The System of Automatic Detection of Penetration through the Protected Perimeter Based on Fiber Optic Sensors and Neural Network

Abstract

The article discusses the principle of constructing the software part of the alarm system's signal processing unit, which will solve the problem of detecting penetration through the protected perimeter by means of a signal coming from a sensitive optical fiber located on a mesh fence. The study is conducted through consideration of such problems as the decomposition of the signal received from a sensitive fiber-optic sensor into components and its further processing, first for training the neural network, and then for analyzing its state. A large part of the work is occupied by consideration of a solution that allows to form an algorithm for building a digital signal processing unit using a neural network and also to reduce the number of false alarms caused by interference and external influence on the security system.