Hybrid Framework for Wind Direction Prediction Using Custom Quantum Feature Extractor

Wind energy along with harnessing wind power for electricity generation, wields added benefits of low installation costs, sustainability and lesser carbon footprint, leading to staggering increase in wind plant installation across the globe.While research in this domain is widespread, this paper explores the technical dimensions of this field, focusing on a vital process of yaw angle prediction in wind turbines.Yaw prediction in wind turbines is essential for aligning turbines with prevailing wind patterns to maximize energy output.This paper proposes a hybrid framework that leverages Custom Quantum Feature Extractor (CQFE) with machine learning and deep learning models for wind direction prediction.The predicted wind direction aids us with determining the probable yaw angle in future and make necessary changes in turbine direction.CQFE along with Principal Component Analysis (n=32) and CNN+GRU model performed promisingly with an R2 score of 0.866, Mean Squared Error of 1160.65 and Mean Absolute Error of 13.02.CQFE along with Principal Component Analysis (n=16) and Ridge Regression model performed promisingly, trained in walk-forward method, with R2 (mean ± std) of 0.85659 ± 0.06661 and MSE (mean ± std) of 0.00868 ± 0.00504.While this paper focused much on a single wind mill in isolation, further research scope includes extending similar principals to large wind farms, where impact analysis of complex wake effects and advance optimization techniques are much required.