Machine Learning Based Synchrophasor Data Analysis for Islanding Detection

Abstract

Twentieth century has witnessed a tremendous growth in the share of renewable resources in the power grid. Along with the various benefits, this has also raised several technical challenges as well as concerns in system operation, control and protection. These concerns have led to the introduction of advanced metering and protection equipments such as Phasor Measurement Units (PMUs). The highly sampled ‘Big Data’ recorded by PMUs contain significant information about the health of the system. Efficient and timely analysis of this data can address most of the power system concerns such as voltage stability, power system modeling, fault event monitoring, unintentional islanding, state estimation etc. This paper presents a method to detect unintentional islanding using machine learning technique on PMU data. A grid connected PV system is simulated in SIMULINK for data generation. Machine Learning algorithm is adapted to train and further test the data. The results show very good accuracy on test data.