Multi-Scale Convolutional LSTM Network for Sewage Flow Prediction

Predicting sewage flow is of great significance in urban sewage treatment and management. However, traditional prediction methods often overlook information at different time scales, failing to capture the complex relationships among these pieces of information. To address this issue, this paper proposes a multi-scale convolutional long short-term memory (LSTM) network. The convolutional layers are used to capture data patterns at different time scales, while the LSTM network is utilized to better capture the long-term dependencies and short-term fluctuations between sequences. Experimental results demonstrate that compared to traditional LSTM networks, the proposed method achieves higher accuracy and stability in sewage flow prediction.