Research On Radar Signal Sorting Method

At present,the core technology of radar reconnaissance equipment is how to sort and identify pulse signals of various radar emitter from high-density and large-quantity pulse streams.The traditional sorting model is based on conventional parameters.The research in this area has encountered a bottleneck.It is necessary to find new algorithms.How to sort the the received radiation emitter signals correctly in a complex and ever-changing electromagnetic environment is a research topic that should not be underestimated.Firstly,the overall structure of the traditional sorting algorithm is introduced.It mainly includes four modules: known-radar pre-sorting,unknown-radar pre-sorting,known-radar main sorting,unknown-radar main sorting.And the basic theory of each module algorithm is elaborated.The development,advantages and disadvantages of the unknown radar main sorting algorithm are introduced in detail.Secondly,some improvements have been made in the structure of the traditional sorting model: The first improvement is to replace the unknown radar pre-sorting template matching with unknown-radar dynamic clustering sorting.The second improvement is to to add a PRI sorting result to merge processing methods.Considering the possibility of separating pulse signals from the same radar emitter after the PRI separation,it is necessary to combine radar emitter of increasing batch to form a real radar emitter.The improved model is verified based on the measured data.Thirdly,for the bottleneck of the characteristic parameters of the pulse,the intrinsic properties and separability of the intra-pulse modulation characteristics are searched.some features are mined by similarity coefficient,ambiguity function and bispectrum transformation.Eight different modulation signals are generated to verify the features,and then each feature extraction is verified based on the measured data,and the experimental results are analyzed.Finally,FCM and SVC clustering sorting experiments are performed based on the intra-pulse feature parameters and the inter-pulse feature parameters of measured data.The analysis shows that when there are multiple sets of data,it is difficult to separate the radar signal with similar inter-pulse feature parameter values or intra-pulse feature parameter values according to a layer of clusters,and clustering must be performed again,parameter setting during cluster sorting also has a significant effect on the results.Therefore,the classification algorithm for accurate,rapid and manual non-intervention has yet to be studied.