Research On Pattern Synthesis Technology Of Phased Array Antenna And Frequency Diverse Array Antenna

The rapid development of modern wireless communication and electronic technology has put forward higher and higher requirements for antenna performance.Array antennas have gained a lot of attention due to their benefits of easy to acquire the beampatterns with high gains,narrow beams,shaped beams,low sidelobes,and easy to realize beam scanning.The array antenna synthesis technology,which can solve the design parameters of each antenna element in accordance with the design requirements,is the basis of array antenna design.However,solving the array antenna synthesis problem is often regarded as a highly nonconvex,nonlinear and time-consuming task,which puts forward extremely high requirements and challenges for the array synthesis technology.Although many synthesis methods have been proposed for phased array antennas,these methods have some drawbacks,such as limited applicability,computational inefficiency,and inability to consider the non-ideality of the actual arrays.Moreover,the increasing demand for large arrays in modern electronic systems makes it difficult for existing synthesis methods to design large array antennas quickly and stably.Therefore,it is of great significance to carry out research on general,flexible,efficient and high-performance array antenna pattern synthesis technology.On the other hand,in order to achieve more flexible beam control,frequency diverse array antennas have been extensively studied in recent years.By introducing tiny frequency increments between the array elements,frequency diverse array antennas can generate angle-and range-dependent beampatterns,which is crucial for building next-generation radar systems.However,the beampatterns of frequency diverse array antennas are range-angle-coupled and non-static.Therefore,it is of great significance to carry out research on focusing and static beampattern synthesis for frequency diverse array antennas.Focusing on the pattern synthesis technology of array antennas,the application of artificial neural networks,ensemble learning,convolutional neural networks,iterative Fourier technique,and evolutionary algorithms in the phased array and frequency diverse array pattern synthesis is investigated.The main contribution of this dissertation are summarized as follows:1.An encoder-decoder-based artificial neural network framework is proposed for focused and shaped beampattern synthesis of linear array antennas to overcome the shortcomings of the existing phased array synthesis methods,such as lack of flexibility,huge computational burden,and difficulty in considering the non-ideality of actual arrays.In this framework,the encoder and decoder are designed as an array synthesizer and an array analyzer,respectively.By pre-training the decoder,the framework has extremely high computational efficiency in array synthesis applications and can potentially realize real-time linear array synthesis.Furthermore,the proposed framework can be used in the linear array pattern synthesis for any given array geometry.Both ideal synthesis and real synthesis with mutual coupling effects can be considered,as well as,different excitation control can be employed with this framework.Therefore,the proposed framework has high flexibility and versatility in array synthesis.2.A multi-branch encoder-decoder-based artificial neural network framework is proposed for the synthesis of multiple-pattern non-uniformly spaced linear arrays.The encoder and decoder in each branch serve as an array synthesizer and an array analyzer,respectively.By minimizing a loss function related the radiation patterns,the array excitation amplitudes,and the array element positions,different encoder-decoder branches work jointly to obtain a common set of array excitation amplitudes and element positions with a minimum element spacing constraint for different radiation patterns.Furthermore,the ensemble learning technique is introduced to the decoder to improve the accuracy of array analysis.By pre-training the decoder with the actual training samples,it can replace full-wave simulation analysis to consider the mutual coupling effects of non-uniform array antennas in the synthesis process in real time.Therefore,the proposed multi-branch framework has high research value and engineering significance.3.An encoder-decoder-based convolutional neural network framework is proposed for efficient shaped beampattern synthesis of planar array antennas to address the problem that there are too many parameters in the fully connected structure of artificial neural networks to be directly employed for planar array beampattern synthesis.In this framework,the encoder and decoder are implemented as two fully convolutional neural networks,thereby reducing the number of neural network parameters required for feature extraction and pattern reconstruction.In addition,instead of using the traditional deconvolutional layer in the decoder to realize the upsampling of the feature map,a sub-pixel convolutional layer is employed to effectively avoid the checkerboard artifacts and achieve a more accurate and efficient array analysis.Therefore,the proposed encoder-decoder-based convolutional neural network framework is effective,efficient and superior in the shaped beampattern synthesis for planar array antennas.4.A fast synthesis method is proposed for large thinned array antennas to address the extremely high hardware and software burden required for large array antenna design.In this method,firstly,aiming at the shortcomings of iterative Fourier technique,a modified iterative Fourier technique with a perturbation mechanism is proposed to prevent the algorithm from trapping into local minima.Then a hybrid algorithm based on the genetic algorithm and the modified iterative Fourier technique is proposed for thinned linear and planar array synthesis.The hybrid algorithm inherits the global search ability of the genetic algorithm and the fast synthesis ability of the modified iterative Fourier method,so it has superior synthesis performance and accelerated convergence speed.5.The range-angle-decoupled and quasi-static beampattern synthesis of frequency diverse array is investigated to address the issue that the frequency diverse array has a coupled and time-varying range-angle-dependent beampattern while the conventional phased array only has an angle-dependent beampattern.Firstly,a general conformal frequency diverse array framework is established and a hemispherical frequency diverse array is utilized to explore the unique elevation-azimuth-range localization ability of the conformal frequency diverse array compared with the low-dimensional frequency diverse arrays.On this basis,a logarithmic static nonlinear frequency offset and an optimized static nonlinear frequency offset are proposed to obtain the quasi-time-invariant 3-D focusing beampattern with low sidelobes for the conformal frequency diverse array,so as to alleviate the time-variant beampattern.