Features of Building Software Simulation to Optimize the Efficiency and Reliability of Automated Production Lines Using AI Methods

This article explores a methodological approach to integrating artificial intelligence (AI) into automated production lines through the development of computer-based simulations. As a case study, the research focuses on an experimental assembly line for agricultural drones, designed to serve as a testbed for intelligent optimization strategies. The simulation is constructed using object-oriented programming principles, enabling modularity, scalability, and architectural clarity. Each component of the production line—conveyors, manipulators, quality control nodes, and drone modules—is modeled as an independent object with defined behaviors and interaction protocols. A key objective of the study is to replace selected elements of the simulation architecture with AI-driven agents capable of learning and adapting to dynamic production conditions. These agents implement reinforcement learning algorithms and heuristic decision-making strategies aimed at improving the overall efficiency, reliability, and responsiveness of the manufacturing process. The simulation environment supports both graphical implementations, using modern game engines such as Unity, and console-based models developed in Python or C++, allowing for comparative analysis of performance, flexibility, and integration potential. The concept of digital twins is central to the proposed framework, providing a virtual mirror of the physical production system that can be used for predictive analytics, resource optimization, and real-time decision support. The study highlights the advantages of simulation-based AI integration, including reduced development risk, accelerated prototyping, and enhanced system transparency. Minor limitations, such as abstraction gaps and computational overhead, are acknowledged and discussed. Overall, the research contributes to the architectural foundations of intelligent manufacturing systems by demonstrating how simulation environments can serve as platforms for testing, training, and deploying AI agents. The findings are relevant to developers of industrial automation, digital twin frameworks, and adaptive control systems seeking to enhance production line performance through intelligent technology.