Design And Implementation Of Radar Radiation Source Signal Reconnaissance And Processing System Software Based On QT

Radar emitter reconnaissance is to intercept and analyze the electromagnetic signal emitted by enemy emitter through electronic reconnaissance equipment,so as to obtain the electronic intelligence such as threat level,communication content and specific location.It is an important research content in modern electronic warfare.With the increasing complexity of electromagnetic environment and the emergence of new radar,the traditional radar emitter reconnaissance has some problems,such as over dependence on feature database,complex and large amount of reconnaissance data.The development of artificial intelligence technology provides a new direction to solve these problems.Therefore,building a radar emitter recognition system based on artificial intelligence to realize radar emitter recognition and method performance evaluation in complex electromagnetic environment will provide research reference for the application of artificial intelligence in the field of electronic reconnaissance,which has important theoretical significance and military value.Traditional radar emitter reconnaissance processing software is usually written in MATLAB or C / C + + language.Due to the lack of deep learning algorithm library,it is difficult to meet the development requirements of radar emitter reconnaissance software under the current artificial intelligence technology.Therefore,in this paper,for the sorting and recognition process of radar emitter signal,on the basis of in-depth analysis of traditional radar emitter reconnaissance algorithm,combined with the one-dimensional CNN neural network algorithm designed in this paper,the design and implementation of radar emitter intelligent reconnaissance system software is completed based on cross platform software QT and python language.This paper mainly studies from the following three aspects Firstly,aiming at the disadvantage that the recognition stage of traditional radar emitter reconnaissance process relies too much on emitter characteristic parameter library,this paper designs the radar emitter intelligent reconnaissance process,and designs and realizes the radar emitter reconnaissance processing architecture based on deep learning.Among them,the parallel processing architecture based on high-performance computing module group +deep learning server is designed at the hardware level,and the CPU + GPU computing platform is adopted for the high-performance computing module group;the emitter sorting and intelligent identification algorithm and application layer function modules are designed at the software level.Secondly,on the basis of the intelligent reconnaissance process,this paper first analyzes the software development requirements,designs the reusable standard software interface on the basis of object-oriented,which is convenient for the later software function expansion;then designs the standardized algorithm interface on this basis,so that the sorting and identification algorithm can follow the unified interface to develop;designs the visual drag and drop component Finally,QT is used to design the classes and functions of each module,unify the code standard,and complete the construction and function realization of radar emitter reconnaissance system software.At the algorithm level,for the emitter signal sorting,this paper first designs a clustering pre sorting algorithm based on pulse width,carrier frequency and angle of arrival.The above three characteristic parameters are used as the conditions of clustering pre sorting to dilute the complex and aliased pulse sequence.Secondly,the main sorting algorithm based on PRI transform is proposed.The PRI transform is carried out on the radar pulse sequence to form the PRI spectrum.The PRI value corresponding to the peak value exceeding the threshold is used to search the sequence,and then a certain type of radar signal is sorted out and its characteristic parameters are extracted.On the other hand,for emitter signal recognition,this paper designs a recognition algorithm based on one-dimensional CNN,trains and tests if emitter data by adjusting network parameters,and finally compares the training time and recognition effect of radar emitter if data under one-dimensional CNN,stacked self encoder and multi-layer perceptron networks,and obtains the most suitable for radar emitter recognition One dimensional neural network based on neural network.Finally,the software of radar emitter intelligent reconnaissance system designed in this paper is tested,the test environment of radar emitter reconnaissance processing experiment is built,and the identification test and comparison are carried out by simulating the signal input system software of various emitter types.The software uses the clustering and PRI sorting algorithm proposed in this paper,which can separate the overlapped measured emitter signals;the one-dimensional CNN network algorithm is ideal for emitter type recognition,and the recognition rate can even reach 93% when the signal-to-noise ratio is 20 d B.The experimental results fully prove that the radiation source intelligent reconnaissance algorithm proposed in this paper can improve the problems of insufficient database characteristic parameters and low recognition rate under the traditional radiation source reconnaissance process,and the radiation source reconnaissance software can also complete the reconnaissance process and the result analysis of the algorithm,which is of great significance for the Research of radiation source reconnaissance intelligent algorithm.