Development of an Intelligent System for Predicting the Forest Fire Development Based on Convolutional Neural Networks

Abstract

Forests are a natural renewable resource and can meet the needs of the society, provided that they are used for a multiple, rational, continuous and sustainable use. Forest fires are a natural component of forest ecosystems and cannot be completely eliminated. However, in recent decades, there has been a tendency to transform forest fires from a natural regulatory factor into a catastrophic phenomenon causing significant economic, environmental and social damage. It is critical to understand the relationships between the underlying environmental factors and spatial behaviour of a forest fire in order to develop effective and scientifically sound forest fire management plans. The key objective of this study is to enhance the efficiency of the formation of a real-time forest fire forecast under the unsteady and uncertain conditions. In the article, the author proposes to develop an intelligent system for predicting the forest fire development based on artificial intelligence and deep computer-aided learning. A key element of the system is forest fire propagation models that recognise data from successive images, predict the forest fire dynamics and generate an image with a fire propagation forecast. It is proposed to build forest fire propagation models by using a real-time forest fire forecasting method. In the article, the author presented a structural diagram of an intelligent system to forecast the dynamics of a forest fire and described the functional structure of the system by constructing its functional models in the form of IDEF0 diagrams.