The article contains a review of methods used to create a learning neural network model intended for intrusion detection system based on “Snort” software. The aim of research: to create an adaptive expert system representing a cascade of intrusion detection and special neural modules, considering local area network traffic features.