Tree roots are vital for tree ecosystems; accurate root detection helps analyze the health of trees and supports the effective management of resources such as fertilizers, water and pesticides. In this paper, a deep learning-based ground-penetrating radar (GPR) inversion method is proposed to simultaneously image the spatial distribution of permittivity for subsurface tree roots and layered heterogeneous soils in real time. Additionally, a GPR simulation data set and a measured data set are built in this study, which were used to train inversion models and validate the effectiveness of GPR inversion methods.The introduced GPR inversion model is a pyramid convolutional network with vision transformer and edge inversion auxiliary task (PyViTENet), which combines pyramidal convolution and vision transformer to improve the diversity and accuracy of data feature extraction. Furthermore, by adding the task of edge inversion of the permittivity distribution of underground materials, the model focuses more on the details of heterogeneous structures. The experimental results show that, for the case of buried scatterers in layered heterogeneous soil, the PyViTENet performs better than other deep learning methods on the simulation data set. It can more accurately invert the permittivity of scatterers and the soil stratification. The most notable advantage of PyViTENet is that it can accurately capture the heterogeneous structural details of the soil within the layer since the soil around the tree roots in the real scene is layered soil and each layer of soil is also heterogeneous due to factors such as humidity, proportion of different soil particles, etc.In order to further verify the effectiveness of the proposed inversion method, this study applied the PyViTENet to GPR measured data through transfer learning for reconstructing the permittivity, shape, and position information of scatterers in the actual scene. The proposed model shows good generalization ability and accuracy, and provides a basis for non-destructive detection of underground scatterers and their surrounding medium.

This article presents a new unconditionally stable finite-difference time-domain (FDTD) method combined with the artificial neural network (ANN) for multiscale problems. The input of ANN-FDTD is the point location of spatial grid division in FDTD, and the output is the field data at these points. At each time step, the hypothetical solution of Maxwell’s equations is constructed using a known forced excitation source and the output of ANN. The error of ANN is determined by the gradient of its output with respect to the input vector. This error value is used to update ANN parameters with the backpropagation (BP) algorithm, so the training does not have to involve labeled samples. Here, ANN is trained to ensure that the hypothetical solution satisfies the boundary conditions. ANN is trained individually at each time-marching step, and thus, the results of the previous step do not affect those of the next step. The time step of ANN-FDTD can be chosen much larger than that of the traditional FDTD for microwave components with fine structures. Moreover, all time steps can be divided into blocks to execute parallel computation. Three numerical examples are provided to validate the accuracy and efficiency of the proposed method.

In this paper, an adiabatic operation slope-loss algorithm (AO-SLA) is developed for designing efficient adiabatic waveguide tapers. The algorithm can compute the efficient adiabatic guided-wave shape of the waveguide taper. A well-studied case of an adiabatic waveguide taper is used to show how it can readily generate a guided-wave shape that gives the same mode-connection power transfer with a much shorter length compared with the linear shape. In the waveguide taper design, it is shown that the device efficiency designed with AO-SLA is better than that based on the parabolic case and the SLA case, and the adiabatic taper designed can give a broad operating bandwidth. Moreover, the adiabatic condition α for AO-SLA can be very large.

This paper proposes a fully numerical method for designing adiabatic mode converters (NM-AMCs). The proposed method can help compute efficient adiabatic guided-wave shapes for mode converters using a nonuniform section division technique. Such a scheme reduces the memory requirement and computational time for the design of adiabatic devices. Furthermore, the adiabatic mode converter comprises a ridge waveguide, which is difficult to design using analytical methods. This complex case of an adiabatic mode converter is used to demonstrate how the proposed method can readily generate a guided-wave shape that gives the same mode-connection power transfer with a far shorter length compared with a linear-shape converter. The efficiency of the device designed using this NM-AMC is significantly better than that of linear-shape converters; this designed adiabatic mode converter can enable a broad operating bandwidth. Moreover, the efficiency of the adiabatic mode converter that was designed using the NM-AMC method considerably exceeds that obtained using an analytical method proposed for a strip waveguide.

This paper investigates the mode conversion in ridge waveguides. Based on the adiabatic mode evolution, several compact adiabatic mode converters with optimized waveguide shapes are proposed. Three types of adiabatic mode conversion from TM to TE modes are proposed. It is found that the designed adiabatic device is greatly shorter than the one based on the linear shape case, but is capable of achieving the same power mode conversion efficiency. Furthermore, an adiabatic TM ₀ mode maintaining waveguide taper is proposed to guide the TM ₀ mode from narrow waveguide to wide waveguide. The efficiency of the designed adiabatic TM ₀ mode maintaining waveguide taper considerably exceeds that obtained with the parabolic design. Then, combining the TE ₁ mode to TM ₀ mode adiabatic mode converter and TM ₀ mode to TE ₃ mode adiabatic mode converter, a TE ₁ mode to TE ₃ mode adiabatic mode converter (which is difficult to design and cannot be achieved even with a linear-shape connection) with a total length of 760 μm and 94% power mode conversion is proposed.