.png){kind=icon}
.png){kind=icon}
Return
A scalable method with synchronous parallelization for computing selected eigenvalues of large-scale power system model
.png){kind=icon}
PRE
AI
DOI：10.1016/j.epsr.2024.111085.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
For small signal stability analysis of power systems, computing eigenvalues of the state space model is a widely used method, but still worthy of study due to computational issues in practical application. For large-scale power systems, high dimension of state space model, unknown distribution of eigenvalues and requirement on computing speed make computation of eigenvalues a challenging task. A computationally efficient method is proposed with utilizing synchronous parallelization for finding eigenvalues of concern in a flexibly specified area on the s-plane. The designed parallel framework achieves exact workload balance among computing units, which is attributed to the kernel eigenvalue solver implemented by Krylov-Schur factorization with fixing subspace dimension and discarding restart process. Satisfactory parallel speedup and numerical stability are obtained. Reliability for finding all target eigenvalues and parallel scalability of the parallelization are validated by numerical experiments on three power systems with different scales.
Keywords:
Eigenvalue computation
Krylov-Schur factorization
Parallel computing
Small signal stability
Synchronous parallelization
Journal
IF:
4.2
Papers:
1.1W
Citations:
2.2W