Research On PD Radar Interference State Recognition Based On Convolutional Neural Network

Pulse Doppler(PD)radar system is widely used because of its characteristics of large transmitting duty cycle,high average power,and long operating distance.With the rapid development of electronic countermeasures technology,radar jamming methods are gradually diversified and intelligent.At present,there are many types of jamming against PD radars,including noise suppression jamming,response repeating jamming,and direct repeating jamming.It is of great significance to identify the type of interference in time and make decisions against interference.This paper focuses on the electronic jamming pattern recognition of pulse Doppler radar based on convolutional neural network.Aiming at the signal time and space characteristics of various types of electronic jamming and the radar signal processing flow,using deep learning technology,a PD radar jamming pattern recognition method based on convolutional neural network is proposed.First,the two-dimensional array of the PD radar digital baseband echo signal after matched filtering and moving target detection is preprocessed by two sampling methods.Second,the deep convolutional neural network is used for feature extraction,and then the Softmax classifier is used to determine the interference category sort.Simulation experiments show that the method described in this article has a high recognition accuracy.Traditional radar jamming pattern recognition algorithms rely on manual feature extraction and discrimination based on statistics or traditional machine learning algorithms.The algorithm is susceptible to human factors,and there is room for improvement in accuracy and robustness.In recent years,some deep learning algorithms directly transplanted to the image field restrict the transplantation of hardware platforms in the network computing speed and model volume.This paper proposes that the algorithm innovatively uses two kinds of sampled data to replace large-capacity complete data,and achieves a recognition accuracy of up to 98% in a variety of interference types.At the same time,it is designed according to the edge computing hardware platform resources.The model is small and the algorithm calculation process is short..On this basis,this article combines actual scientific research projects to design a VPX-3U standard signal processing board.The board adopts the structure of DSP chip + FPGA chip,which not only has the calculation ability of DSP's complex algorithm,but also has the parallel calculation ability of FPGA.Completed schematic design,printed board design and circuit debugging and testing.On this hardware platform,this article has completed the programming,debugging and testing of the convolutional neural network interference recognition algorithm.The experimental results show that the algorithm runs correctly and has a high correct recognition rate for interference patterns.