Comparative Analysis of Integral-Based Indices for On-line Assessment of Power System Transient Stability

Abstract

This paper assesses the ability of four integralbased indices calculated using the post fault rotor angles, speed deviations, and accelerations to evaluate power system's transient stability. First, the impact of the integration time for the calculation of each of the indices to provide an adequate assessment of the system stability status is assessed. Then, a more detailed evaluation and calculation of the accuracies achieved by the indices is done by looking into the direct relationship between the index values and the stability status of test simulations. Results show that the proposed indices are able to represent correctly the instability degree of the system to different extents, while at the same time identifying the limitations in their use.

Composite Index for Comprehensive Assessment of Power System Transient Stability

“…In this paper, four individual stability indices describing the change in generator rotor angle, rotor speed and acceleration during the transient are defined and calculated first, as discussed in detail in [24], and then used as the sub-indicators for constructing the novel composite transient stability index. The weights for these four indices are selected using PCA to provide an accurate and automatic method for constructing the composite index without the need of human expert input.…”

“…Four sub-indicators, calculated as the integral of some postfault properties of the generators, including acceleration, speed deviation and rotor angle, over an integration period [24], are used as contributing components of the composite index in this paper.…”

“…Although according to (1), ( 2) and [24], ACI and AAI are linearly dependent when the integration period is constant, AAI would yield larger values, resulting in more obvious phenomenon for stability identification and accuracy calculation. Additionally, when the fault duration of each of the simulations is used as the integration period, according to Section III-B, this value is determined randomly for different simulations by a normal distribution with a mean value and standard deviation of 14 cycles and 6.67%, respectively.…”

“…This demonstrates that the four indices are sufficient to reflect the power system dynamic behaviour. Further discussion on the effectiveness and adequacy of the four sub-indicators for representing the dynamic behaviour of the system can be found in [24].…”

“…A slightly higher accuracy can even be observed for lower values of integration periods, for example, 500 ms. This is due to the low accuracy of index pair II (RAI and SDI) for shorter integration periods, as can be seen in [24]. As a result, neglecting the sub-indicators that contribute little to the composite index according to the PCA method can alleviate the computational burden while preserving the accuracy for longer integration periods, while at the same time this can increase the accuracy of classification of the composite index for shorter integration periods.…”