Application of an improved BHESS-BR method to the small signal stability analysis of power systems

Abstract

The BHESS-BR method, which consists of BHESS reduction and BR iteration, has been used in the complete eigenvalue analysis for power system small signal stability. Despite more efficient than the QR method, the BHESS-BR method has its numerical stability degrading significantly as the matrix order grows. In order to solve this issue, this paper presents an improved BHESS-BR method to analyze the power system small signal stability. For improving accuracy, orthogonal similarity transformations are taken to perform column eliminations in both BHESS reduction and BR iteration. Furthermore, a new multiplier tolerance criterion is derived to improve both accuracy and convergence by means of restricting the 2-norm condition number of similarity transformation matrix. To accelerate the computing speed, row eliminations in the BR iteration are refined so that the occurrences of row spikes are prevented. Test results of three standard systems and three practical systems with up to 4720 state variables demonstrate that the improved BHESS-BR method is more accurate and robust than the original BHESS-BR method for the complete eigenvalue analysis of small signal stability in power systems. Because the improvements belong to the scope of algebraic eigenvalue problem, the proposed method can well benefit the small signal stability analysis and offers a new choice for other eigenvalue problems in power systems. Copyright © 2014 John Wiley & Sons, Ltd.