Prediction of False Alarms of Security Alarm Systems for Centrally Guarded Facilities based on the Use of a Neural Network

This article proposes a method for predicting false alarms of security alarm systems for centrally protected facilities using a neural network. In centralized security of facilities, all necessary information for analyzing the level of alarms of facility security alarm systems is stored in the protocols of automated workstations of notification transmission system operators, protocols of the centralized security duty station, and the database of protected facilities. The paper proposes a method for identifying factors that determine the level of false alarms, assessing and converting the qualitative values of these factors into quantitative ones. Based on the collection of information on the statistics of the functioning of the centralized security station, a dataset is collected to predict false alarms of security alarm systems. It is shown that using a neural network for the regression problem can achieve high accuracy in predicting the level of false alarms of facility security alarm systems.