- Apache Spark and MLlib were suitable raw data analysis, processing and modelling tools.
- Raw data contains many outliers, primarily due to zero power production (wind speed < 3 m/s).
- March, August and November are the top three months with significant average power production. Conversely, the warm months in Turkey (from April to July) present lower wind speed averages.
- The average wind speed decrease during the morning and noon.
- Power generation is strongly correlated with wind speed and not with wind direction.
- The power production is asymptotic to 3500 kw for higher wind speed than 13 m/s, approximately.
- Most higher wind speeds have a direction between 170-225° and 15-90°.
- The best performance metrics were obtained using the GBTs model (R2 = 0.9768). However, this model could produce inaccurate predictions for wind speeds lower than 5 m/s.
- The use of an ensemble model is encouraged: theoretical power for wind speed lower than 5 m/s and GBT model for higher.
Wind energy is a crucial source of renewable energy that has seen rapid growth in recent years. However, the performance, efficiency, and reliability of wind turbines are often affected by variable wind conditions. To address this issue and optimize the operation of wind turbines, it is essential to carries out big data analysis and power prediction. In this way, insights of optimal operating conditions can be found to increase the power productivity of this equipment.
The development of data analysis and power prediction model based on wind features was performed using the following steps:
- Real-data collection.
- Data cleaning and formatting.
- Data analysis: Data distribution, Data visualization, Correlation heatmap, Violin Plot, Outlier handling.
- Machine learning development: Gradient-Boosted Trees (GBTs), Generalized Linear Regression, Decision Tree Regressor, Random Forest Regressor, Linear Regression, FM Regressor, Isotonic Regression.
Big data analysis and power prediction of wind turbines offer various benefits to stakeholders, contributing to efficient wind energy production and a sustainable future. Key benefits for each stakeholder group include:
- Wind farm owners/operators: Improved decision-making for maintenance, grid management, and energy production. Increased efficiency and reduced operational costs.
- Energy consumers: Stable and reliable energy supply. Lower energy prices due to improved efficiency.
- Government/regulatory agencies: Achievement of renewable energy targets. Progress towards greenhouse gas emission reduction goals.
- Turbine manufacturers/technology providers: Enhanced wind turbine technologies driven by data analysis. Competitive advantage in the market.
- Researchers/academic institutions: Opportunities for research, innovation, and development of new models/algorithms. Collaboration with industry stakeholders for practical applications.
