How to Get Real Benefits from Real Intelligence in the Grid

Abstract The concept of "intelligent electric power grid" has mutated in many ways since the joint workshops of NSF and the Electric Power Research Institute (EPRI) developed a new vision for the future, in 2002. New White House policy charges NSF with a specific goal: to apply true intelligent optimization to slash the cost of shifting to 80%renewable electricity and 80% use of plug-in hybrid cars, without compromising reliability. There are many new opportunities for useful work, developing new algorithms together with new hardware, discussed in IEEE Computational Intelligence Magazine, August 2011, cited in the White House policy. Another key opportunity emerges from a new paper by Liang, Harley and Venayagamoorthy: to cut the effective total cost of PV and wind farms in half, by performing dynamic stochastic optimal
power flow (DSOPF) every 2 seconds (instead of the usual 5-15minutes), so that the power electronics in these farms can be used as an asset in stabilization and dispatch rather than a cost. This requires use of massively parallel chips, like cellular neural networks, to speed things up, and more advanced RLADP -- an ambitious but attainable research goal, probably worth at least $1-2 trillion per year to humanity.