Explainable district heating load forecasting by means of a reservoir computing deep learning architecture

The European Union (EU) stands at a critical juncture in its energy policy, particularly in the face of evolving global energy dynamics and the urgent need for climate action. This necessitates a paradigm shift towards a more efficient, interconnected, and digitally enhanced energy market, where the integration of renewable energy sources is prioritized. In this context, the role of load forecasting for district heating and cooling systems becomes increasingly significant, especially in the low temperature grids introduced with the 5th generation district heating system. This study presents a Deep Learning methodology based on Reservoir Computing, designed to forecast district heating and cooling loads measured at the generation level. To demonstrate its applicability, a Combined Heat and Power (CHP) plant is considered. The integration of eXplainable Artificial Intelligence (XAI) with black-box models in district heating forecasting not only enhances transparency but also builds trust and facilitates better decision-making by providing clear explanations of the model predictions. The performance of the models was assessed using the Root Mean Squared Percentage Error (%RMSE) and the Mean Absolute Percentage Error (%MAE). The Reservoir Computing-based solution demonstrated superior performance, achieving %RMSE values that were 6.39% and 4.90% lower than those of the second-best baseline model for the training and test sets, respectively. Additionally, the %MAE for the training data was reduced by 3.02%. Furthermore, this approach highlights the increasing significance of Explainable Artificial Intelligence (XAI) in improving the understanding and interpretability of complex forecasting models. By making these models more accessible and comprehensible, XAI plays a crucial role in energy forecasting, particularly in district heating and cooling (DHC) systems. These systems are typically managed by local authorities, whose primary objectives center on sustainable and transparent operations.