A Load Monitoring Network Integrating CBAM Attention Mechanism with BiTCN-GRU

Load monitoring provides more granular electricity usage data and feedback, aiding in the optimization of household energy management. However, existing deep learning methods struggle to effectively capture both global dependency information and local feature details of appliances simultaneously. To address this issue, we propose a network model that integrates a Convolutional Block Attention Module (CBAM) with a Bidirectional Temporal Convolutional Network and Gated Recurrent Unit (BiTCN-GRU) to predict load monitoring in this study. The model integrates the CBAM attention mechanism into the residual block structure of BiTCN, forming an enhanced TCN network that is then combined with GRU to enhance the extraction of both long-term and short-term features from sequence data. Extensive experiments conducted on the REDD and UK_DALE datasets demonstrate that the proposed model reduces the Mean Absolute Error (MAE) index by 30.74% compared to other methods, verifying the effectiveness of the model. Results indicate that the proposed network offers significant advantages in terms of monitoring accuracy.