Advanced Modelling of Full Characteristic Curve of Tubular Turbine Based on Back-Propagation Neural Network

Abstract

For the comprehensive characteristics of hydro-turbine, it is generally grasped through the full characteristic curve map of hydro-turbine obtained from the experiment. However, due to the factors such as time cost, fund cost, experimental difficulty and safety, the experimental results cannot hold onto each working condition of hydro-turbine. And for the specific conditions of unknown working conditions, linear interpolation needs to be carried out through the existing experimental results, which will have a certain hidden danger to the safe and stable operation of the turbine. This paper uses the advantages of neural network to predict the overall operation space of hydro-turbine through back-propagation neural network under the condition of existing experimental points. The results show that the neural network method can effectively ensure the accuracy of the operation characteristics of the hydro-turbine, which can guide the safe and stable operation of the hydro-turbine more clearly, and ensure the efficient and stable operation of the hydro-turbine.