Objective. To conduct a dynamic analysis of the factors affecting the reliability of inverters used in large-scale solar power plants and to develop predictive monitoring algorithms for their technical condition. Methods. The study employed methods of systematic classification of reliability factors, thermal, electrical, and mechanical analysis, as well as machine learning techniques based on autoencoders for anomaly detection. Sensor technologies and IoT architecture were utilized for real-time data acquisition and processing. Results. A classification of factors influencing inverter reliability was developed, including an assessment of their sensor monitoring capabilities. An adaptive system architecture for technical condition analysis was constructed, incorporating a block diagram of dynamic monitoring. An Autoencoder + Threshold-based Anomaly Detection model was proposed to evaluate the inverter health index in real time, enabling early detection of potential failures. Conclusion. The proposed approach enhances the reliability and operational efficiency of centralized inverters by implementing an intelligent monitoring system. The use of predictive analytics and sensor-based architecture contributes to reduced maintenance costs, improved operational stability of solar power plants, and preemptive failure detection.
Rakhmonov et al. (Tue,) studied this question.