Small signal stability constrained rescheduling using sensitivities analysis by Neural Network as a preventive tool

Small signal stability that appears as low frequency oscillations play an important role when increasing power transfer in interconnected power systems. In this paper, a novel approach for generation rescheduling constrained to small signal stability based on neural network sensitivity analysis is presented. The proposed method considers system minimum eigenvalue as stability index for generation rescheduling. A Security Analyzer Neural Network (SANN) is trained to estimate minimum eigenvalue associated with each generation pattern. The main motivation for neural network application is its ability for sensitivity analysis of eigenvalue. For any given operating point of power system, SANN evaluates minimum eigenvalue using corresponding generation pattern as input. Then sensitivity of the minimum eigenvalue with respect to generation pattern is used as a measure for changing generator outputs in order to increase system damping. Effectiveness and applicability of proposed approach is demonstrated on WSCC 9-bus test system.