Search

The identification of coherent areas is an important step to fortify the power system against the potential spread of cascading failures. This processing task helps to establish the preliminary strategies to implement online control methodologies that increase the power system reliability. In this regards, a methodology to identify coherent areas based on sliding power spectral density algorithm is presented in this paper. A coherency index is estimated from self- and cross-power spectral densities information in order to identify the coherent areas for a specific frequency related to an inter-area oscillation mode. In order to facilitate and automatize the process, an identification criterion based in the coherency index is proposed. This enables the proposed approach to be applied offline or online. In addition, the computation of the self-power spectral density using a sliding window opens the possibility to detect the onset of transient processes. Synthetic signals and two test power systems with different dynamic characteristics are used to evaluate the potential application in large systems. [ABSTRACT FROM AUTHOR]

It is not uncommon for oscillatory electric power system modes to move close to a resonance in which eigenvalues coincide. In a weak resonance, the modes are decoupled and the eigenvalues do not interact. We analyze general perturbations of a weak resonance and find two distinct behaviors, including interactions near strong resonances in which the eigenvalues quickly change direction. The possible perturbations are illustrated with interactions between electromechanical modes in a 4-generator power system. Some of the interactions are similar to subsynchronous resonance and can lead to oscillatory instability. [ABSTRACT FROM AUTHOR]

Power system normal form analysis has developed co-efficients and indices in modal coordinates to quantity nonlinear modal Interactions. We study the changes in the coefficients and indices when the power system equations are expressed in different coordinates or units and show that they can be normalized to be invariant to coordinate changes and thus intrinsic to the power system. The results are illustrated on a 4-generator System. An example shows that the coefficients and indices not only detect non-linear interactions but also can become very large near a strong resonance in the system linearization. [ABSTRACT FROM AUTHOR]