Artificial intelligence methods in deregulated power systems operations

With the introduction of the power systems deregulation, many classical power trans- mission and distribution optimization tools became inadequate. Optimal Power Flow and Unit Commitment are common computer programs used in the regulated power industry. This work is addressing the Optimal Power Flow and Unit Commitment in the new deregulated environment. Optimal Power Flow is a high dimensional, non-linear, and non-convex optimization problem. As such, it is even now, after forty years since its introduction, a research topic without a widely accepted solution able to encompass all areas of interest. Unit Commitment is a high dimensional, combinatorial problem which should ideally include the Optimal Power Flow in its solution. The dimensionality of a typical Unit Commitment problem is so great that even the enumeration of all the combinations would take too much time for any practical purposes.

This dissertation attacks the Optimal Power Flow problem using non-traditional tools from the Artificial Intelligence arena. Artificial Intelligence optimization meth- ods are based on stochastic principles. Usually, stochastic optimization methods are successful where all other classical approaches fail. We will use Genetic Programming optimization for both Optimal Power Flow and Unit Commitment. Long processing times will also be addressed through supervised machine learning.