Virtual inertia regulation based high-order observer for weak microgrids using deep sequential learning

Area of Science:

• Electrical Engineering
• Control Systems
• Renewable Energy Integration

Background:

• High penetration of Sustainable Energy Resources (SERs) in Hybrid Micro-grid Systems (HMGS) reduces inertia due to power electronic interfaces.
• Low inertia in HMGS leads to frequency instability and challenges in grid management.

Purpose of the Study:

• To introduce an adaptive virtual frequency controller (VFC) for Energy Storage Systems (ESSs) in low-inertia HMGS.
• To enhance the dynamic stability and responsiveness of HMGS with high SER penetration.

Main Methods:

• Implementation of a high-order extended disturbance observer (HOEDO) for disturbance estimation and compensation.
• Application of deep sequential action-value learning (SAVL) for unsupervised controller parameter adaptation.
• Emulation of virtual rotor dynamics in control loops for improved responsiveness.

Main Results:

• The adaptive VFC effectively improves system inertia and damping characteristics.
• The proposed scheme demonstrates superior performance compared to conventional virtual inertia control methods.
• The controller exhibits faster dynamic response and suppressed fluctuations under varying conditions.

Conclusions:

• The adaptive VFC, utilizing HOEDO and deep SAVL, offers a robust solution for stabilizing low-inertia HMGS.
• This approach enhances grid resilience and facilitates higher integration of SERs.
• The method provides a significant advancement over existing control strategies for microgrid stability.