Improving the Efficiency of Multistep Short-Term Electricity Load Forecasting via R-CNN with ML-LSTM

Area of Science:

• Electrical Engineering
• Computer Science
• Artificial Intelligence

Background:

• Accurate electricity load forecasting is crucial for efficient smart grid management in urban environments.
• Existing models often struggle with the complexity and variability of commercial and residential energy consumption patterns.
• Intelligent grid operations require robust forecasting to optimize resource allocation and ensure customer financial benefits.

Purpose of the Study:

• To develop an innovative and efficient framework for short-term electricity load forecasting.
• To improve the accuracy of power consumption predictions for smart grid applications.
• To enhance operational strategies in smart cities through precise energy management.

Main Methods:

• A two-phase framework was developed, starting with rigorous data cleaning and preprocessing.
• A deep Residual Convolutional Neural Network (R-CNN) was employed for feature extraction from electricity consumption data.
• Features were then fed into a multilayered Long Short-Term Memory (ML-LSTM) network for sequence learning, followed by fully connected layers for final forecasting.

Main Results:

• The proposed R-CNN with ML-LSTM architecture demonstrated superior performance in short-term electricity load forecasting.
• Evaluated on residential IHEPC and commercial PJM datasets, the model achieved significantly lower error rates compared to established baseline models.
• The framework effectively extracts relevant features and captures temporal dependencies in electricity consumption data.

Conclusions:

• The developed framework offers a highly accurate solution for short-term electricity load forecasting in smart grids.
• This approach can significantly enhance the efficiency of electricity management systems for both residential and commercial users.
• The integration of R-CNN and ML-LSTM provides a powerful tool for intelligent energy management and resource optimization.