We present an efficient wideband multilevel fast adaptive cross approximation-characteristic basis function method (Wideband MLFACA-CBFM) for wideband electromagnetic scattering problems. In this method, the multilevel fast adaptive cross approximation (MLFACA) of the impedance matrices and the generation of the CBFs are performed only once at the highest frequency within the given band. The resulting decomposition and compression information is reused across the entire band, eliminating the need to repeatedly perform the time-consuming MLFACA decomposition and CBF generation at each analyzed frequency. Moreover, impedance interpolation is employed to efficiently construct the approximate decomposed impedance matrices at the analyzed frequencies. As a result, the proposed method effectively reduces the time required to compute the wideband radar cross section.

Efficient computation of multiangle monostatic radar cross section (RCS) using the method of moments (MoM) is hindered by the need to restart the iterative solution process for each incident angle due to the changing excitation. To mitigate this burden, this letter proposes an autoregressive (AR) model-based method for predicting improved initial guesses to accelerate convergence. The AR model parameters are estimated using the Burg algorithm. Predicting each initial guess incurs only O(N) computational complexity, which is lower than the O(N-2) cost of a single iteration in the iterative solver. Compared with those from the conventional phase-corrected initial guess method, the initial guesses predicted by the proposed method are closer to the converged solution. The proposed method effectively reduces the number of iterations, particularly under stringent convergence criteria. Numerical results demonstrate the effectiveness and efficiency of the proposed method.

A new method that utilizes multibranch Rao-Wilton-Glisson (MB-RWG) functions and the electric and magnetic current combined field integral equations is proposed for analyzing piecewise homogeneous dielectric targets of arbitrary shapes with junctions. The proposed method allows for the independent discretization of the outer surface of the object and the interfaces of different materials within it, thus enabling the generation of quasi-nonconformal meshes. This provides superior flexibility during preprocessing, especially when handling interfaces connected to the junctions. At the junctions, where more than two material regions meet, there are quasi-nonconformal boundaries between the meshes, and these quasi-nonconformal meshes are then connected in the form of MB-RWG functions. Due to the natural, normal continuity of MB-RWG, no continuity requirements or special treatments need to be imposed at these junctions. Numerical experiments are given to demonstrate the feasibility and versatility of the proposed method.

A conformal discretization for magnetic field integral equation (MFIE) based on the characteristic basis function (CBF) using only RWG basis functions is proposed. The proposed method achieves conformal discretization and well-conditioned system formulation of MFIE, improving its accuracy and maintaining satisfactory iterative convergence. Moreover, the CBF-based nature enables substantial reduction in the number of unknowns.

The dual testing scheme using the Buffa-Christiansen (BC) basis function not only maintains the good convergence of the magnetic field integral equation (MFIE) but also improves its accuracy. However, in practical applications, fewer unknowns are desired to reduce the computing resources of the nBC-tested MFIE. In this letter, a novel conforming discretization based on characteristic basis functions (CBFs) is proposed to mitigate the problem of the conventional nBC-tested MFIE. The proposed discretization can acquire a well-conditioned discretization of the identity operator leading to a satisfactory iterative convergence while keeping a conforming testing scheme. The resulting system matrix is smaller than the original system matrix of the conventional nBC-tested MFIE while inheriting satisfactory iterative convergence and accuracy. Numerical results are shown to demonstrate the performance of the proposed method.

It is widely agreed that the infrared stealth should be considered in the preliminary design phase of turbofan engine so as to decrease the infrared radiation of engine exhaust system. In the present study, a turbofan engine with afterburner at small bypass ratio (0.44, 0.5) in the flight envelope (altitude h < 12.8 km, Mach number M < 1.5) was studied, in which afterburner state was not taken into account. The design method of geometrical model of the exhaust system at the preliminary design phase was developed, and the Latin hypercube sampling method was used to design sample cases, which were numerically simulated to study the infrared radiation characteristics of the exhaust system at different inlet parameters of the exhaust system. A neural network model was developed to predict the infrared radiation of the exhaust system, in which the inlet parameters of the exhaust system in the flight envelope were applied as input parameters. Results show that the maximum error of the developed neural network model for predicting infrared radiation characteristics is less than 10 %, which can be used to evaluate the infrared radiation characteristics of the exhaust system at the preliminary design phase of turbofan engine.

A novel Shooting and Bouncing Ray (SBR) method, suitable for thin dielectric layers on Perfect Electric Conductors (PEC) surfaces and lossy dielectric materials, is proposed. This method is based on the propagation characteristics of nonuniform plane waves at the interface between two lossy media. The accuracy of the proposed SBR is verified by comparing the results with those obtained from the commercial software.

