Abstract:
Fault detection in a microgrid is a challenging task. Specially, in a dc microgrid, protection system has stringent requirements. This is due to the fact that the dc fault current has a very high rate of rise. There is an immediate need to detect and isolate the fault in a dc microgrid before it reaches the peak value that typically takes around 5 to 10 milliseconds. The high value of fault current can damage the converters installed at various buses in the system. Hence, the protection system is required to operate the circuit breakers within 2 to 3 ms to isolate the fault and prevent harm to the sensitive power electronic devices.
This thesis includes working on a 5 bus ring type dc microgrid that contains distributed sources of energy such as solar plant, wind energy system, ac grid and battery energy storage. Single line to ground fault is created at different locations in the network and analyzed for varying values of fault resistances. 2 different time domain techniques have been used for protection namely – current derivative based and parameter estimation based method. These suffer from certain disadvantages as mentioned in subsequent chapters and hence cannot be fully relied on for fault detection in a dc microgrid.
This thesis proposes a novel frequency domain based fault detection technique that works on the principle of finding IMF (Intrinsic mode functions) which is then operated upon by the Hilbert Huang Transform to find the variation of frequency with time. Single line to ground fault is created on various cables in the system and frequency spectrum is analyzed for various values of fault resistances
