State-space partitioning method for composite power system reliability assessment

Abstract

This study describes an efficient technique for reliability assessment of composite power generation-transmission system. The state-space partitioning (SSP) method combines the advantages of the state enumeration method (SEM) and the Monte Carlo simulation (MCS). The SEM is quite effective at handling low-order contingencies that constitute a significant portion of the total state space. However, it is computationally inefficient for higher order contingencies. On the other hand, the MCS is very effective at sampling larger state spaces. With the SSP, the system state space is partitioned into two regions. The system states are first enumerated in descending probability order using the fast sorting technique (FST). Then the remainder of the state space is sampled using the MCS to catch low-probability states. Since the likelihood of encountering failed states in the non-enumerated part of the state space is higher than in the full state space, the simulation converges much faster than a direct MCS (DMCS). A comparison of SSP with the SEM and DMCS using IEEE reliability test system shows that the SSP considerably reduces the computational effort. The effectiveness of the proposed method is also demonstrated on a part of the China southern power grid.