A recent increment in power load and the fast reconciliation of sustainable power sources into the power grid requires an improvement in situational awareness. Furthermore, with this increase in power load, the aging grid infrastructure is under stress more than ever causing more event and fault occurrences on the grid. To avoid major faults and blackouts, better monitoring and optimal use of these assets has become necessary. With the advent of phasor measurement unit(PMU) technology, high resolution measurements of the power system has been made possible. However, the vast amount of data generated by PMUs brings the challenge of effectively leveraging the useful information.This work presents unsupervised learning methods to detect critical events taking place on the grid using historical PMU measurements. The most prominent feature of this method is that it does not require prior knowledge of disturbance samples or grid topology information as opposed to the methods present in the existing literature. Different categories of events are proposed based on visual characteristics of the data and event detection using existing unsupervised anomaly detection methods have been studied. Further, a meta heuristic method particle swarm optimization is explored in order to improve the performance of one of the detector.