Intelligent digital twin utilization for real-time forecasting and optimization of the ship's power system

The paper presents the concept and mathematical model of an intelligent digital twin of a ship’s power system, designed for real-time operation. The proposed solution integrates dynamic energy balance modeling, telemetry signal processing using a Kalman filter, load forecasting with long short-term memory (LSTM) neural networks, anomaly detection mechanisms, and optimization modules. The digital twin is implemented as a modular software architecture capable of integration with onboard control systems and cloud-based fleet analytics platforms. A series of computational experiments in MATLAB/Simulink simulates both typical and critical operational conditions, including stable load, overloads, generator failures, voltage instability, and energy-saving modes. The results demonstrate strong convergence between simulated and computed values, as well as timely system responses to emerging anomalies and effective optimization decisions. The developed model highlights the potential of digital twin technology to enhance energy efficiency, operational reliability, and environmental sustainability in modern maritime transport. It provides a foundation for advanced autonomous energy management and supports compliance with evolving IMO decarbonization and safety requirements.