Volume equivalent shooting and bouncing rays (VESBR) overcomes the limitation of conventional high-frequency algorithms that usually ignore the internal polarization effect when precisely analyzing the electromagnetic scattering of pure mediums and thick-coated metal targets. In this paper, VESBR incorporated with the theory of thin medium coating of metals is proposed to process the electromagnetic scattering problem of complex structures containing the thick mediums and the thin coatings. Numerical simulation verifies the effectiveness of the proposed method.

This paper proposes a ResNet-assisted Characteristic Basis Function Method (CBFM) for fast electromagnetic scattering prediction of cylindrical conformal finite periodic arrays. To address the high computational cost of solving characteristic basis function (CBF) expansion coefficients in large-scale conformal structures, two separate residual neural networks are constructed to predict the magnitude and phase of complex CBF coefficients, respectively. The predicted coefficients are then used to compute radar cross sections (RCS). Numerical results demonstrate that the proposed method achieves high accuracy and reduces computational time compared with conventional CBFM, offering an efficient solution for conformal array analysis.

Using only the RWG functions, the combined source integral equation (CSIE) with weak form combined source condition can achieve fine accuracy and fast iterative convergence for conductor objects. However, compared with a conventional integral equation in the method of moments (MoM), the conventional CSIE involves more matrices and more complex numerical processing, and these make the CSIE inefficient, especially for multiple excitation problems. In this article, a characteristic basis function (CBF)-based CSIE with initial guess is proposed to mitigate this problem. The CBF is employed to reduce the number of unknowns as well as the storage consumptions and iteration time. In the meantime, an initial guess especially for CBFs is proposed to reduce iterations when solving multiple excitation problems. Numerical results are given to demonstrate the performance of the proposed method.

In this article, the artificial neural networks (ANNs) are combined with the characteristic basis function method (CBFM) to fast predict the electromagnetic scattering from finite periodic structures. The amplitude and the phase of the expansion coefficients of the characteristic basis functions (CBFs) are predicted by two different ANNs. Then, the predicted expansion coefficients are used to calculated the radar cross sections (RCSs) of the finite periodic structures. Numerical results are given to show the performance of the ANN-assisted CBFM.

In this letter, an efficient solution of the wideband partial modification problem involving an original structure with small geometric modifications is proposed. Compared with the conventional partial modification analysis method, the proposed method is more efficient in two aspects: first, the inverse of the impedance matrix of the original structure is solved implicitly in a compressed form. Based on the compressed inverses, all the subsequent modification problems can be quickly solved with the aid of the Sherman-Morrison-Woodbury formula and the partitioned-inverse formula each time the modification takes place. Second, solutions of the original structure at different frequencies are related by the adaptive cross approximation and thus can be quickly solved. Numerical examples are performed to demonstrate the accuracy and efficiency of the proposed method.

Frequency-polarization multiplexing metasurfaces (FPMMSs) capable of generating vortex beams carrying orbital angular momentum (OAM) have facilitated the actualization of high information capacity in optical and microwave communication. However, poor frequency controllability and the deficiency of theoretical methodology for frequency modulation of meta-atom remain challenges for existing FPMMSs with orthogonal linear polarization. In this work, a strategy to customize the operating band of the “I”-shaped meta-atom (ISMA) based on phase analysis of the equivalent circuit is proposed. A frequency modulation factor is introduced to adjust the operating band of the ISMA over a wide range based on the relationships between element geometry parameters, capacitance and inductance values in the equivalent circuit model, and the reflected phase of ISMA-type meta-atom. Then, a dual-band dual-polarized meta-atom is proposed by placing subunits operating at different bands cross-orthogonally, allowing the subunits to modulate the phase independently. A dual-band dual-polarized dual-mode OAM beam generator is designed and constructed as a proof-of-concept to verify the methodology. The measured results are in excellent agreement with the simulations. The proposed methodology for customizing frequencies establishes the groundwork for the implementation of OAM-based FPMMSs for secure and high-capacity communication in 6G massive-MIMO systems.

In characteristic mode analysis (CMA), characteristic modes (CMs) on wideband frequency need to be solved for analyzing physical characteristics. This paper proposes an efficient method to obtain wideband CMs by solving the generalized eigenvalue problems (GEPs). In this method, the wideband Sherman-Morrison-Woodbury algorithm (WSMWA) is embedded into implicitly restarted Arnoldi method (IRAM). Numerical results show that the proposed method can save much CPU time.

