Use of a Neurofilter to Extract a Periodic Signal from Noise

Abstract

An increase in the requirements for airborne radars necessitates continuous improvement of methods for processing radar information. The relevance of the development of new approaches and breakthrough technologies is due to a contradiction between the promising characteristics of the radar and the extreme capabilities of the well-known technical solutions for processing radar information. The neuroprocessor model considered in the article is built on analog circuits. The development goal is to achieve high performance hardware recognition of signals by trained neuro-like devices. The developed analog neuro-filter is designed to solve the problems of recognizing periodic signals under conditions of high-intensity interference in the megahertz range. The experiments showed the possibility of separating signals from interference with a speed of several tens of microseconds, which was achieved through parallel processing of information and the use of analog logic in control circuits. An analog neuro filter can be a competitive alternative to digital signal recognition systems in compact equipment.