Related Experiment Videos
A short-term temperature forecaster based on a novel radial basis functions neural network
1Institute of Cybernetics, Mathematics and Physics, ICIMAF, La Habana, Cuba. pedro@cidet.icmf.inf.cu
International Journal of Neural Systems
|April 20, 2001
Summary
This study introduces a novel short-term temperature forecasting method using a Radial Basis Functions Neural Network initialized by a Regression Tree. This approach accurately predicts building temperatures, proving useful for electric load forecasting applications.
Area of Science:
- Building energy systems
- Artificial intelligence in engineering
- Environmental modeling
Background:
- Accurate temperature prediction is crucial for electric load forecasting in buildings.
- Existing methods may lack the precision needed for dynamic building environments.
Purpose of the Study:
- To present a novel, hybrid method for short-term temperature forecasting.
- To evaluate the effectiveness of this method for building load forecasting applications.
Main Methods:
- A Radial Basis Functions Neural Network (RBFN) was employed for temperature prediction.
- The RBFN was initialized using a Regression Tree (RT), where each terminal node informed an RBF hidden unit.
- The model utilized current coded hour and temperature as inputs to predict the subsequent hour's temperature.
Main Results:
- The proposed hybrid RBFN-RT model demonstrated effective short-term temperature forecasting capabilities.
- The method's accuracy suggests its viability for practical applications.
Conclusions:
- The developed temperature forecasting model, integrating Regression Trees and Radial Basis Functions Neural Networks, is a promising tool.
- This predictor can be successfully applied to enhance the accuracy of electric load forecasting in buildings.