Hybrid forecasting model for central air conditioning load based on CEEMDAN and WTCN-GRU

Accurate load prediction is an important foundation for the energy-saving and optimized control of central air conditioning systems, and it is crucial for energy conservation and emissions reduction in buildings. To address the issues of low accuracy in existing load forecasting models, this paper proposes a load prediction method of air conditioning based on hybrid model combining Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) and an improved Temporal Convolutional Network (WTCN) -Gated Recurrent Unit (GRU). Firstly, Pearson correlation analysis is used to select highly correlated influencing factors as feature inputs. CEEMDAN is then applied to decompose and reconstruct the original data to mitigate data non-stationarity and improve data quality. Secondly, the first-layer convolution of each residual block in the TCN is improved to enhance feature extraction capability. Thirdly, the gating mechanism in the GRU is utilized to handle the temporal relationships in the data and predict the air conditioning load. Finally, experiments are conducted using the central air conditioning load data from office building for validation. The results show that the proposed model outperforms other benchmark models, significantly improving the accuracy of building air conditioning load forecasting. It holds promising application prospects in optimizing building energy consumption control.