Analysis of the detection of an attack based on SQL injection using an impulse artificial neural network

This article presents the results of testing to create a specialized system that helps prevent cyberattacks, thus popularizing the construction of intelligent applications. Based on the results obtained, it can be argued that the tests carried out are satisfactory. The mathematical basis for building a neural network model is the HESADM model (Hybrid Artificial Intelligence Framework). The presented system allows you to form a set of rules using fuzzy logical neurons. This paper presents an approach to the formation of a fuzzy neural network used for detecting SQL injection attacks. The methodology used in this paper is an impulse artificial neural network (SANN), which uses an evolving neural network system (eCOS) and a multi-layer approach of an impulse artificial neural network to classify the exact type of intrusion or network anomaly with minimal computational potential. The impulse artificial neural system forms itself continuously, adapting to the input data, being in a functioning or not state, being under the supervision of an administrator. This system finds application to several other complex problems of the real world, proving its efficiency, including in the field of information security. The considered model is a hybrid evolving pulse anomaly detection model (HESADM), which works on impulses that occur in the system, while neurons are used to monitor the algorithm using a single training pass. In the system, traffic-oriented data is used by importing classes that use variable encoding. The data used is obtained by converting the real characteristics of network traffic into certain time stamps.