Machine Learning Based Prediction Model for Voltage Fault in the Hybrid AC/DC Microgrids

Recently, the use of alternative energy sources such as renewable energy sources has increased; hence, the incorporation of various monitoring, measuring, and communication infrastructures into modern power system networks offers the opportunity to build resilient and efficient hybrid AC/DC microgrids.

However, it creates various threats of instabilities in the form of voltage instability, power quality (PQ) disturbance, among others, to the complex network. Faults are the major issues that can badly affect the voltage stability in the microgrids and lead to protection system failures in the microgrids. Hence, it is essential to efficiently monitor the voltage stability in real-time by detailed data analytics on voltage measurements acquired from the sensor units.

In practice, the fault properties are random, and the topology of the microgrid may change before/after the fault clearance. Under such circumstances, online fault monitoring and timely prediction of the voltage instability can improve the resilient operation of the hybrid AC/DC microgrids.

In the proposed research work, a new data-driven methodology will be proposed for the timely prediction of the post-fault voltage stability in the hybrid AC/DC microgrids using the framework of the multi-agent deep reinforcement learning (MA-DRL) approach.