UFNGBM (1,1): A novel unbiased fractional grey Bernoulli model with Whale Optimization Algorithm and its application to electricity consumption forecasting in China
| Autor: | Ye Yuan, Wanli Xie, Bin Pu, Fengtao Nan, Ningbo Zhu |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2021 |
| Předmět: |
Consumption (economics)
Mathematical optimization Electric power distribution Computer science business.industry Deep learning Nonlinear grey Bernoulli model Stability (learning theory) Power (physics) Electricity economics TK1-9971 Six-year-plan Nonlinear system General Energy Electricity consumption forecasting Electricity Artificial intelligence Grey model Electrical engineering. Electronics. Nuclear engineering business China |
| Zdroj: | Energy Reports, Vol 7, Iss, Pp 7405-7423 (2021) |
| ISSN: | 2352-4847 |
| Popis: | The electricity distribution in a planned way has been a difficult issue that the power supply bureau wishes to solve, and it is the lifeblood of economic development. Forecasting annual electricity consumption is very crucial for the planning of the power supply bureau and for booming economic development. We propose a novel unbiased fractional nonlinear grey Bernoulli model [i.e, UFNGBM (1,1)] to forecast China’s annual electricity consumption based on the nonlinear grey Bernoulli model [i.e, NGBM (1,1)]. First, UFNGBM (1,1) approach is designed to derive calculation formula of the novel model, and validity of the model is proved by the matrix perturbation theory. Second, a novel optimization algorithm is introduced based on the whale algorithm to find the optimal parameters (i.e., o r d e r and p o w e r ) of the proposed model. Third, the accuracy, stability, and effectiveness of our method are verified through three real-world cases in China. Finally, we collect the electricity consumption data of three provinces in China and successfully apply the proposed algorithm to predict the electricity consumption from 2019 to 2024. The experimental results demonstrate that our proposed model is significantly superior to nine alternative models on the electricity consumption data of Jilin and Jiangsu. The performance of our novel method is close to the state-of-the-art deep learning method on the electricity consumption data of Shandong. It is noticed that our method [as an extended version of NGBM(1,1)] is significantly better than NGBM(1,1) on these three real-world datasets, which further shows the effectiveness of the our proposed algorithm. Meanwhile, the electricity consumption of these three provinces in the next six years (2019–2024) is forecasted, which has a very good guiding significance and provides a more reliable reference for the economic and power bureau. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|