Next-Generation Solvers for Complex Microwave Engineering Problems

该项目旨在设计一种互补的物理指导，数据驱动的方法，可以及时准确地解决复杂的微波工程问题。到目前为止，阻止这种方法的主要瓶颈是训练的理论模型与现实之间的差距，将使用多频复值域自适应技术来克服。该方法将使用深度神经网络来可靠地学习微波工程问题的物理概念。该项目将产生重大的经济和社会效益，例如支持通信系统、采矿、基础设施检查、安全、遥感和微波成像的有效设计、安装和运行。

This project aims to design a complementary physics-guided, data-driven method that can accurately solve complex microwave engineering problems in a timely manner. The primary bottleneck so far preventing that approach, which is the disparity between the trained theoretical model and reality, will be overcome using a multi-frequency complex-valued domain adaptation technique. The method will use deep neural networks to reliably learn the physical concepts of microwave engineering problems. This project will have significant economic and societal benefits, such as supporting the efficient design, installation and operation of communication systems, mining, infrastructure inspection, security, remote sensing, and microwave imaging.