| Summary: | The techniques of big data analysis hold immense potential in optimizing indoor energy consumption and enhancing comfort levels. This paper proposes a predictive method for effectively forecasting energy usage in libraries through a multi-step ahead time series-based long short-term memory-backpropagation model, integrated with building energy consumption sub-metering analysis technology. Experimental results indicate that the proposed multi-input multi-output model significantly outperforms traditional recursive and direct models in terms of predictive performance, adeptly capturing the intricate characteristics and temporal dependencies of energy consumption data, thereby offering a novel technological pathway and practical implications for building energy management. © 2025 The Author(s). Published by Oxford University Press.
|
|---|