Power system oscillations are occurring worldwide, some of which have significant consequences. The conventional model-based approaches for investigating the oscillations are meeting an obstacle in that the system is not sufficiently and precisely modelled. Such model inadequacy can hopefully be compensated by the abundance of data, e.g. PMU data, DSM data, and SCADA data, leading to data-driven approaches. There are different possibilities for mixtures of data and models in the data-driven approaches. At one extreme of the mixture, which is desired by many system operators, is the model-free approach where data alone speaks without a model needed at all. This note discusses the possibilities of the model-free method.
