Моделі машинного навчання для аналізу ефективності виробництва та логістики готового бетону

The relevance of the problem. In the post-war period, the issue of efficient organization of production and logistics of construction materials, in particular ready-mixed concrete, which is the basis for the restoration of residential and industrial infrastructure, is of particular relevance. Concrete plants face numerous logistical challenges, including order management, on-time delivery, transportation constraints, and volatile demand. In such conditions, the role of digital twins and intelligent decision support systems is growing. Digital twins allow modeling and optimizing production and logistics processes based on real data in near real time. Methodology. The study applies a systematic approach to the design of digital twins of concrete plant production and logistics using modern methods of simulation modeling, machine learning, and artificial intelligence. A modular information technology architecture has been developed, covering the physical IoT layer, data collection and storage layer, modeling and analytics layer, as well as integration with ERP, CRM, WMS, TMS, MES/EAM, and SCADA. Decision trees, random forests, and neural networks are used as machine learning models to analyze the efficiency of ready-mixed concrete logistics in real time. Results. A generalized model of a digital twin of a concrete enterprise is proposed, which allows combining production processes, order management, logistics, and analytics into a single intelligent information system. Machine learning models for analyzing the efficiency of logistics in real time, which take into account the state of the transport fleet, capacity utilization, delivery schedules, order priority, etc., are considered. Novelty. For the first time, a holistic information technology of a digital twin of the production and logistics system of a concrete industry enterprise is proposed, which integrates the technologies of the industrial Internet of Things, simulation multi-agent modeling, and machine learning methods. The integration of ML models with IoT data allows for automatic adjustment of production and logistics scenarios, unlike the existing order management system, IoT platform of concrete mixers, and dispatching system, which allows for an increase in the efficiency of predictive analytics of ready-mixed concrete logistics. Practical significance. The proposed approaches can be implemented in concrete enterprises of various sizes, both as part of the reconstruction of existing plants and in the construction of new mobile concrete plants. They provide increased forecasting accuracy, reduced costs, improved customer service, and reduced environmental impact through logistics optimization.