To achieve the infrared/radar integrated stealth of a helicopter,the integrated infrared suppressor was incorporated into the design of stealth helicopter.Through numerical simulation,the temperature field,infrared radiation characteristics and radar scattering characteristics of stealth hel-icopter in hover were calculated.Results showed that under the combined effect of the ejected flow of integrated infrared suppressor and downwash airflow,the mixing duct and rear fuselage of stealth helicopter could be effectively cooled.Under the influence of tail rotor airflow,the local heating effect of the high-temperature exhaust on left fuselage was more significant.Due to the integrated design scheme of the exhaust system and rear fuselage,the latter could effectively shield the high-temperature mixing duct,therefore,total infrared radiation in 3～5 μm band was mainly from gas and fuselage skin.However,for 8～14 μm band,gas infrared radiation could be neglected.At 3 GHz,6 GHz,10 GHz and 15 GHz frequencies,the RCS mean of the stealth helicopter throughout the entire circumference was controlled below 4 dBsm,which indicated that the integrated design scheme could not only reduce infrared radiation but also have good radar stealth capability.

准确高效的电磁散射仿真方法对设计隐身航空发动机排气系统非常重要.将特征基函数法(CBFM)、多层快速多极子算法(MLFMA)、插值分解(ID)算法和并行技术相结合,对发动机排气系统的单站雷达散射截面(RCS)进行仿真计算.插值分解(ID)可以对单站激励矩阵进行低秩压缩,因此可以减少矩阵方程的求解次数,显著提高了传统CBFM-MLFMA的计算效率.为了验证算法的正确性,对轴对称排气系统模型进行了加工和RCS试验测试.与测试结果相比,仿真结果与其吻合良好,验证了算法的精度.

An interpolation method for the reduced matrix in the wideband characteristic mode basis function method (WCMBFM) with the wideband adaptive cross approximation (WACA) algorithm is proposed to efficiently compute the wideband electromagnetic scattering problems. In this method, the characteristic mode basis functions (CMBFs) and the selected dominant basis functions of the far-field block pairs obtained by the ACA decomposition are extracted at the highest frequency within the band of interest. Then, the approximate compression of the reduced matrix at other required frequencies within the band of interest can be efficiently generated by the polynomial interpolation. Compared with the conventional interpolation method for the reduced matrix, the proposed method reduces the dimension of the matrix to be interpolated, which makes the construction of the reduced matrix more efficient. The numerical result of the frequency selective surface (FSS) is investigated to show the advantages of the proposed method.

The efficient broadband monostatic radar cross section (RCS) of the electrically large and deep morphing S-shape cavity calculation is a challenge in computational electromagnetic. Moreover, the traditional electromagnetic computing method can be very time-consuming because the highly complex computations of broadband monostatic RCS must be repeated each time the S-shape cavity structure is modified. To tackle these difficulties, artificial neural networks (ANNs) are proposed to reduce the computational cost of conventional methods, and the data-driven ANNs model can be trained with the inputs scheme datasets include three parts, which are the morphing cavity structure parameters, the incident plane wave information, and RCS of the initial morphing cavity. With careful design and proper training, the broadband monostatic RCS of morphing S-shape cavity can be obtained efficiently by ANNs. In the last, three examples are presented to demonstrate the accuracy and efficiency of the proposed method.

Performing 3D dense captioning and visual grounding requires a common and shared understanding of the underlying multimodal relationships. However, despite some previous attempts on connecting these two related tasks with highly task-specific neural modules, it remains understudied how to explicitly depict their shared nature to learn them simultaneously. In this work, we propose UniT3D, a simple yet effective fully unified transformer-based architecture for jointly solving 3D visual grounding and dense captioning. UniT3D enables learning a strong multimodal representation across the two tasks through a supervised joint pre-training scheme with bidirectional and seq-to-seq objectives. With a generic architecture design, UniT3D allows expanding the pre-training scope to more various training sources such as the synthesized data from 2D prior knowledge to benefit 3D vision-language tasks. Extensive experiments and analysis demonstrate that UniT3D obtains significant gains for 3D dense captioning and visual grounding.

A multibranch curvilinear Rao–Wilton–Glisson (MB-CRWG) basis function is proposed to solve surface integral equations with nonconformal curvilinear meshes. The derivation of the MB-CRWG basis function is given. The structure of the MB-CRWG basis function can be divided into positive and negative parts. The positive part is one curvilinear triangular patch, and the negative part is composed of several curvilinear triangular patches. The MB-CRWG can conveniently connect two surfaces discretized with curvilinear triangular patches of different sizes. Numerical examples of the electromagnetic scattering from multiscale perfect electrically conducting targets are given to demonstrate the effectiveness of the proposed basis function.