Sparse representation algorithm applied to power systems signal compression

Abstract

Power system signal compression is an important task due to the nature of electrical disturbances, which must be analyzed offline to extract useful information. This paper presents a comparative study of greedy algorithms for obtaining sparse representations of signals applied to power system signal compression. The algorithms are evaluated based on two parameters: the quality of the compressed signal, measured by its correlation coefficient relative to the original signal, and the number of elements in the representation, which is related to the compression ratio. Additionally, computational complexity, in terms of required floating-point operations (FLOPs), is analyzed to identify which algorithm is most suitable for real-time implementation. Finally, a comparison with the wavelet transform is performed using a database of real power system signals.