Short-term reliability evaluation of protection systems in smart substations based on equivalent state spaces following semi-Markov process

摘要

The reliability analysis of protection systems in smart substations helps to improve their designed functions, whose difficulty lies in the complex configuration with various components. To avoid the dimension disaster, the state spaces of the protection systems may be aggregated to several lumped states. When the states do not satisfy the balance criterion, it is newly found that the lumped states not following the Markov process introduce notable error to the short-term reliability. To avoid this, an equivalence algorithm based on the semi-Markov process (SMP) is newly proposed. It is derived that the residence time distributions of the lumped states are the weighted averages of the distributions of the states to be lumped. The kernel matrix of the SMP, the instantaneous probability, and its sensitivity, are derived from the transfer rates and distributions. The numerical results validate accuracy of the proposed algorithm, which shows that the traditional model underestimates the availability. It underestimates sensitivity of the time source (TS), while overestimates those of the others. The probabilities of the proposed algorithm are more sensitive to the TS, while those of the traditional model are more sensitive to the merging unit and the ethernet switch.