Research on Radar Plot Classification Based on Fully Connected Neural Network

Clutter is the inherent environment of radar signal detection and processing. In this paper, the full connected neural network is used to study the classification of radar clutter and real targets. First, according to the engineering experience, the multi-dimensional feature index of radar plot is given, and the fully connected neural network is built in the Tensorflow architecture. According to the application scenarios in this paper, the effect evaluation indexes of the training model were defined, and the classical loss function was optimized. Finally, the training results were given.