ALGORITHMS FOR INTELLIGENT SUPPORT OF MANAGEMENT DECISIONS IN CASE OF THREATS FROM FOREST FIRES

The relevance of the study is due to the need to improve the validity and effectiveness of management decisions when responding to forest fire threats in conditions of high uncertainty and dynamic development of fire hazard situations. Existing approaches, based mainly on expert methods, do not always allow for prompt and accurate consideration of the entire range of factors influencing the choice of management decisions, which requires the implementation of intelligent decision support systems. The work uses a hybrid approach combining deep neural tree and fuzzy logic methods, implemented in the form of algorithms for intelligent decision support at the strategic and tactical levels of management. Machine learning methods are used to predict key parameters of forest fires (area, duration of liquidation, distance to settlements, number of forces and means involved), and the fuzzy logic apparatus is used to generate sound management recommendations based on forecast data. The main results of the study are the development and implementation of algorithms that automate the processing of initial data, forecast target indicators and determine recommended management decisions for the Commission for the Prevention and Elimination of Emergencies and Fire Safety. Examples of the implementation of algorithms are given, high values of the quality metrics of machine learning models and the performance of the proposed solutions on test data are demonstrated. The scientific novelty of the work lies in the integration of deep neural tree methods with Takegi-Sugeno fuzzy inference into a single system designed to support decision makers in responding to forest fire threats. Flowcharts of algorithms are proposed that allow formalizing the decision-making process and increasing their interpretability. The practical significance of the results lies in the possibility of using the proposed algorithms as part of intelligent decision support systems for government agencies in emergency situations to improve the validity, timeliness and quality of management decisions, as well as optimize time and human resources in responding to forest fire threats.