Knowledge-based learning for emergency voltage control

Abstract

A new supervised genetic learning control for maintaining voltage profiles after an emergency in power systems is proposed in this study. Search efficiency is improved after introducing system knowledge into the search process of genetic learning. The optimization of the coordinated voltage control is considered a multi-objective optimal problem. Thus, a set of effective controls can be found, and a knowledge base is formed. Effective controls are stored in a long-term memory and exploited for further application. After an emergency, the stored knowledge is used to provide guidance in addition to an online learning process. The search efficiencies of genetic learning can be significantly improved. A system simulation of the New England 39-bus system shows the efficiency of the proposed genetic learning control.