The basis of this project is a new and deep partnership between Steve Potter, a world pioneer in searching for functional capabilities of neural circuits in vitro ("living neural networks, LNN"), and the Venyagamoorthy team, which has led the application of adaptive, anticipatory optimization to components of the electric power grid.
The two groups are combining together to address the challenge of spatial complexity. Previous work on LNNs has focused on challenges like managing a single control variable, but electric power grids entail thousands of interconnected variables which must be managed in real-time. The new work in vitro will probe the ability of LNNs made up of thousands of neurons and glia to predict the behavior of a complicated power grid simulator, and test the ability of new biological learning models to explain their capabilities. New mathematical concepts for how to cope with complexity will also be tested in addressing the same prediction challenge, and in attempting to apply adaptive, anticipatory control for the first time to large scale power grid control in simulation. Testing on commercial electric power grids will mainly occur through their collaborations with Mexico, Brazil, China, Nigeria, Singapore and South Africa.
The use of wind power to displace coal and reduce CO2 emissions is currently limited to about 20%, because of the lack of anticipatory optimization (and optimal time-shifting, as demonstrated in the work of Venayagamoorthy et al.) and storage. If combined with adequate storage, the new algorithms aimed at here should make it possible for both China and the US to assimilate enough wind (or solar) power to be able to zero out their emissions of CO2 in power generation. It currently appears that the US and China both have enough onshore wind resources to make this possible.
