Radar emitter signal sorting multi-objective optimization method based on membrane system

Abstract

The invention discloses a radar emitter signal sorting multi-objective optimization method based on a membrane system. The traditional clustering is firstly converted to a multi-objective optimization problem; then, with a membrane computation optimization theory as a main line, intelligent information processing techniques such as a genetic algorithm and computation intelligence are integrated to the framework of the membrane structure for crossing, the self improvement ability of the algorithm is investigated, comparison with the computation efficiency and the integration ability with the current typical intelligent algorithm is carried out, the convergence accuracy and the effectiveness of the membrane system are verified, multiple Pareto frontier solutions with different clusters are expected to be acquired, the diversity and the difference of sorting results are kept, and accuracy of later sorting is facilitated. In combination of the nature of emitter signal sorting, cluster effectiveness indexes, feature evaluation indexes and other evaluation criteria related to sorting form a corresponding problem model and a multi-objective function. In view of intercepted unknown radar signals, the nature of unsupervised classification can meet radar emitter signal sorting requirements.

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