A Rapid Recognition Model Of Multi-source Mixed Radar Signals Based On A Priori Information Database

One of the key technologies in electronic support system is sorting and identification the radar emitter signals.Nowadays,radar emitter signal sorting and recognition mainly depends on five conventional parameters.However,the electromagnetic environment is extremely complicated where radar,communication devices and guided weapons are used on a large scale in the modern battlefield.Meanwhile,with the continuous advancement of radar technology,the radar operating frequency is continuously improved and the modulation method is becoming more diverse.So,in the electromagnetic environment with dense signals and various systems,how to effectively sort and identify multisource hybrid radar signals is one of the key problems that the radar detection systems need to solve.Establishing the prior database is one of the effective way to identify high-density signals.Pre-identifying signals with a priori database can reduce subsequent processing pressure.Nonetheless,the signal waveforms and parameters emitted by complex radars are constantly changing.The traditional method of using Pulse Description Word(PDW)as the prior database has not been able to meet the requirements of identification in an accurate and timely manner.The Ambiguity Function(AF)can comparatively describe the internal structure of the signals.The new features can be found through the extraction AF in the signals,which is conducive to constructing a new prior database.In the factional domain,it takes a lot of time to calculate and extract the section in Ambiguity Function Main Ridge(AFMR).Even with the application of the intelligent search algorithm,the processing time is still about 2 second.The massive radar signals cannot be handled timely and accurately.Based on the above points,this thesis comprehensively illustrates the AFMR rapid search,the construction of multidimensional prior information database,the extraction of section feature in AFMR,etc.The main work and achievements are as follows:(1)This thesis briefly introduces the research status of radar emitter signal recognition,and analyzes the advantages and limitations of several identification methods.At the same time,the theories of the AF and the relationship between the autocorrelation function and AF are explained.In order to improve the drawbacks in the current optimization method without prior information,the intelligent search method of AFMR based on the adaptive evolution is put forward.The method uses the characteristics of instantaneous frequency to pre-identify the signal and the parameter of the differential evolution algorithm is adjusted to greatly improve the adaptability and convergence speed in the algorithm.The experimental results show that the convergence speed in the proposed method is 25 times faster than the traditional method,80% higher than other intelligent search algorithms on the basis of not reducing the accuracy.The search time of the section in AFMR is reduced to less than 500 ms.(2)Because the electronic detection objects mostly belong to non-cooperative signals,the research on the construction of the prior database is relatively lagging behind.In this thesis,based on the prior database with PDW parameters,a mufti-dimensional prior database is constructed based on the instantaneous frequency and AFMR.At the same time,the unknown signals are designed to match the existing records in the formation base with the methods of threshold matching,the phase coefficient and hierarchical matching.The experimental results show that the constructed prior information and the recognition model have promptness and accuracy.The unknown signals can be accurately identified when the traditional method is almost invalid.In this way,the identification content in the prior database is enriched and the application range of AFMR is expanded.(3)The local differential characteristic in AFMR is not recorded in the prior database and the fuzzy kernel clustering algorithm with effective evaluation is used to perform automatic clustering.A preliminary exploration was carried out to automatically add unknown radar parameters into the prior database.The experimental results show that the accuracy is satisfactory under high SNR,which lays a foundation for the subsequent automatic addition of unknown radar parameters into the prior database.