Methods and Algorithm for AI-based Monitoring of the Automated Power Supply System Status in Mobile Control and Communication Complexes Under Crisis Conditions

The stable operation of mobile control and communication complexes under any challenging conditions is directly dependent on the performance of their Autonomous Electrical Power Supply System (AEPSS). This scientific article presents a hierarchical monitoring architecture designed for real-time remote acquisition of power source parameters — including voltage, current, fluctuations, and environmental metrics using PLC and Edge-AI technologies. Anomalies are detected through a hybrid model combining LSTM-Autoencoder and Decision Tree ensemble methods, while the Remaining Useful Life (RUL) of the system is estimated accordingly. Field experiments were conducted during 48-hour operational cycles of mobile power stations in simulated tactical scenarios and field exercises. During these tests, early warning notifications were issued approximately 10 seconds prior to system failure, achieving an F1-score of 97%. The proposed approach improved system reliability from 0.982 to 0.996 and reduced error messages by 42%.