Learning-based Frequency Synchronization with NTP for Low-cost Phasor Measurement Units

Synchronization and high-resolution measurement is essential in power gird. In this respect, low cost phasor measurement units (PMUs) replacing global positioning system (GPS) will be rapidly deployed in the grid. This cost reduction can be achieved by simple network time protocol (NTP). However, previous researches do not guarantee accuracy of synchronization since network delay variation is not taken into account. In this paper, a learning based frequency synchronization is proposed aiming at increasing the synchronization accuracy by learning network delay variation.