Dynamic security enhancement of power systems using mean-variance mapping optimization

In this paper, a preventive control action that involves both generation rescheduling and load curtailment is proposed for enhancing the dynamic security of large interconnected power systems. The control action is formulated as a security-constrained optimization problem that is solved by mean-variance mapping optimization (MVMO) integrated with a self-adaptive penalization technique and artificial neural networks to develop a fast and effective methodology. The proposed methodology is applied to a 16-generator 68-bus test system to solve the security-constrained optimization problem with both continuous and discrete decision variables. To find a proper and cost-effective solution for the control actions within an acceptable time, dynamic security assessment methodology based on artificial neural networks is integrated into the optimization process for predicting the violations of security constraints brought about by the candidate solutions. The proposed method effectively integrates a variety of popular heuristic optimization algorithms, including MVMO, differential evolution, particle swarm optimization, genetic algorithms, big bang-big crunch, and artificial bee colony. MVMO outperforms all the others in various aspects such as reliability and robustness.