International Journal of Numerical Modelling-Electronic Networks Devices and Fields最新文献

This study investigates interfacial heat transfer and phonon scattering effects in AlGaN/GaN heterostructures using a modified ballistic diffusive equation (BDE) that incorporates a radiative transfer formulation for quasi-ballistic heat flux and integrates multiple phonon scattering models (Callaway, Born–Von Karman, Holland). Finite element simulations reveal that Umklapp scattering markedly reduces the effective thermal conductivity as both temperature increases and layer thickness decreases. A comparative analysis between Al_0.17Ga_0.83N and Al_0.32Ga_0.68N layers highlights distinct thermal conductivity trends, primarily due to enhanced boundary scattering in the higher aluminum-content layer. The temperature distribution exhibits a pronounced peak within the GaN channel, followed by a rapid decline across the Al_0.32Ga_0.68N barrier, emphasizing the localized nature of heat generation. Overall, the results demonstrate that phonon scattering mechanisms and material composition have a decisive role in controlling phonon transport at nanoscale interfaces, providing essential insights to improve thermal reliability of GaN HEMT devices.

This work introduces a highly sensitive and reconfigurable terahertz (THz) biosensor developed for detecting liver cancer biomarkers. The proposed sensor consists of a hybrid graphene-gold THz biosensor patterned on a dielectric substrate, ensuring strong plasmonic confinement and tunable resonance behavior. A machine learning framework employing the K-Nearest Neighbors (KNN) algorithm was implemented to optimize the interaction among graphene's chemical potential, relaxation time, and temperature. Electromagnetic simulations combined with KNN predictions reveal outstanding sensing performance, achieving 76.36 GHz/RIU for normal liver tissue, 130.43 GHz/RIU for AFP, 187.38 GHz/RIU for DCP, and 220.41 GHz/RIU for VEGF. The structure exhibits high quality factors up to 38.85 and a figure of merit of approximately 1.69 RIU⁻¹, confirming its excellent capability to distinguish healthy from malignant tissues. This compact and dynamically tunable design demonstrates strong potential for real-time, label-free THz biomedical diagnostics, particularly in early liver cancer detection.

This study investigates the nonlinear and chaotic dynamics of a modified Hindmarsh–Rose (HR) neuron model through a unified framework based on piecewise fractional differential operators. The classical HR neuron model is widely used to describe the electrical activity of neuronal membranes; however, it does not fully account for memory effects and regime-switching phenomena observed in real neurobiological processes. To address this limitation, we formulate a piecewise dynamical model in which the membrane potential evolves under different fractional operators, including the Caputo, Atangana–Baleanu, and Caputo–Fabrizio derivatives. A fractional order parameter $��\beta �\in �\left(�0�,�1�\right)�$ is introduced to regulate the memory intensity of the system and to construct a generalized fractional-order HR model. The piecewise structure enables the modeling of switching behaviors and crossover effects between distinct neuronal activity regimes. Numerical simulations are carried out to examine the influence of the fractional order and the piecewise operator structure on neuronal firing patterns, chaotic attractors, and complex oscillatory dynamics under external current stimulation. The numerical results clearly show that changes in the fractional order and the choice of switching thresholds have a pronounced influence on the stability properties, oscillation amplitudes, and chaotic behavior of the neuron model. Through a broad set of numerical experiments, the proposed framework is shown to reliably reproduce key features of neuronal dynamics, demonstrating that the adopted numerical schemes are well suited for capturing memory effects and regime switching. Overall, this study advances fractional and piecewise modeling approaches in neuroscience by providing both new theoretical perspectives and practical computational tools for the investigation of complex neuronal systems.

This work introduces a modified Chua circuit, distinguished by a hyperbolic tangent nonlinearity, designed to exhibit complex dynamics pertinent to neuromorphic engineering. The proposed system affords exceptional parametric control over attractor topology, enabling the systematic manipulation of multi-scroll chaotic structures via a single control parameter. Furthermore, the system exhibits bistability through the coexistence of distinct attractors under specific initial conditions. Through a comprehensive bifurcation analysis, we identify and characterize the critical Hopf bifurcation points that govern the transitions between stable, periodic, and chaotic regimes. A principal contribution of this research is the identification of a well-defined, bifurcation-free parameter corridor. Within this zone, the system generates robust, low-frequency oscillations analogous to neural delta rhythms while maintaining sustained chaotic behavior without secondary bifurcations, which ensures highly predictable frequency tuning. The theoretical framework is substantiated by a practical analog circuit implementation, demonstrating excellent fidelity between the mathematical model and its physical realization. A comparative performance analysis reveals that the proposed oscillator possesses superior characteristics including continuous stability windows, comprehensive delta-band coverage, and minimal parameter sensitivity when compared to classical and contemporary designs. These findings establish the proposed circuit as a robust and versatile platform for investigating nonlinear brain dynamics, developing tunable neuromorphic oscillators, and advancing chaos-based technologies.

