Can Cui

To address the issue of iterative Fourier transform method (IFT) easily falling into local optima in thinned array synthesis, this communication proposes a two-stage alternating iterative Fourier transform (TSAIFT) approach for large thinned planar circular array synthesis, which involves an optimization stage utilizing a modified iterative Fourier transform density tapering (IFTDT), and a relaxation stage employing the IFT with constrained amplitude range. Specifically, the dynamic reference excitation mechanism is proposed to enhance the global exploration ability, which allows the number of active array elements in each ring to be iteratively adjusted during the optimization stage. In the relaxation stage, the excitation amplitudes are no longer constrained to be 0 or 1. Instead, a relaxation factor is introduced to gradually accumulate the small variations in the excitation coefficients. When the optimization stage gets trapped in local optima, it transitions to the relaxation stage to escape from the local optima. The two stages alternate to approach the optimal solution gradually. Due to the adoption of the low-computational IFT for both optimization and escaping from local optima, the proposed TSAIFT is capable of achieving competitive optimization results within a relatively short period of time, even when dealing with large-scale arrays. Several typical thinned planar circular arrays formed by isotropic element and microstrip patch element are synthesized, and the experimental results demonstrate the effectiveness of the proposed approach in optimization performance and computational efficiency.

In this letter, a method based on an encoder-decoder fully convolutional network is proposed for the efficient synthesis of planar arrays fulfilling user-defined beampattern masks. The decoder introduces a novel sub-pixel convolutional layer to yield an accurate and fast upsampling regression from the array excitations to the radiated pattern. The encoder is trained to minimize the deviation between the desired shaped beampattern and the actual one. By collaborating with the pre-trained decoder, an efficient training of the encoder can be realized to yield a set of array excitations that fits the design objectives. Representative numerical results are reported to assess the effectiveness and efficiency of the proposed method.

In this letter, a novel static-dynamic convex optimization algorithm is proposed for array synthesis, which can be divided into static and dynamic successive optimization stages. At the first stage, the synthetic direction graph is constructed as a second-order cone programming problem so that an initial excitation is quickly obtained. Next, by serving the results obtained previously as the initial excitation, an iterative operation is implemented to obtain a high-precise solution. Moreover, in order to effectively ensure the continuity of the solution process and approach the feasible solution, shrinkage factor and crossover operator are introduced. Four tough examples of synthesizing different kinds of patterns are conducted to validate the feasibility and effectiveness of the proposed method. Results show that compared with the existing representative methods, the proposed algorithm is of high computational efficiency and universality.

The tremendous progress in artificial intelligence promotes the wide application of machine learning (ML) technology in the field of electronic science. Recently ML-based antenna optimization provides a distinct candidate and attracts considerable attention. However, the large number of training samples generated through time-consuming EM simulations becomes a significant challenge. In this article, an efficient online data-driven enhanced-XGBoost (E-XGBoost) method for antenna optimization is proposed, which is mainly composed of two parts, i.e., an input variable filter module (IVFM) and an antenna optimization module (AOM). Specifically, IVFM serves as a variable sensitivity analyzer, which is accomplished by E-XGBoost to efficiently reduce the dimension of design variable and hence save the training samples. Next, the design variables obtained by IVFM are fed into AOM to find the near-optimal solution. In AOM, an online learning strategy is proposed to train a local E-XGBoost model to evaluate the population in the metaheuristic optimization algorithm (MOA). Compared to the global ML model that can mimic the entire design space, this local E-XGBoost model can further cut down the training samples. To verify the performance of the proposed method, several different antenna examples, i.e., U-slot patch antenna, Fabry–Perot resonant antenna, and dual-polarized cross dipole antenna and 5G MIMO antenna array, are simulated. Numerical results support the proposed method in terms of its superior performance and potential advantage of saving computational overhead.

In this article, a novel multibranch encoder-decoder-based artificial neural network (ANN) framework is proposed for pattern-reconfigurable nonuniformly spaced linear array synthesis. In the proposed framework, different encoder–decoder branches are devoted to synthesizing different radiation patterns to satisfy the desired radiation patterns. In each encoder–decoder branch, the encoder, and the decoder serve as the array synthesizer and analyzer, respectively. By minimizing a suitably-defined loss function with respect to multiple radiation patterns, element amplitudes, and positions, not only multiple mask-constrained radiation patterns can be successfully achieved, but also the common element amplitudes and positions with minimum inter-element spacing control can be obtained. Thus, different radiation patterns can be switched by phase-only control. In addition, by using different training samples, the proposed method can consider the ideal synthesis of the array factor, as well as the actual synthesis of antenna arrays to consider the mutual coupling effects in the synthesis process. Simulation results are provided to validate the capability, efficiency, and attractiveness of the proposed method.

The beampattern synthesis of antenna array is always known as a computationally cost task that needs efficient strategies to deal with. In this article, a novel artificial neural network (ANN)-based array synthesis method is proposed. By establishing an encoder–decoder-based ANN framework, the mask-constrained beampattern in terms of focused or shaped beam for linear arrays with arbitrary given array geometry is successfully synthesized. More in detail, the encoder serves as an array synthesizer, while the decoder behaves as an array analyzer. Thanks to the designed pretrained decoder, such an approach is computationally efficient so that real-time array synthesis can be potentially achieved. Moreover, the proposed method allows one to consider both ideal and actual linear arrays with mutual coupling effects and nonidealities. The results of a wide numerical validation involving amplitude-only, phase-only, and amplitude–phase excitation syntheses are presented to assess the flexibility and the versatility of the proposed method also in comparison to the competitive state-of-the-art synthesis techniques.

Frequency diverse array (FDA), by virtue of its range‐angle‐dependent beam pattern, has drawn a substantial attention in recent years. However, as far as the time‐variant range‐angle‐coupled beam pattern is concerned, such inherent flaw has become a bottleneck of its further development. To address this issue, herein, a logarithm‐based optimised static non‐linear frequency offset (LOSNFO) for FDA is proposed, which can successfully alleviate its inherent flaw, thus producing a short‐time range‐angle‐decoupled beam pattern with low sidelobe levels (SLLs) and narrow half‐power beam widths (HPBWs). Moreover, for practical applications, the proposed LOSNFO considers the propagation effect, which is always neglected in its counterpart. Numerical results are reported to demonstrate the effectiveness and superiority of the proposed LOSNFO‐FDA.

Pipeline is important for gas transportation. Free span of buried pipeline may cause safety problem as a result of decreasing the load-carrying capacity of the pipeline. A method for detecting the free span of buried pipelines based on the inner pipe excitation signal was proposed. Vibration characteristics analysis of the pipe and its surrounding coupling system were discussed by means of the finite element method. A three-dimensional pipe model containing a free-spanning segment with a certain length was established. Transient dynamic response analysis was adopted. Frequency response function plots were generated to analyze the results of simulation. The effects of soil type, the length of the free-spanning segment, and the excitation signal on detection were studied. The results show that (1) the obvious change of the first natural frequency of the pipe with different surroundings can be used to detect the free-spanning segment of pipeline; (2) the harder the soil medium is, the higher the first natural frequency of the pipe and soil surrounding system will be; (3) the longer the length of the free-spanning segment is, the lower the first natural frequency will be.