Unsupervised radar signal sorting method based on clustering and SOFM

Abstract

The invention relates to an unsupervised radar signal sorting method based on clustering and SOFM, and belongs to the technical field of deep learning and radar signal sorting. According to the radarsignal sorting method, density dynamic clustering is combined with an SOFM neural network capable of automatically achieving structure adjustment, and density clustering and SOFM combined unsupervisedradar signal sorting without preset parameters is constructed; the method specifically comprises the following steps: firstly, carrying out density dynamic clustering on radar signals and carrying out pre-sorting processing to obtain related parameters required by a main sorting stage; and secondly, constructing an SOFM network capable of automatically realizing structure adjustment by using related parameters obtained by pre-sorting to perform main sorting processing on radar signals to obtain a final sorting result. According to the radar signal sorting method, an initial value does not need to be preset, unsupervised sorting of radar signals is achieved, the accuracy of radar signal sorting is improved, and a good processing effect is achieved for radar signals with overlapped parameters.

Description

technical field [0001] The invention relates to an unsupervised radar signal sorting method based on clustering and SOFM, and belongs to the technical field of deep learning and radar signal sorting. Background technique [0002] Radar signal sorting is a key technology in electronic warfare. The environment faced by the modern radar reconnaissance system is becoming more and more complex, the density of radar signals is increasing, and the parameter overlap is becoming more and more serious. The real-time signal processing of the radar reconnaissance system is facing severe challenges. K-means clustering has the advantages of simplicity, speed, and good practicability. SOFM neural network has the characteristics of self-organization learning and is suitable for high-dimensional data clustering. They are all commonly used methods for radar signal sorting. [0003] Although these two methods are classic signal sorting methods, they both have certain shortcomings: (1) The tra...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to solve the problem that when the traditional K-means clustering and SOFM neural network are used for radar signal sorting, the preset parameters have a great influence on the correct rate of sorting, and it is easy to fall into local optimum, the network structure is fixed, and it is impossible to sort The technical status of overlapping radar signals with parameters, and an unsupervised radar signal sorting method based on clustering and SOFM

Images