This article analyzes emergency situations arising from natural disasters (fires, explosions, earthquakes, floods, landslides, etc.) and man-made accidents. The event in which the values of fire risks in the area decrease uniformly and at the same time the amount of fire risks is minimal is considered optimal. Models, methods, and logistic regression algorithm have been used to predict fire hazard situations and identify potential fire safety measures.
The report is devoted to improvement of algorithms for separating functions by Bezier curve, which are based on the algorithm design with optimal classifying objects. The developed algorithm performs classification in the systems with complex order.