Real-time vulnerability assessment (VA) is one of the essential tasks of the so called Smart Grid, since it has the function of detecting the necessity of performing global control actions. In view of this, the present paper will introduce a novel data-mining-based approach to map post-contingency Dynamic Vulnerability Regions (DVRs), taking into account three short-term instability phenomena. Based on probabilistic models of relevant inputs (e.g. nodal loads and occurrence of contingencies), the approach applies Monte Carlo (MC) simulation to recreate a wide variety of possible post-contingency dynamic data of some electric variables, which could be directly available from PMUs in a real system (e.g. voltage phasors or frequencies). From this information, a pattern decomposition method, based on empirical orthogonal functions (EOF), is used to approximately pinpoint the DVR spatial locations. The identified DVRs are then used to ascertain the actual dynamic state relative position with respect to their boundaries, which is accomplished by using a support vector classifier (SVC). The proposal is tested on the IEEE New England 39-bus test system. Results show the feasibility of the approach in finding hidden patterns in dynamic electric signals as well as in numerically mapping power system DVRs.
展开▼
