AI-Driven Real-Time Condition Monitoring and Reliability Enhancement of Centralized Inverters in Large-Scale Solar Power Plants

Abstract

Relevance: Reliability degradation and fault diagnosis limitations significantly affect centralized inverter stability
in large-scale solar power plants.
Objective: This study investigates modern SCADA-, AI-, and thermal-model-based approaches for real-time
condition monitoring, fault diagnosis, reliability assessment, and fault-tolerant operation of centralized inverters
operating under dynamic renewable energy conditions.
Methods: Comparative analysis of degradation models, SCADA monitoring, machine learning diagnostics, and
Neutral Point Shift compensation methods.
Results: Hybrid monitoring approaches improved fault detection accuracy, enabled adaptive post-fault recovery,
reduced unplanned shutdown risks, and enhanced inverter operational reliability under variable operating
conditions.