Deep Learning
Solar Radiation Prediction
Recurrent neural network model to predict solar radiation for the next 24 hours.
Course: Neural NetworksCo-authors: Diego Quezada
Objectives
- 1Predict solar radiation for the next 24 hours using recurrent neural networks.
- 2Identify correlations between meteorological variables to improve prediction accuracy.
- 3Evaluate different temporal feature representations for time series forecasting.
Conclusions
- There is a relationship in the behavior of average hourly temperature and humidity by displacing temperature over time.
- There is a high correlation between temperature and radiation per hour.
- Using a two-dimensional representation for the time of day increases network performance.
Technologies
- NumPy
- Pandas
- Matplotlib
- Seaborn
- Scikit-learn
- TensorFlow
- Keras