Intelligent Event-Triggered Distributed Control for Islanded Microgrid: Adaptive Communication Scheduling with Reactive Power Minimization

This paper presents a Machine Learning-based Event-Triggered Mechanism (ML-ETM) distributed secondary control framework for islanded microgrids with inverter-interfaced Distributed Generators (DGs). The principal objective is to attain simultaneous voltage restoration, frequency regulation, proportional current sharing, and reactive power minimization while significantly reducing communication burden. The proposed ML-ETM framework dynamically regulates triggering thresholds based on real-time system operating conditions, enabling higher communication density during transients and sparse communication during steady-state operation. Critically, the learning module adjusts only communication scheduling while maintaining the stability-critical control law, thereby preserving theoretical stability guarantees. The framework achieves 75-85% communication reduction and 92% lower reactive power circulation with convergence times of 2.3 to 3.0s. Lyapunov-based analysis confirms global practical stability and zeno-free behavior. Comprehensive simulation studies validate robustness under constant load, multiple load disturbances, plug-and-play operations, DG faults, cyber-attacks, and packet loss scenarios. The proposed approach offers a highly scalable and robust solution for intelligent microgrid operation with significant practical benefits including reduced infrastructure requirements, improved energy efficiency, and enhanced power quality.