dependent (anisotropic).Hence, their macroscopic depend on orientation distribution--the