Widespread application of microgrid technology and clean energy in recent years has brought people a challenge of how to eliminate the faults caused by ground harmonic currents for multi-type non-linear loads and cable lines or local distributed power generation systems. As the harmonic components of the ground current are complex, there is no clue to the different types of ground faults and fault locations. In the case that the fault data cannot be processed in advance, this paper proposes a model integrated with the CMAC (Cerebellar Model Articulation Controller) and the constraint conditions to tackle this mess of the ground harmonic current detection and compensation. Here is a network design defined in this way that CMAC is used to detect harmonic current. Next, an example of concrete microgrid earth fault is cited for the network training. Experimental results reveal that the CMAC can effectively analyze and estimate the ground harmonic components

CMAC, Ground Fault, Ground Harmonic Current