PREDICTIVE POWER QUALITY MANAGEMENT OF MILITARY INFORMATION SYSTEMS BASED ON MACHINE LEARNING
This paper presents a predictive approach to power quality management in military information systems based on machine learning and optimization-based control. An ML model is developed to forecast voltage deviation, total harmonic distortion, and voltage unbalance using historical electrical-state data and candidate control sequences. The proposed framework integrates power quality forecasting, predictive assessment, and optimization of control actions subject to power quality and critical-load constraints. Four control mechanisms are considered: measurement-based, threshold-based, ML-based predictive, and ML-based predictive-optimization control. The evaluation combines sMAPE and RMSE with predictive time margin, integrated power quality indices, constraint-violation ratio, and critical-load power supply. The approach shifts power-supply management from reactive detection of electrical deviations toward predictive assessment and anticipatory control, providing a basis for improving the continuity and reliability of power supply to critical military information-system components.
Authors
- Natalia Amurova
Publication Details
- Journal
- Techscience.uz.
- Published
- 2026-11-15
- DOI
- https://doi.org/10.47390/ts-v4i11y2026n16
- Primary Topic
- Power Quality and Harmonics
- Type
- article
- Field-Weighted Citation Impact
- 0.00