Generation of Two Turbine Hill Chart Using Artificial Neural Networks

Abstract

The hill chart is an important tool for the study of the turbine performance, the energy production as well as management and hydropower control. This paper propose a model to generate two hill chart based on feed-forward Artificial Neural Network (ANN-FF). The dataset used for training the ANN-FF model is obtained from a small-scale test model of hydroelectric turbine, installed on the Madeira River in the state of Rondonia, Brazil. Predicted values obtained by applying the proposed ANN-FF model for each parameter is similar to the values measured from the small-scale test model of the turbine. The training errors of the proposed ANN-FF model have significant values from the third decimal point. It is concluded that ANN-FF is a good strategy for the generation of hill charts for the study of hydroelectric turbine efficiency.