Back to Search
Start Over
Load forecasting via Grey Model-Least Squares Support Vector Machine model and spatial-temporal distribution of electric consumption intensity.
- Source :
-
Energy . Sep2022, Vol. 255, pN.PAG-N.PAG. 1p. - Publication Year :
- 2022
-
Abstract
- Load forecasting analysis plays an important role for regional electric power project planning as well as consumption management. For improving the long-term load forecasting accuracy and usability, this paper proposes a new residual-type combined Grey Model-Least Squares Support Vector Machine forecasting model for the component loads by extracting the load characteristics. In this model, each component decomposed load is forecasted using characteristic-matched model. Meanwhile, Moran's I and α convergence are used to identify the spatial-temporal distribution characteristics and levels of electric consumption intensity from the perspective of spatial econometrics. The results show that the proposed model can improve the forecasting accuracy, and can avoid the poor usability in the application phase. According to a case study for Beijing, the electric consumption intensity of the city has a positive spatial correlation, and the low electric consumption intensity districts are mainly concentrated in the central region, which reveals a trend of spreading from the center to periphery. The overall electric consumption intensity shows a downward trend, but the internal differences increase with time. The results may help to carry out electric construction planning, and to formulate regionally energy-saving policies. • Electric load characteristics are captured with Empirical Mode Decomposition. • The component decomposed loads are forecasted by characteristic-matched models. • Building a residual-type combined Grey Model-Least Squares Support Vector Machine model for load forecasting. • Electric anomalies and substitution potential districts are identified from spatial econometrics. • Analyzing convergence trend and level of electric consumption intensity. [ABSTRACT FROM AUTHOR]
Details
- Language :
- English
- ISSN :
- 03605442
- Volume :
- 255
- Database :
- Academic Search Index
- Journal :
- Energy
- Publication Type :
- Academic Journal
- Accession number :
- 158262785
- Full Text :
- https://doi.org/10.1016/j.energy.2022.124468