A Multi-objective Optimization Method for Radar Radiator Signal Sorting Based on Membrane System

Abstract

The invention discloses a multi-objective optimization method for radar radiation source signal sorting based on a membrane system. First, the traditional clustering is transformed into a multi-objective optimization problem, and then the membrane calculation optimization theory is used as the main line, and the genetic algorithm is integrated into the membrane structure framework. Algorithms, computational intelligence and other intelligent information processing technologies are intersected, and the self-improvement ability of the algorithm and the calculation efficiency and fusion ability of the current typical intelligent algorithms are compared to verify the convergence accuracy and effectiveness of the membrane system, in order to obtain multiple data sets with different clusters. The Pareto frontier solution maintains the diversity and difference of sorting results, which is conducive to the accuracy of later sorting. Combined with the nature of radiation source signal sorting, the clustering validity index, feature evaluation index and other evaluation criteria related to sorting are constructed into corresponding problem models and multi-objective functions. For the intercepted unknown radar signals, the essence of unsupervised classification is in line with the classification requirements of radar emitter signals.

Description

technical field [0001] The invention belongs to the technical field of radar radiation source signal sorting and processing, and in particular relates to a multi-objective optimization method for radar radiation source signal sorting based on a membrane system. Background technique [0002] Since the conventional five-parameter sorting method cannot meet the complex system coexistence and high-density signal environment, it is not only in the ordinary time domain and frequency domain to deeply analyze the pulse flow characteristics and waveform variation rules of complex system radar emitter signals. And in the time-frequency domain (via Wigner-Viller transform, wavelet and wavelet packet transform, etc.), nonlinear transform methods are also used in other mathematical transform domains to fully extract intra-pulse features of radar emitter signals, and to construct effective The feature parameter system provides support. During the research process, we also realized that s...

AI Technical Summary

Problems solved by technology

The clustering and sorting involved in the past all used traditional cluster analysis or improved and combined algorithms based on it, subject to the limitations of traditional clustering methods,

Images