Neural networks for optimal operation of a run-of-river adjustable speed hydro power plant with axial-flow propeller turbine

This paper analyzes the regulating capabilities of both turbine speed and guide vanes position in an axial-flow propeller turbine. Two neural networks are implemented in order to simulate the turbine behavior and the turbine efficiency. A maximum-efficiency-tracking algorithm is developed to set the guide vanes position. An experimental power plant built in the Hydraulics Laboratory is described. In order to validate the proposed operation control, the dynamics of this run-of- river pilot plant has been simulated. Substantial increases in the turbine efficiency have been found.