AI-POWERED VELOCITY PREDICTION AND ENHANCEMENT IN INVELOX WIND TURBINE

Wind energy is a sustainable and viable alternative to fossil fuels. A novel high-performance design, I2NS2F (Integrated omni-directional Intake funnel, Natural fan, Straight diffuser, Splitter, and Flange), has been developed for wind turbines. This design features an intake funnel, natural fan, flow section, exit splitter, and flange. Four configurations were analyzed and optimized using MATLAB Simulink and Ansys Fluent.The I2NS2F design achieved a remarkable wind velocity of 53 m/s at the turbine region with an inlet speed of 5.5 m/s, outperforming other configurations. To address the intermittent nature of wind velocity, an enhanced Deep Learning (DL) model utilizing Long Short-Term Memory (LSTM) optimized with Black Widow and Mayfly algorithms was implemented for velocity prediction. Validated through subsonic wind tunnel tests using 3D-printed miniature models, the model demonstrated high accuracy, establishing its effectiveness for wind velocity prediction and power optimization in INVELOX type wind turbines.

Autorentext

Dr. K. Ramesh Kumar, Head and Senior Trainer at Garuda Aerospace's RPTO, Chennai, has 11 years of industry experience in India and Singapore, along with 7 years in academia.Dr. M. Selvaraj, Associate Professor of Mechanical Engineering, has 30 years of teaching experience across India and Ethiopia.