A novel prediction model for integrated district energy system based on secondary decomposition and artificial rabbits optimization

Abstract

Energy predictions for buildings are the basis for energy efficiency and the implementation of smart technologies to cope with operational and energy planning issues in buildings, playing a crucial role in the implementation of environmental protection measures. Despite numerous methods proposed in current research to forecast energy, dealing with seasonal and non-linear data, particularly heat loads, presents significant volatility, resulting in less precise and poorly fitted predictions. This study introduces an artificial rabbits optimization architecture based on secondary decomposition to provide a solution for the prediction of heat loads. Leveraging secondary decomposition proves effective in discerning data trends and seasonality while simplifying the original data, thereby boosting prediction accuracy. Intelligent optimization is added for neural network parameter optimization and the trained model is used to predict the individual decomposed data to improve the fitness between the data and the model. Extensive assessments show that the proposed framework excels with an R2 of 98.87% and outperforms other models, achieving the highest 6.11% accuracy boost. Accurate prediction of building heat loads is necessary for the energy transition in the construction industry, driving the development of new technologies in building technology and accelerating the transition to clean and renewable energy.