Using wavelet synopsis techniques on electric power system measurements

Abstract

The elaboration of power system data is of crucial importance to the study of power system quality, control and development. Considering the widely interconnected grid and its future expansions, the integration of historical data (namely time series of various scales) to databases, implies the exponential increase of their size. Even in the case of small island systems, the logging of numerous values in sampling time of seconds can lead to similar results. Moreover, since some central assessment by Transmission System Operators (TSOs) has to be executed, the need to transmit this information over a network of given capacity also rises. To face the above issues, methods to compress time series data are examined in this paper. The wavelet synopsis techniques of Garofalakis-Kumar and the Greedy are applied for queries evaluated according to the L2 metric, while the Garofalakis-Kumar and the Selection of the Top-k Haar coefficients are used for queries evaluated according to the L∞ metric. The reconstructed time series, after the application of each synopsis technique, is compared to the original one according to specific criteria. Suggestions for further study and research are pointed out.