Enhancing Energy Forecasting in Combined Cycle Power Plants using a Hybrid ConvLSTM and FC Neural Network Model

Abstract

Accurate forecasting of net hourly electrical energy output (EP) within Combined Cycle Power Plants is critical for optimizing operational efficiency and resource allocation. In this research, we propose a groundbreaking hybrid model that merges Convolutional Long Short-Term Memory (ConvLSTM) networks and Fully Connected (FC) neural networks. This innovative approach harnesses temporal and spatial intricacies, revolutionizing energy output predictions.

The objective of this study is to introduce and evaluate the performance of the proposed Hybrid ConvLSTM and FC layers model in energy forecasting. By merging temporal and spatial dimensions, we seek to improve prediction accuracy and contribute to informed decision-making within the energy sector.

Our model integrates ConvLSTM layers to capture temporal dependencies and convolutional operations to unravel spatial relationships. FC layers enhance predictive accuracy, combining temporal insights with spatial awareness. Evaluation metrics encompass r2_scores, Mean Squared Error (MSE_values), Root Mean Squared Error (RMSE_values), and Mean Absolute Error (MAE_values).

Empirical analysis demonstrates the Hybrid ConvLSTM and FC layers model's superior predictive capabilities. It exhibits competitive r2_scores, achieving a value of 0.933192. With low MSE_values of 19.080595, RMSE_values of 4.368134, and MAE_values of 3.559722, the model outperforms existing techniques.

The proposed model surpasses traditional linear approaches by capturing non-linear relationships. It excels over existing deep learning models by assimilating both temporal and spatial dimensions. The hybrid architecture enhances generalization, mitigating overfitting concerns.

This research introduces a novel Hybrid ConvLSTM and FC layers model, enriching energy forecasting by bridging the gap between temporal and spatial complexities. The model's ability to integrate multi-dimensional insights empowers stakeholders with actionable information, paving the way for informed energy production decisions. This hybrid approach underscores the potential to reshape the energy landscape by optimizing operational strategies and fostering sustainability.