Enhanced Decentralized Machine-Learning Protection for AC Microgrids in DIgSILENT PowerFactory

Topic scope and research objective

A suitable research question is: how can the Power Quality & Protection approach represented by “Enhanced Decentralized Machine Learning-Based Protection Scheme for AC Microgrids Using DIgSILENT PowerFactory 2024” be evaluated using DIgSILENT PowerFactory so that fault detection time and classification accuracy are improved or maintained without creating unacceptable degradation in fault-location error?

This article approaches Enhanced Decentralized Machine Learning-Based Protection Scheme for AC Microgrids Using DIgSILENT PowerFactory 2024 from a methodology perspective so the page answers a different research need rather than repeating the main project description.

Technical focus

• decentralized local-measurement protection architecture
• grid-connected and islanded AC microgrid fault cases
• machine-learning classification using relay-zone features
• dependability, security, detection time and selectivity metrics

Model architecture and implementation plan

The workflow should remain modular enough to support baseline comparison, sensitivity testing and parameter revision.

• AC microgrid / distribution network
• Inverter-based sources and loads
• Fault event library
• Voltage/current/sequence measurements
• ML detection-classification-location logic
• Relay / breaker decision and performance logging

Recommended methodology

1. Validate normal grid-connected and islanded operating points. Relate the step to the Power Quality & Protection objective and record the relevant parameters.
2. Generate labeled faults across type, location and resistance. Relate the step to the Power Quality & Protection objective and record the relevant parameters.
3. Extract local or network measurement features. Relate the step to the Power Quality & Protection objective and record the relevant parameters.
4. Train / configure the ML protection logic and hold out test cases. Relate the step to the Power Quality & Protection objective and record the relevant parameters.
5. Measure detection time, classification accuracy, location error and selective isolation. Relate the step to the Power Quality & Protection objective and record the relevant parameters.

Study cases for comparative research

• normal no-fault operation
• multiple fault classes at representative locations
• fault-resistance sweep including high-resistance cases
• grid-connected versus islanded / weak-source operation
• measurement-noise and unseen-operating-point robustness

Outputs and quantitative validation

The recommended validation evidence includes fault detection time, classification accuracy, fault-location error, selectivity / dependability. The strongest validation comes from repeating identical test cases for the reference and proposed methods, then explaining why the measured differences occur. The final discussion should also explain sensitivity to fault-resistance sweep including high-resistance cases, grid-connected versus islanded / weak-source operation.

• Fault voltage and current waveforms
• Sequence / frequency protection features
• Predicted fault class and location
• Relay / breaker status
• Accuracy, detection-time and robustness metrics

PhD and journal research extensions

• decentralized or federated protection using local measurements
• robust learning across high-resistance and inverter-limited faults
• explainable feature importance for relay decisions
• online adaptation to topology and grid-connected / islanded changes

For PhD researchers and postgraduate scholars working internationally, this topic can be adapted to a university proposal, published reference paper or independently defined research gap. The model scope can be aligned with the required software version, parameter set, dataset, disturbance profile, geometry, controller structure and reporting format while preserving reproducibility and clear technical attribution.

Continue from this research guide