Electric Load Forecasting based on Wavelet Transform and Random Forest. Issue 12 (27th October 2021)

• Record Type: Journal Article
• Authors: Peng, Li‐Ling; Fan, Guo‐Feng; Yu, Meng; Chang, Yu‐Chen; Hong, Wei‐Chiang
• Is Part Of: Advanced theory and simulations. Volume 4:Issue 12(2021)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-10-27
• Subjects: electric load forecasting -- random forest (RF) -- short‐term load forecasting (STLF) -- wavelet transform (WT); Science -- Simulation methods -- Periodicals; Science -- Methodology -- Periodicals; Engineering -- Simulation methods -- Periodicals; Engineering -- Methodology -- Periodicals; 507.21
• DOI: 10.1002/adts.202100334
• Languages: English
• ISSNs: 2513-0390

Abstract

Abstract: Aiming at the problem of strong randomness and low forecasting accuracy in short‐term electric load, a method based on wavelet transform (WT) and random forest (RF) are proposed. In the proposed method, the noise is removed by WT, and the original data are decomposed into several groups with low or high frequencies, and then the decomposed column variables are used as characteristic variables to forecast by RF. It has three advantages: 1) due to the instability of electric load data, the decomposition and denoising of WT can be used to characterize the nonstationary signal characteristics; 2) WT has more advantages in time domain analysis because of its correlation to signal removal and the tendency of noise whitening after transformation; and 3) based on WT, RF still maintains forecasting accuracy even after the features of the analyzed data are lost. Electric load data from Australian‐Energy‐Market‐Operator are taken as an example for a case analysis. By comparing with other existed methods, the results have showed that the proposed model can reduce the influence of random noise during forecasting processes and improve the associated accuracy and reliability. Abstract : Aiming at the problem of strong randomness and low forecasting accuracy in short‐term electric load, a method based on wavelet transform and random forest are proposed. Electric load data from Australian‐Energy‐Market‐Operator are taken as a case analysis. The comparing results demonstrate that the proposed model can reduce the influence of random noise and improve the accuracy and reliability.