Energy Systems: Technology and Management

A New Hybrid Intelligent-based Model for Building Electrical Consumption Forecasting

Document Type : Original Article

Authors

Nasser Shahsavari-Pour 1
Mojtaba Afshar-Jahanshahi 2

1 Professor Department of Industrial Engineering, Vali-e-Asr University of Rafsanjan, Iran

2 Assistant Professor of Accounting, Faculty of Administrative Sciences and Economics

https://doi.org/10.48306/estm.2025.232463
Abstract
Energy information systems are crucial for the operational optimization of smart buildings. These systems can offer benefits such as high energy-saving potential, efficiency, and intelligent services. Therefore, accurate electricity consumption forecasting requires a smart estimation strategy that considers the parameters affecting electricity consumption patterns. This paper proposes an intelligent prediction model that accurately forecasts and analyzes building electricity consumption. The proposed model includes the following components: a feature selection model based on mutual information for selecting input variables, and a deep learning time series prediction model based on Long Short-Term Memory (LSTM) neural networks to forecast the target value. The model's performance was evaluated using real-world data from a two-story smart home located in Houston, Texas, USA. A comparative analysis with other benchmark models was also conducted. The comprehensive comparison demonstrated that the hybrid model is more accurate than individual models, and the proposed intelligent model outperforms other benchmark hybrid and standalone models, as indicated by the achieved prediction performance.

Keywords

Building
Energy consumption
load forecasting
smart building
deep learning models
energy management

Home

Guide for Authors

Submit Manuscript

Reviewers

Contact Us