Possibilities of using software complexes with elements of artificial intelligence in fortification
Abstract
Modern armed conflicts require high speed and validity of planning engineering equipment of defense lines. The article is devoted to the analysis of the implementation of software complexes with elements of artificial intelligence (AI) as decision support systems (DSS) in fortification planning. The concept of GeoAI is considered - a combination of GIS (ArcGIS, QGIS) with machine learning to automate terrain analysis and create a "digital twin" of the territory based on LiDAR and remote sensing for high-precision digital elevation models (DEM). The scientific novelty lies in the formalization of fortification planning as a multifactor optimization problem. To solve it, reinforcement learning algorithms (DQN) are proposed, which allow the AI agent to find optimal configurations of structures. To balance conflict goals (cost, time, protective properties), the NSGA-II algorithm is used to search for Pareto solutions, as well as TOPSIS and entropy weights methods for their ranking. Special attention is paid to modeling the enemy's counteraction using game theory, in particular Stackelberg games and Bayesian Markov games, which will allow implementing the "worst-case scenario" strategy and should provide proactive defense. The key aspect is human control (HITL), where the system acts as an intelligent assistant, and the final decision remains with the person. The software implementation is focused on Python (PyTorch, Ray RLlib, Gurobi) with local deployment for cybersecurity and secrecy. The results of the study confirm that the use of intelligent software complexes will significantly optimize planning, increase the robustness of engineering solutions to the uncertainty of the combat environment, and provide a technological advantage in the management of defense engineering support.
