Метод проактивного оцінювання стійкості будівельних проєктів на основі нейро-нечіткого моделювання

Effective management of sustainable development projects in the construction industry is impeded by the reactive nature of traditional control methods and the challenge of integrating heterogeneous data, which includes the quantitative metrics and qualitative expert assessments inherent to Environmental, Social, and Governance (ESG) criteria. To address this problem, this paper proposes a method for proactive sustainability assessment, the implementation of which involves an integrated set of models to function as an early warning system. The method is based on an Adaptive Neuro-Fuzzy Inference System (ANFIS), selected for its unique ability to model complex nonlinear systems while handling the ambiguity and uncertainty of sustainability indicators. The implementation of the proactive sustainability assessment method for construction projects begins with the application of a developed structural-semantic data model, which transforms heterogeneous project inputs into a unified input vector through hierarchical structuring, normalization, and fuzzification. The resulting vector serves as the foundation for the method’s computational core – a predictive neuro-fuzzy model implemented on the ANFIS architecture. This model, by learning from historical data, autonomously generates a knowledge base of IF-THEN fuzzy rules, identifies nonlinear dependencies, and forecasts a Proactive Sustainability Index (PSI). The final results from the predictive model are interpreted using a diagnostic decision-support model, which visualizes the dynamics of the PSI and, by analyzing the most activated fuzzy rules, performs a root-cause diagnosis of potential deviations, thereby converting computational results into practical managerial tools. The key conclusion of this research is that the proposed method, grounded in the integration of structural-semantic, predictive, and diagnostic models, operationalizes proactive management through a data-driven system. It objectifies the assessment of complex sustainability factors, bridging the gap between qualitative expert knowledge and quantitative data. In contrast to existing «black-box» artificial intelligence models, the method ensures transparent diagnostics due to the interpretability of the predictive model’s fuzzy rules, which enhances trust in the results. Ultimately, the developed method provides the management of construction organizations with a scientifically grounded and adaptive toolkit for anticipating sustainability-related risks, optimizing managerial interventions, and improving overall project outcomes in a dynamic environment.