The growing demand for compact and efficient dual-band antennas in modern wireless and satellite systems motivates the need for designs that offer high gain, low cross-polarization, and broad operational versatility. Based on concentric spiral shape, developed from square iterations, this research proposes a low-profile linearly polarized printed fractal radiator. Both the scaling factor ($��S_f�$) and the rotation angle ($��\theta �$) control the design of the antenna, allowing it to operate in two bands with a fixed aperture. The suggested radiator attains resonance at $��10.50�$ and $��12.50�$ GHz, which correspond to X- and Ku-band frequencies, using a RO-5880 substrate with a dielectric constant 2.2. The simulated and measured findings demonstrate excellent agreement with gain of $��9.67�$ and $��11.28�$ dBi, cross-polarization levels below $��-�37�$ dB, and radiation efficiencies over $��80�\%�$. The proposed small-sized dual-band antennas meet the need for efficient, innovative wireless and satellite systems. These findings confirm that the antenna is suitable for wireless and DBS applications, and they hint at its potential applicability to IoT and 5G with the minimal design up-gradation.

This study investigates the performance and thermal behavior of induction motors used in electric vehicles under different gear ratios. In line with increasing energy efficiency requirements, the dynamic characteristics of the motor during the WLTP Class 1 driving cycle have been analyzed in detail. A four-pole, squirrel-cage induction motor with a rated power of 5 kW was designed, and its electromagnetic performance and loss components have been calculated using ANSYS Motor-CAD software. Thermal analysis revealed that, under S2-60-min duty cycle, the stator winding temperature reached approximately 155°C, and under prolonged operation, the temperature approached insulation limits. In simulations where 10 consecutive WLTP Class 1 cycles have been applied, temperature rise trends and energy consumption data have been evaluated over a total operating period of 2.84 h. To examine the effect of gear ratio on performance, various ratios between 5 and 15 have been tested, and the motor's torque, speed, efficiency, and regenerative energy recovery characteristics have been compared. The results showed that selecting a gear ratio of 7 provided the most favorable operating conditions, with an average efficiency of 88.88%, cycle-averaged instantaneous efficiency of 68.44%, and 60.17 Wh of regenerative energy recovery. Furthermore, it has been determined that high-ratio operating scenarios led to a rapid increase in thermal load, necessitating additional cooling measures.

Reconfigurable reflectarray antenna (RRA) has been considered a potential technology for future communication. However, few studies focus on investigating the intermodulation distortion induced by its active devices, which are employed to provide different compensation phases for the RRA elements. In this paper, the radiation pattern prediction for third-order intermodulation (IM3) distortion of the RRA with embedded varactor diodes is first proposed. It is implemented with a nonlinear model of the varactor and full-wave electromagnetism (FEM) simulation. The presented varactor model is first employed to design the RRA and then to acquire the power and phase of the IM3 signal in each element of the RRA, which is realized with a nonlinear harmonic-balance and FEM co-simulation. With the help of the HFSS software, the radiation pattern is achieved when the co-simulation results of the RRA elements are input. The IM3 radiation measurement of the RRA is presented to verify the radiation pattern prediction of the IM3 signal. The consistent comparison results between the simulation and the measurement indicate that the proposed method is accurate. Significantly, the radiation pattern of the IM3 signal is different from that of the fundamental signal based on the designed RRA. It is meaningful and implies that it may have linearization approaches for the RRA. The proposed method will enhance the application of the RRA since the linearization approach is still a scientific challenge in future wireless networks.

In addressing the pattern synthesis of multiconstraint thinned planar antenna arrays, existing intelligent optimization algorithms encounter limitations such as premature convergence and insufficient solution accuracy. Therefore, we propose an adaptive mutation strategy dung beetle optimizer (AMSDBO) for optimizing thinned planar antenna arrays. The AMSDBO algorithm innovatively constructs a three-stage collaborative optimization framework, effectively achieving high-performance design for thinned planar arrays under the constraints of fixed aperture size and sparsity rate through a parameter adaptive adjustment mechanism. Firstly, initial solutions for the dung beetle populations are generated using a chaotic mapping reverse learning (CMRL) strategy to enhance both population diversity and the quality of the initial solutions. Next, the local mutation search (LMS) mechanism is introduced. An adaptive T-distribution mechanism is imposed to effectively enhance the early global search capability. Then, the Lévy flight variation strategy is employed to adaptively adjust the positions of the dung beetle population in the later stages, helping the algorithm avoid local optima and accelerating convergence. Simulation results with classical test functions demonstrate that AMSDBO offers significant advantages in convergence accuracy and robustness compared to traditional algorithms (DBO, PSO, and IWO). Additionally, two sets of typical planar thinned array experimental results indicate that the optimization performance of the AMSDBO algorithm is significantly improved compared to traditional optimization algorithms. Specifically, there is a peak sidelobe level (PSLL) reduction of 15.5% for DBO, 11.64% for PSO, and 14.56% for IWO. This confirms the effectiveness and superiority of the AMSDBO algorithm.