Wind-Hydro coordination

This paper reports the application of neural networks denoted “autoencoders” in order to reduce the dimension of the search space in complex optimization problems. This allows a more efficient search by meta-heuristic algorithms, with a reduction in computing time and an improvement in the quality of results. The technique, coined as miranda, is illustrated with an application of an EPSO (Evolutionary Particle Swarm Optimization) algorithm to problems of medium term wind-hydro coordination, where the operation of cascading river dams with pumping-storage capability must be combined with decisions on the available wind power generation, depending on tariffs and market prices. One shows that an EPSO running of a reduced space generated by an autoencoder with solutions evaluated in a reconstructed space runs many times faster to obtain the same results as an EPSO running in the original problem space.

Tr 2008 LCosta B mir subm.pdf — PDF document, 237Kb