Artificial Neural Network (ANN) Prediction of Guide Vane for Effective Utilization of Pump as Turbine (PAT)

This research paper investigates the application of an Artificial Neural Network (ANN) model as a soft computing approach for optimizing the settings of guide vanes in a pump operating as a turbine (PAT). Correct adjustment of guide vanes is essential for maximizing energy conversion efficiency in PAT applications. Using a dataset collected over 16 months (January 2020 to April 2021), the ANN model was trained and tested on nine operational parameters, including speed regulator, speed motor, inlet angle, wrap angle, impeller size, flow rate, and power output. Statistical measures indicated strong prediction accuracy for the ANN model: in the training phase, it achieved R2=0.999683 R2=0.999683 R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>=0.999683, R=0.99984, R=0.99984, R=0.99984, and MSE of 0.000010with RMSE of 0.00330 During testing, the model maintained high accuracy with R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>=0.94968 R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> = 0.94968 R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>=0.94968, R=0.94984 R=0.94984, R=0.94984, MSE of 0.005010, and RMSE of 0.005330. Performance metrics such as MSE, RMSE, Determination Coefficient (DC), and Correlation Coefficient (CC) further validated the model's efficacy. The ANN's predictive capability highlights its suitability for determining optimal guide vane angles, improving the efficiency of water pumps operating as turbines. This study offers a comprehensive analysis of ANN's role in enhancing PAT efficiency, showing its superiority over traditional rule-based approaches in predicting optimal vane settings and maximizing energy output.