Electricity Load and Price Forecasting Using Jaya-Long Short Term Memory (JLSTM) in Smart Grids.

Area of Science:

• Smart Grid Technology
• Artificial Intelligence in Energy
• Data Science

Background:

• Smart grids enable consumers to modify electricity usage based on price signals, influencing price patterns.
• Accurate electricity price and demand forecasting are crucial for smart grid reliability and sustainability.
• Big data analytics is a growing area in smart grid research due to massive data generation.

Purpose of the Study:

• To develop an advanced forecasting model for electricity price and demand in smart grids.
• To enhance forecasting accuracy by utilizing multiple input variables.
• To improve the training and predictive capabilities of forecasting models through hyperparameter optimization.

Main Methods:

• Utilized a recurrent neural network (RNN), specifically Long Short-Term Memory (LSTM), for forecasting.
• Employed a multi-variable input approach to improve prediction accuracy.
• Tuned LSTM hyperparameters using the Jaya optimization algorithm.
• Preprocessed data using the z-score method for outlier and missing value removal, followed by normalization.

Main Results:

• The proposed multi-variable LSTM model, optimized with Jaya, demonstrated superior forecasting accuracy.
• Performance was evaluated using Root Mean Square Error (RMSE) and Mean Absolute Error (MAE).
• The optimized LSTM model outperformed both univariate LSTM and Support Vector Machine (SVM) models.

Conclusions:

• The integration of multi-variable inputs and the Jaya optimization algorithm significantly enhances LSTM's forecasting performance in smart grids.
• The proposed model offers a more accurate and efficient solution for electricity price and demand prediction.
• This approach contributes to improved load management and grid stability.