The article discusses the transformation of approaches to managing the life cycle of capital construction objects through the implementation of hybrid intelligent systems. A paradigm shift towards a proactive model based on risk prediction and prevention using hybrid intelligent systems is substantiated. Unlike existing solutions, the proposed approach overcomes the gap between physical models and artificial intelligence methods through their symbiotic integration. An innovative architecture of a hybrid system has been developed, combining physical models of degradation of building structures with modified recurrent neural networks for predicting residual resource based on developed mathematical models. A multi-level implementation architecture is proposed, describing specific computer vision and predictive analytics algorithms, as well as their validation methodology. The study found that the use of hybrid systems will improve the accuracy of predicting building defect development by 35-40% and reduce operating costs by 20-30% through optimization of repair strategies. The proposed approach demonstrates potential for increasing inter-repair periods by 25-40% while ensuring regulatory safety indicators.
Абакумов et al. (Thu,) studied this question.