Research Objective
To develop a decentralized ML-assisted protection workflow for AC microgrids by using local voltage/current features to detect, classify and isolate faults under grid-connected and islanded operation.
System Architecture
The model includes AC microgrid feeders, inverter-based sources, BESS, renewable generation, loads, relays, breakers, fault events and exported measurement channels for machine-learning feature extraction.
Simulation Methodology
PowerFactory simulates LG, LL, LLG, LLL and high-impedance faults at different locations and resistances. Local features are used for ML classification while breaker status and relay response are compared for coordination validation.
Validation Scenarios
- Grid-connected and islanded operating modes
- Different fault types and fault resistance levels
- Feeder and bus fault-location variation
- Voltage-current feature extraction for local relays
- Protection decision and breaker clearing-time evaluation
Expected Graphs and Result Discussion
A complete result section should include the main waveforms, controller response, operating status and comparison tables needed for engineering thesis documentation. For this project, the important graph set includes:
- Three-phase voltage/current waveforms
- Sequence current features
- Fault classification confusion matrix
- Relay/breaker operating status
- Detection time and protection selectivity
Thesis and Research Extension Ideas
XGBoost, Random Forest, SVM, CNN-LSTM and explainable AI can be added to strengthen the thesis and SCI paper contribution.