Modified Segmentation Prony Algorithm and Its Application in Analysis of Subsynchronous Oscillation

摘要

Based on traditional Prony algorithm (TPA) and segmentation PA (SPA), a modified SPA (MSPA) is proposed in this paper. By dividing the whole observation window into several segments according to certain rules and then utilizing data in all of the segments, the overall sample function matrix is constructed and the average error effect of the whole observation window is taken into account. Meanwhile, a new approach to extracting oscillation modes is presented by means of Kaiser window FFT analysis and a concept of energy class. The MSPA and time domain simulation method are integrated and used in analysis of subsynchronous oscillation (SSO). Their results can be confirmed and complemented by each other. Simulation results show that the MSPA not only enhances SPA's computational efficiency and credibility, but also avoids the lost of oscillation modes existing in modified PA (MPA) which uses singular value decomposition to determine the rank of Prony model.