Journal
INTERNATIONAL JOURNAL OF ENERGY RESEARCH
Volume 46, Issue 2, Pages 1766-1788Publisher
WILEY
DOI: 10.1002/er.7293
Keywords
electricity production source; hybrid model; prediction; wavelet transform
Categories
Ask authors/readers for more resources
A novel combination prediction model based on wavelet transform, long short-term memory, and stacked autoencoder was proposed to predict the future growth trend of electricity production. Empirical results showed that this combination model outperformed benchmark models, indicating its potential in electricity production forecasting.
Electricity is an important indicator for economic development, especially electricity production (EP), which is electricity industry managers making strategic decisions. There are many ways to produce electricity, which is the source of rapid growth in EP is rarely studied. Due to the nonstationary and nonlinearity of the EP time series, traditional methods are less robust to predict it. In this study, a novel combination prediction model is proposed based on wavelet transform (WT), long short-term memory (LSTM), and stacked autoencoder (SAE). Comparisons between the SAE-LSTM and the advanced prediction model. We compared SAE-LSTM and the advanced prediction model including BP (Back Propagation) etc. In addition, the performance comparison of the different wavelet layers based on SAE-LSTM and the performance comparison of the EMD and EEMD based on SAE-LSTM are also compared. At last, future average growth rates (June 2021 -> December 2022) are predicted. The empirical result shows that the combination model in view of SAE-LSTM exceeds the benchmark models. The results also imply that WT-SAE-LSTM outperforms the EMD, EEMD-SAE-LSTM, and SAE. Based on the optimal orders and layers of Coiflets combining with SAE-LSTM, natural gas is the fastest-growing source of EP in the United States.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available