Leveraging Deep Learning and Hybrid Models for Enhanced Electricity Load and Price Forecasting in Deregulated Utilities Market

In a deregulated electricity market, accurate electricity Load price forecasting (ELPF) is essential for participants to manage volatility and maximize profits. This study introduces a novel approach that leverages various advanced deep learning models, including Deep Belief Networks (DBN), Enhanced Convolutional Neural Networks (ECNN), and Temporal Convolutional Networks (TCN), to enhance forecasting accuracy. We also explore hybrid models that combine Long Short-Term Memory (LSTM) and Gated Recurrent Units (GRU) with traditional forecasting techniques like ARIMA to better integrate external factors such as weather, electricity consumption, and natural gas prices. The proposed hybrid model incorporates a robust feature selection technique to refine the input data, allowing for more effective prediction of price spikes in both occurrence and magnitude. The forecasting engine utilizes probabilistic neural networks to process selected features, ensuring a comprehensive analysis of the complex relationships inherent in electricity pricing. Our experimental results, validated with real data from the electricity markets, demonstrate that the proposed models outperform traditional techniques, achieving significant reductions in mean absolute percentage error. This research contributes valuable insights into optimizing electricity price forecasting methodologies in competitive electricity markets.