Distributed neural network systems for collective management of construction machinery: from group interaction to swarm intelligence

Abstract

The present study is devoted to the study of the prospects of using distributed neural network systems for collective management of construction machinery based on the principles of swarm intelligence. The concept of swarm intelligence, based on modeling the collective behavior of decentralized self-organizing systems, provides fundamentally new opportunities for improving the efficiency, safety and autonomy of construction machinery. Based on the system analysis of modern approaches to the construction of distributed control systems, a multilevel architecture of neural network management of construction machinery is proposed, transforming individual operations of individual machines into a single self-organizing system of group interaction. Empirical verification of the developed model was carried out on the basis of a comprehensive analysis of data obtained during field tests of the prototype system on real construction sites. The results of the study demonstrate a 37,4% increase in operational efficiency, a 41,2% reduction in complex task completion time, and a 28,9% reduction in energy consumption compared to traditional management methods. The proposed approach opens up new prospects for the development of adaptive, fault-tolerant and self-learning systems for collective management of autonomous construction machinery capable of operating effectively in complex dynamic conditions of construction sites.