ARCHITECTURE OF AN AUTONOMOUS INTELLIGENT SYSTEM FOR MANAGING CONSTRUCTION PROCESSES UNDER THE RISK OF EXTERNAL IMPACTS

Abstract

This paper explores the problem of ensuring the autonomy and fault tolerance of intelligent control systems used in construction process management under conditions of external risks, such as power outages, loss of network connectivity, or cyber-attacks. As artificial intelligence (AI) becomes increasingly involved in construction scheduling, monitoring, and resource coordination, there is a pressing need to develop system architecture capable of maintaining critical functionality under infrastructure failures. The main objective of this research is to design and formalize a model of an autonomous AI-driven system that can operate independently from centralized infrastructure and seamlessly transition to fallback control logic when needed.

The methodology combines system architecture analysis, the use of UML diagrams (use case, component, and state diagrams) to model functional logic and interactions, risk scenario modeling using failure analysis techniques, and a comparative evaluation of centralized and edge-based computing approaches. The proposed three-layer architecture (physical, computational, and communication) is centered on a local edge server. This server hosts AI modules, handles anomaly detection, power switching, and provides decision-making capacity independently of cloud services.

The results demonstrate that the implementation of a hybrid autonomous control system significantly enhances the resilience of construction operations. The edge-based architecture outperforms centralized models in response time, offline operability, and stability in unstable environment. A comparative analysis shows that the deployment of such a system may increase initial costs by 5–10%, yet these costs are justified by a substantial reduction in risk of downtime, delays, and data loss.

The proposed system is particularly suitable for construction sites operating in constrained or high-risk conditions, including strategic infrastructure projects or military-affected zones. Future research should focus on optimizing AI algorithms for offline operation, developing industry-wide integration protocols, and validating the proposed model through real-life implementation in pilot projects.