| Electronic reconnaissance refers to a means of intercepting enemy signals and obtaining the location,category,and threat level of radiation source by electronic equipment.In the military field,electronic reconnaissance provides important intelligence support for battlefield reconnaissance,battlefield command,and battlefield interference.In the civilian field,electronic reconnaissance has played an important role in the security of key areas.In recent years,electronic reconnaissance satellites have become the research focus of the world’s military powers due to their advantages of all-weather,wide range,low environmental impact,and not easy to be attacked by the enemy.However,satellite experiments have the problem of high cost,and it is difficult to directly carry out physical experiments.Computer simulation technology provides a low-cost and highly feasible solution to this problem.At the same time,in the long-period and multi-platform observation scenario,how to effectively integrate massive observation data has become another problem of the spaceborne reconnaissance system.Therefore,this paper designs and implements a spaceborne electronic reconnaissance simulation system based on data fusion.First of all,this paper describes the data fusion technology,signal processing algorithm and transfer learning technology involved in the spaceborne electronic reconnaissance system.Secondly,based on the spaceborne electronic reconnaissance activities in the real world,this paper analyzes the composition of the spaceborne electronic reconnaissance system and the execution process of the satellite reconnaissance work,and proposes the functional requirements and non-functional requirements of the system.Among them,functional requirements can be divided into radar management,satellite management,electronic reconnaissance management,signal processing management,etc.At the same time,this paper puts forward the algorithm requirements for the important link of electronic reconnaissance(the problem of radar emitter identification).As far as we know,the current radar emitter identification algorithm has shortcomings such as a single technical route for data acquisition,and the identification of unknown radar radiation sources(radars not included in the database)needs to be studied.Based on this,this paper has completed three tasks.First,this paper simulates the scene of spaceborne electronic reconnaissance,designs a signal processing algorithm based on datalevel fusion,and enriches the simulation technology of radar radiation source data.Second,this paper designs an unknown radar judgment method based on K nearest neighbors,which accurately distinguishes the radar to be identified into known data and unknown data,and the judgment accuracy rate can reach 93.69%.Third,this paper designs an algorithm for identifying unknown radar radiation sources based on feature fusion and transfer learning.Using data fusion,transfer learning and deep learning methods,the internal correlation between unknown radars and existing radar databases is fully excavated,and the detection of unknown radars is realized.The identification and threat assessment of the system,the identification accuracy rate can reach 92.52%.After that,this paper combines functional requirements,non-functional requirements and algorithm requirements to design the architecture,database and functional modules of the spaceborne electronic reconnaissance system in detail.With the help of professional radar toolbox and satellite toolkit,this paper develops a simulation system of spaceborne electronic reconnaissance.Finally,this paper conducts a comprehensive test on the function and performance of the system.The test results show that the system can meet the needs of users for various services such as radar simulation and satellite receiver simulation.At the same time,the system runs stably and can quickly interact with users.The spaceborne electronic reconnaissance system can simulate the scene of satellite reconnaissance radar,and realize the functions of transmitting,intercepting,processing and identifying radar signals.This paper has completed the expected work. |