|
Modeling and Short-Term Prediction of National Electricity Consumption Using Recurrent Neural Network and TPE Optimization Algorithm
|
Mojtaba Zahaki Rahat *1   , Hossein Sadeghi saghdel2 |
1- Tarbiat Modares , mojtaba.zahaki@modares.ac.ir
2- Tarbiat Modares |
|
|
Abstract: (322 Views) |
| The intermediary role of electricity in various industries and its correlation with societal well-being have gained increased significance. Awareness of the demand for this energy plays a crucial role in guiding the country towards development. In recent decades, with the advancements in deep learning models and their heightened accuracy, their usage has become prevalent. In the realm of modeling and predicting electricity consumption, incorporating influential variables is crucial for enhancing prediction accuracy. Thus, in this research, variables such as non-oil Gross Domestic Product (GDP), average country temperature, holidays, and electricity consumption trends are utilized. The TPE optimization algorithm is employed to optimize the LSTM model. For result comparison, an alternative model excluding two variables, non-oil GDP and holidays, is designed and optimized using the TPE algorithm. The study results indicate that the model incorporating variables like non-oil GDP and holidays exhibits higher accuracy compared to the model without these two variables |
|
| Keywords: Forecasting, power consumption, neural network, optimization |
|
|
Full-Text [PDF 1583 kb]
(136 Downloads)
|
Type of Study: Research |
Subject:
E.Economic
Received: 2024/02/29 | Accepted: 2024/11/27 | Published: 2024/12/30 | ePublished: 2024/12/30
|
|
|
|
|
|
|
| Send email to the article author |
|
|
|
| Add your comments about this article |
|
|
|
