Conventional numerical heat conduction solvers are exceedingly computationally expensive and memory demanding. Recent advances in deep learning have witnessed its extensive applications in computational physics field. Compared to traditional methods, deep learning techniques emerge superior computational efficiency, providing a substitute for speeding up the calculation. In this paper, we propose an innovate deep learning framework to predict the 3D temperature field in a cube region filled with random objects of various geometries and materials. The framework is capable of resolving both passive and active heat conduction problems. After being fully trained, it can achieve a similar precision compared to the finite element method (FEM), while the calculation speed is accelerated by about 280 times. Furthermore, the deep learning model has demonstrated robust generalization ability in predicting real-world objects not existing in the data set. We believe that the framework paves the way for solving complex heat conduction problems in engineering, as well as inverse problems in the future.

For wireless communication systems, receivers usually work under complex electromagnetic environments and are often susceptible to electromagnetic interference (EMI). With the wide application of pulse signals in various fields, the impact of pulse signals on the receivers of communication system has not been extensively studied. On the one hand, the existing receiver electromagnetic sensitivity (EMS) interference coupling effect is mainly analyzed from the perspective of energy only, without considering of different electromagnetic parameter characteristics of pulse signal, such as pulse width, repetition period, duty cycle and so on. On the other hand, there is a lack of quantitative characterization of typical performance indexes of receiver subject to pulsed interference environment, resulting in insufficient reliability and accuracy of receiver interference degree prediction and electromagnetic compatibility (EMC) evaluation. This paper focuses on the EMS interference coupling effect analysis and EMC evaluation method of receiver under pulsed interference environment. First, based on the analysis of the interference mechanism of the pulse signal on receivers, the formula for the bit error ratio (BER) is derived. Then a system model is proposed to verify the theoretical analysis results through numerical simulation. With the established relationship between the parameters of the pulsed interference and the BER performance of the receiver, a novel EMC evaluation method has been proposed. As a practical application example, the evaluation method is verified via a well-designed experiment on BeiDou Navigation Satellite System (BDS). The experiment shows that the observed phenomena are in good agreement with the conclusions of the proposed evaluation method, proving that the method is applicable to the EMC evaluation of receiver under pulsed signal interference environment.

In this letter, an attention-based residual network (ResNet) is proposed to predict the radiation pattern of phased array antenna (PAA). The proposed model consists of two input and preprocessing modules, a ResNet-attention module and a multi-layer perception (MLP) module. It enables simultaneous prediction of the radiation patterns for PAAs with multiple arrangements and frequencies, which supports the design of PAA and ensures the optimal performance of the equipment in the same electromagnetic environment (EME). Various antenna arrays are utilized to verify the effectiveness and superiority of the proposed model. Discussions also encompass the model's performance under a large array and its generalization capabilities.

The finite-difference time-domain (FDTD) method based on nonuniform face-centered cubic (FCC) grids, termed as NU-FCC-FDTD method, is proposed in this article. The FCC-FDTD method with uniform grids has a smaller numerical dispersion error than the traditional FDTD methods. To further enhance its efficiency and scalability, the NU-FCC-FDTD method is proposed to solve complex electromagnetic problems. Unlike the traditional FDTD methods based on Yee’s grids, for which the nonuniform grid technique is relatively straightforward, instability may occur when trying to apply nonuniform grids into the FCC-FDTD method. To ensure stability, cell-size-based correction factors are carefully defined in the time-marching formulations after taking cell sizes into consideration. Through comprehensive numerical dispersion analysis, the NU-FCC-FDTD method shows numerical dispersion advantage over the FDTD methods and the FCC-FDTD method with uniform grids. Furthermore, the stability condition is also given by using the parameter scanning method and verified in numerical examples. Four examples, including electromagnetic wave propagation, a dielectric rod (DR) resonator, a composite metallic-dielectric radome and the electromagnetic energy absorption of human head, are carried out to demonstrate its performance. Results show that the proposed NU-FCC-FDTD method can significantly improve efficiency compared with that of the FCC-FDTD method without compromising accuracy.

The widespread deployment of multiple radio-frequency (RF) devices on complex platforms poses critical challenges for spectrum coexistence. Excessive radiated emissions and insufficient radiated susceptibility can cause intolerable interference, whereas traditional electromagnetic compatibility (EMC) design typically enforces uniform limits across all devices. Such uniformity neglects system heterogeneity, resulting in constraints with excessive interference margins or compromised system-level robustness. To address this issue, we propose a Joint Tailoring of Radiated Emission and Susceptibility Limits (JTRESL) framework. Explicitly accounting for the physical link between emission and immunity, a radiation coupling network model is first developed to characterize electromagnetic interference among devices. Based on this model, a multiobjective optimization formulation is established to balance interference immunity, implementation cost, and spectral smoothness, while incorporating engineering constraints. The resulting nonconvex problem is solved using an iterative algorithm based on sequential quadratic programming, which updates tailoring variables frequency-by-frequency and jointly optimizes emission limits and susceptibility thresholds. Experiments in a multidevice anechoic chamber demonstrate that the proposed method effectively eliminates interference states, controls engineering cost, and produces smooth limit curves. This enables targeted tightening in high-risk bands and appropriate relaxation in overly conservative bands. Overall, JTRESL provides an efficient and feasible solution to enhance EMC and spectrum coexistence in heterogeneous multiple RF platforms.

Based on the auxiliary differential equation (ADE), a performance-enhanced convolution perfectly matched layer (CPML) for the face-centered cubic finite-difference time-domain (FCC-FDTD) method is developed in this letter. Unlike the traditional FDTD methods on Yee’s grids, for which the CPML technique is relatively straightforward to be implemented, poor performance, and even instability may occur in the FCC-FDTD method. To avoid these problems, the ADE technique is incorporated into the derivation, and the impact of node locations on absorbing boundary constitutive parameters is carefully considered. In addition, the derivation of the time-marching formulations is substantially simplified. The optimal ranges of constitutive parameters are found through a numerical sweep. Two numerical examples, the electromagnetic pulse propagation and absorbed power density of a brain model during mobile communication, are examined to validate the performance. The results suggest that the proposed ADE-CPML for the FCC-FDTD method shows good stability and accuracy.