РОЗРОБКА КОНЦЕПЦІЇ ЦИФРОВОГО ДВІЙНИКА ЕЛЕКТРОПРИВОДА ЛІФТА НА ОСНОВІ СТОХАСТИЧНОЇ МОДЕЛІ ARIMAX

Objective. To develop a concept for a digital twin of an elevator system’s electric drive that combines physical models, stochastic forecasting methods, and sensor data streams to reproduce the system’s dynamic operating modes in real time. Methodology. The study involved the analysis and generalization of approaches to technical diagnostics and the construction of digital twins, mathematical modeling of electromagnetic and mechanical processes in an asynchronous motor, and structural modeling of the system architecture. In addition, stochastic modeling and analysis of the corresponding time series were performed. To enable the description of degradation, physical-statistical and stochastic models were considered. Results. A mathematical model of an asynchronous electric drive in a rotating coordinate system was proposed, which describes acceleration, deceleration, and load change modes. In addition, the main malfunctions, monitored parameters, and methods of vibration, temperature, electrical, and spectral diagnostics were summarized. To predict the technical condition, the use of the ARIMAX model was justified, taking into account phase currents, temperature, speed, and electromagnetic torque. Within the scope of this study, a multilevel structure of a digital twin is proposed, comprising physical and sensor levels, peripheral processing, a time-series repository, an analytical core, and visualization tools. Scientific novelty. A concept for a hybrid digital twin of an elevator electric drive has been developed, combining mathematical models of electromechanical processes, degradation modeling, stochastic forecasting, and real-time analysis of sensor data. The proposed approach ensures comprehensive consideration of the system’s electrical, mechanical, and temperature parameters. Practical significance. The proposed structure can be used to create a condition-based maintenance system for elevator equipment, enable early detection of degradation changes, predict remaining service life, and generate maintenance alerts. The results obtained are primarily aimed at practical applications.