Data driven intelligent forecasting of electric heating and cooling loads in CCHP systems for sustainable energy management

Accurate short-term forecasting of electric, heating, and cooling loads is critical for the efficient operation of combined cooling, heating, and power (CCHP) systems. However, existing studies largely focus on single-load forecasting, employ computationally complex hybrid models, or treat load prediction as a detached task rather than an integrated component of CCHP operational planning. This study addresses this gap by proposing a computationally efficient long short-term memory (LSTM)–based forecasting framework capable of simultaneously predicting electrical, thermal, and cooling demands for CCHP systems. The proposed methodology integrates correlation-driven feature selection, noise reduction through moving average filtering, and iterative hyperparameter optimization to enhance prediction accuracy while maintaining a lightweight architecture suitable for real-time deployment. The model is evaluated using publicly available energy and weather datasets, and its performance is assessed using root mean square error. Comparative analysis against benchmark machine learning models—including Decision Trees, Random Forest, Bagging, and XGBoost—demonstrates superior accuracy and stability of the proposed LSTM across all load types. Beyond forecasting accuracy, this work establishes a forecasting-driven operational configuration that enables coordinated scheduling of electric, heating, and cooling subsystems within energy management and supervisory control and data acquisition environments. The results indicate that the proposed approach offers a practical and scalable solution for data-driven CCHP operation, supporting improved efficiency, reduced operational uncertainty, and enhanced suitability for smart energy systems. • LSTM accurately predicts CCHP loads across all metrics. • Correlation analysis refines input features. • Moving average filter reduces data noise. • LSTM outperforms Bagging, RF, DT, and XGB models. • Optimized model suits EMS and SCADA deployment.