ІНТЕЛЕКТУАЛЬНА МОДЕЛЬ ФОРМУВАННЯ ІННОВАЦІЙНО-ІНЖИНІРИНГОВИХ КЛАСТЕРІВ

Anton Rogovyi, Олена Aхієзер, Світлана Решетнікова et al. · The Scientific Issues of Ternopil Volodymyr Hnatiuk National Pedagogical University Series pedagogy · 2026

In the context of digitalization and the development of the knowledge economy, innovation and engineering clusters are transforming into complex dynamic systems characterized by a high level of uncertainty, multifactor interactions, and significant dependence on external managerial influences. Traditional economic and mathematical approaches do not provide sufficient flexibility for modeling such systems, which necessitates the application of integrated artificial intelligence methods. The aim of this study is to develop and validate a hybrid model for the formation of innovation and engineering clusters that combines machine learning methods, graph analysis, agent-based modeling, and digital twins, followed by an assessment of the synergistic effect of their functioning and an analysis of the impact of managerial factors on cluster development dynamics. The study employs a multilevel approach that includes agent clustering (k-means, DBSCAN), construction of a graph-based interaction model, application of graph neural networks to account for topological dependencies, and agent-based modeling to investigate the evolution of the system over time. Additionally, agent compatibility and risk functions are introduced to formalize the synergistic effect and cluster resilience within an optimization framework. The integration of all components is implemented in the form of a digital twin, enabling scenario analysis capabilities. Research Results. The model was validated on a simulated regional cluster consisting of enterprises, universities, investors, and a governmental body. Simulation results demonstrated that government support significantly increases the innovation activity of cluster participants, enhances the density of network connections, and raises the number of cooperative projects. Key nodes of network interaction were identified, particularly the role of universities as knowledge transfer centers and investors as financial hubs. Conclusions. The obtained results confirm the effectiveness of the proposed approach and its suitability for forecasting the development of innovation ecosystems and supporting strategic managerial decision-making. The developed model provides a comprehensive integration of machine learning methods, graph analysis, agent-based modeling, and digital twins, enabling the transition from static cluster analysis to adaptive and predictive management. The practical significance of the study lies in the possibility of applying the proposed approach to support regional innovation policy, evaluate the effectiveness of managerial decisions, and form sustainable innovation and engineering ecosystems.

Cited by 0 works. Open-access PDF

Authors