NEURAL NETWORK BASED WIDE AREA COORDINATION AND LOCAL CONTROL OF ELEMENTS IN A LARGE COMPLEX SYSTEM

点击翻译按钮获取中文摘要

Many, many researchers have called out for new systems to manage the nation's infrastructure networks which would be adaptive, self-healing, integrated and intelligent. This project is a unique effort to make those words literally true, by scaling up new approaches to computational intelligence so as to provide greater performance, robustness and foresight in the control of electric power grids. The work will build on prior successes of the PIs, as described, for example, in www.eas.asu.edu/~nsfadp, in which intelligent control allowed generators to stay up and running in the face of disturbances three times as large as what forces a generator shutdown, under the previous best control schemes. It will explore new ways of scaling up to larger systems, involving multiple generators and advanced power switching components, to be managed in an integrated, optimal fashion. It will use power grids as a testbed for truly general concepts of intelligent systems, and of adaptive management of complex systems.