Computational Intelligence in Smart Grids

A smart electric power grid increases the connectivity, automation and coordination between generating plants, the networks and consumers by intelligent use of information. This paper focuses on the generation, transmission and distribution parts of the grid. These networks are nonlinear non-stationary stochastic systems and have expanded to become too complicated for humans to operate safely during severely changing conditions, as evidenced by catastrophic blackouts. In addition high power semiconductor switches have found application in power networks but the high switching speeds of these devices have not yet been fully exploited to stabilize a power grid in distress. Computational intelligence (CI) can smarten a grid in many ways. In the first place offline CI-based methods can assist the operator with anticipatory information. Furthermore, when the grid is stressed, CI-based multi-variable nonlinear optimal controllers can control the grid more reliably and more rapidly than humans, without requiring a traditional mathematical model of the grid. Some applications, as well as some ideas of smarts, are described. Keywords-Computational intelligence, smart grid, neural networks, adaptive critic designs, fuzzy systems, evolutionary computation.