REAL-TIME MANAGEMENT OF CRITICAL IT-SYSTEMS BASED ON NEURAL NETWORK TECHNOLOGIES

Abstract

The paper investigates a relevant applied problem associated with building decision support systems for critical information services. An original approach is proposed, based on neural network forecasting, within which a method of dynamic local approximation using neural network models has been developed. The principles of constructing and implementing the operational algorithm (under conditions of uncertainty of the external load profile) of a combined proactive system for managing computational resources are outlined. Experiments have been conducted that confirm the effectiveness of the method and the approach as a whole.