Modeling of digital management processes of an organization using artificial intelligence and adaptive fuzzy logic mechanisms

Abstract

In the modern architectural and construction industry, which is undergoing a fundamental transformation due to the mass introduction of digital technologies in the processes of design, construction and operation of capital construction facilities, the modeling of digital management processes of organizations based on artificial intelligence and adaptive fuzzy logic mechanisms is of particular relevance. The complexity of managing such organizations is determined by the multifactorial nature of production processes, the dynamic nature of project implementation conditions, the high degree of uncertainty in the external and internal environment, and the need to simultaneously consider technical, economic, organizational, and regulatory requirements, which limits the effectiveness of traditional deterministic models and linear decision-making algorithms, especially when management actions rely on expert assessments and qualitative characteristics. Hybrid models that integrate machine learning methods with fuzzy logic mechanisms based on the theory of linguistic variables and the formalization of expert knowledge, combined with adaptive approaches, provide processing of both quantitative and qualitative parameters, allowing for continuous adjustment of model parameters in accordance with the changing conditions of the organization's functioning. In the context of digitalization, which includes information modeling of buildings, digital twin technologies, and automated design and management systems, such intelligent systems integrate formalized algorithms with the heuristic knowledge of experts through fuzzy production rules and neuro-fuzzy networks, replicating expert thinking and improving algorithms based on feedback from implemented management actions. This is particularly important for coordinating large-scale infrastructure and urban planning projects, where it enhances the accuracy of predicting deadlines and costs, optimizes resources, and reduces conflicts. A comparative analysis of various types of fuzzy inference demonstrates the superiority of a hybrid architecture combining Mamdani fuzzy inference with genetic algorithms and the error backpropagation method over alternatives: the ANFIS model achieves a prediction accuracy of 92.37% with a stability of 1.84% standard deviation, surpassing Mamdani by 4.94 percentage points, while the zero-order Sugeno provides minimal computing time 38.29 ms, and Mamdani has the highest adaptability to expert knowledge at 9.21 points. The implementation of the adaptive model in the activities of construction organizations leads to an increase in the accuracy of time planning by 25.69%, a reduction in actual cost deviations from the planned by 53.24%, a reduction in production capacity downtime by 49.76%, a reduction in decision approval time by 61.33% and an increase in customer satisfaction by 22.80%, with a maximum effect of 47.94% for large organizations, implementing multidisciplinary infrastructure projects, where the dependence of efficiency on scale is described by a power function with an indicator of 0.684, reflecting increasing returns. The learning dynamics of the model show a logarithmic increase in prediction accuracy from 73.42% after the first month to 95.17% by the 24th month, with a reduction in the mean squared error by a factor of 4.02, confirming statistical significance through a variance analysis with high F-values and a determination coefficient of 0.8472. The developed approach, which complies with national standards and industry digital transformation strategies, is scalable to regional and industry systems, integrating with digital twins and e-governance platforms, opening up the prospects of incorporating deep learning, natural language processing, computer vision, and big data technologies for the transition to autonomous management in the Construction 4.0 concept, which contributes to improving the quality of facilities, reducing the time and cost, minimizing the environmental impact, and achieving national development goals through the formation of a unified digital ecosystem in the construction industry.