基于改进 Fuzzy ART 的自适应雷达信号分选

Being able to remove the interweavement in waveforms came from radiation sources for signal separation, signal sorting can be used as a core technology in electromagnetic spectrum warfare systems of reconnaissance receivers. In order to solve the problems about the high density of pulse streams and severe overlapping of temporal waveforms with domain characteristics in complex electromagnetic environments, an Adaptive Vigilance Threshold Fuzzy Adaptive Resonance Theory (AVT Fuzzy ART) clustering algorithm was proposed to adjust the vigilance threshold adaptively based on the Manhattan distance sensitized to attribute differences, solving the algorithm inadequate problems for most unsupervised model-based signal sorting algorithms. And the proposed algorithm was also designed to weight dynamically for the vigilance threshold based on the transient clustering probability of radiation sources derived from online accumulated data. Simulation results show that the algorithm is competent to eliminate the interweavement of multi-category radiation source signals without prior historical information.