The objective of this proposal is to explore the feasibility of exploiting recent advances in the understanding of power system voltage relationships and phenomena, and in the structuring and application of artificial intelligence methodologies, for control of power system voltage profiles. Since the relationship between voltage control means and their effects are predominantly local in nature, hierarchical structures are naturally indicated. The complexities of the problem, including mixed mode controls and radical nonlinearities, have defeated practicable implementation of conventional control synthesis. Recent advances in artificial intelligence methodologies, including artificial neural net applications, suggest that their application to this problem might be feasible. We propose to develop an appropriate problem definition, and to assess possible training and control strategies, including several that have recently been developed or suggested in the literature.
