Title :
The multi-stage combination farecasting model TA-PS for energy demand
Author :
Chen, Wei-Dong ; Zhu, Peng-fei ; Guo, Qi
Author_Institution :
Inst. of Syst. Eng., Tianjin Univ., Tianjin, China
Abstract :
With certainty plus a random time series analysis method, model TA is created with a combination of the trend analysis and ARMA. At the same time, using principal component analysis of input variables pre-set to eliminate the factors that affect the overlap between the information, support vector machine regression model, energy demand forecast to be the PS model. Then, model TA combining with model PS, constructed serial and parallel of two multi-TA-PS-order model, proved that the optimal combination forecasting method has been forecast by the square of error and certainly not more than the combination of the individual to participate in the various forecasts square prediction error method and the minimum value. The model is verified that TA-PS series models have a high explanatory, which means that this research has reference value to the establishment of energy policy.
Keywords :
load forecasting; power engineering computing; principal component analysis; regression analysis; support vector machines; energy demand; multi-stage combination forecasting model; optimal combination forecasting method; principal component analysis; random time series analysis method; support vector machine regression model; Autocorrelation; Cybernetics; Demand forecasting; Economic forecasting; Energy consumption; Load forecasting; Machine learning; Predictive models; Statistics; Time series analysis; Energy demand; Multi-stage combination forecast; Supported Vector Machines; TA-PS model;
Conference_Titel :
Machine Learning and Cybernetics, 2009 International Conference on
Conference_Location :
Baoding
Print_ISBN :
978-1-4244-3702-3
Electronic_ISBN :
978-1-4244-3703-0
DOI :
10.1109/ICMLC.2009.5212198