Research Guide Overview
Machine Learning Decentralized Protection Scheme for AC Microgrids in PowerFactory explains the related simulation project as a complete research workflow for engineering scholars, PhD research scholars, MTech and MSc thesis students. The guide connects the video/project page with objectives, model blocks, methodology, result graphs and thesis discussion.
The related implementation uses DIgSILENT PowerFactory 2024, Python ML and can be developed into a dissertation chapter, journal extension, conference paper model explanation or final-year project report.
Why This Topic Matters
Conventional overcurrent protection can struggle in inverter-dominated microgrids because fault currents are limited and direction can change. Machine learning protection supports faster classification and decentralized decision-making.
Suggested Modelling Workflow
- build an AC microgrid with DERs, feeders, breakers and relays
- simulate LG, LL, LLG, LLL and high-impedance faults
- extract voltage, current, sequence and relay features
- train a classifier for fault detection and classification
- map predicted fault decisions to breaker or protection zones
Important Simulation Outputs and Graphs
- three-phase voltage/current during faults
- fault classification accuracy matrix
- relay zone decision table
- breaker status timeline
- grid-connected versus islanded protection comparison
Thesis Writing and Result Discussion Structure
A strong thesis section should include the problem statement, literature gap, block diagram, parameter table, controller or algorithm design, simulation cases, waveform labels, baseline comparison, performance indices and conclusion. The result chapter should clearly explain why each graph proves improvement over the reference case.
Research Extension Ideas
- XGBoost, SVM and neural network benchmarking
- high-impedance fault feature engineering
- communication-delay robust decentralized protection
- online adaptive relay setting strategy
Related Project Page
The project page contains the video demonstration and full research scope. Use it with this guide to prepare the model explanation, output graph list and thesis methodology.