Full Connected Neural Networks substitute traditional electromagnetic simulations to calculate target objects' RCS

The target object is known to be a cone-like gyratory body, and after designing and building a suitable model, a dataset is established for training a Full Connected Neural Networks (FCNN). During the process of model building, the Bspline is selected as the envelope for the target slewing out-ofbody profile. The four-layers Full Connected Neural Networks is used to learn the dataset of the target object, so that the trained neural network can replace the traditional electromagnetic simulation calculation to calculate the radar scattering crosssectional area of the target object. Taking advantage of the nature of multi-layers neural networks, it is possible to reduce computing demand and increase the speed of calculations.