A Hybrid Quantum-Classical Machine Learning Approach to Offshore Wind Farm Power Forecasting

Abstract

Wind energy holds a significant position among renewable energy sources. Wind turbines generate electricity by harnessing wind power, and wind farms, typically consisting of several turbines, are commonly employed. The location of wind farms can greatly influence the efficiency of the energy produced. Offshore wind farms, in particular, offer various advantages over onshore installations. In a well-optimized energy production process, the in-tegrity of the grid structure enhances the operational efficiency of wind energy. Therefore, accurately pre-dicting the energy output of wind farms is critical. While classical machine learning (ML) approaches, such as regression and deep learning, are widely used for wind power forecasting, these methods often require large datasets and substantial computational power. Although parallel computing methods can help improve performance, they are typically limited in scope. In contrast, quantum computing represents a new computational paradigm, offering advantages in natural parallelization and efficient data processing. This paper proposes a hybrid quantum-classical model for power forecasting of offshore wind farms. In the proposed model, a quantum neural network is employed as a feature extractor, while support vector regression performs the forecasting task. This study marks one of the initial steps in exploring the opportunities offered by quantum computing beyond classical methods in wind power forecast. The results demonstrate the applicability of quantum-classical hybrid models to computationally intensive problems such as energy production.