With the continuous expansion of water resource allocation projects, accurate prediction of electricity consumption is crucial for energy conservation, cost control, and construction efficiency. Traditional power consumption prediction methods, such as LSTM and Transformer, are difficult to capture both short-term and long-term dependencies when processing complex time-series data. To address this challenge, this paper proposes using xLSTM (Extended Long Short Term Memory Network) to predict power consumption in multiple regions. XLSTM combines the short-term dependency modeling advantages of sLSTM with the long-term dependency modeling capabilities of mLSTM, and can effectively process power consumption data between multiple regions, considering the temporal correlation between different regions. The experimental results show that xLSTM performs well in multi regional power consumption prediction, with a mean square error (MSE) of 0.0030 and an average absolute error (MAE) of 0.035, which is superior to other models. This model provides effective technical support for precise prediction of electricity consumption, and can provide strong guarantees for accurate decision-making and intelligent scheduling management in large-scale water resource allocation projects.