Artificial Neural Networks for the Management of Dokan Reservoir Operation Under Stochastic Conditions

Abstract

This paper presents a decisional support variant of heuristic type intended for the management of the hydroelectric power station, in particular to Dokan hydropower reservoir. It is based on ANN of the simplest topology, which relates a decision variable to a series of deterministic and stochastic parameters influencing it. A comparison of operating performances based on ANN with those resulted from GA and consisting in equations of non-linear regression was presented, also analysis of ANN rules performances under normal hydrological conditions with no extreme flood periods is given.