W.-B. Lin, C.-M. Lee, K.-R. Shih, and C.-H. Huang (Taiwan)
Lifting scheme, Box-Jenkins approach, short-term load forecasting, multi-revolution analysis (MRA), seasonal autoregressive integrated moving average model (SARIMA).
The paper presents a new algorithm embedding lifting scheme into Box Jenkins approach for short-term load forecasting. The lifting scheme is a general and flexible method for the construction of bi-orthogonal wavelets entirely in the spatial domain. Based on wavelet multi revolution analysis (MRA), the original load series are decomposed through lifting scheme into different sub series at different levels of revolution, which show the different frequency characteristic of the load. Then the sub-series are forecast by Box-Jenkins approach- seasonal autoregressive integrated moving average model (SARIMA). Finally, the forecasting results in different levels are reconstructed to the prediction of final load by the inverse lifting scheme. In the paper, the coiflet 12 wavelet factored into lifting scheme step is complemented. To confirm the effectiveness of the proposed approach, the approach has been tested on the prediction of 24-hours ahead through the daily load data provided by Taipower Company. Test results reveal that the performance of the proposed approach is better ensured; improving the load forecast accuracy significantly.
Important Links:
Go Back
