大型水库重要站点的水位预测是水库防洪中的重要问题,目前主要采用水动力学方法计算。但是由于该方法对输入边界条件的准确性要求较高,而实时调度情况下又很难完全满足此条件,因此容易造成较大的水位计算误差。对此,提出了循环神经网络模型,从水库运行的历史数据中挖掘知识,学习入库流量和坝前水位到目标站点水位的映射关系,不需要使用地形资料等数据以避免系统性误差的影响,既可以降低模型对输入边界准确性的要求,又可以在一定程度上提升水位预测精度。以三峡水库长寿站水位计算为例,使用2009—2019年数据训练、验证和测试模型,试验结果显示长寿站水位在接近三峡水库土地征用线时,计算误差在±0.4 m以内,精度优于水动力学模型,可满足实时调度对水位精度的要求。

The prediction of water level for important stations of large reservoirs is an essential issue in flood control. At present, hydrodynamics method is mainly used to calculate the water level. However, due to the high accuracy requirements of input boundary conditions, which is difficult to meet under real-time dispatching, the calculation error is likely to be large. In view of this, we present a recurrent neural network model to mine the knowledge from historical data of reservoir operation and to learn the mapping relationship between inlet flow (including main stream and tributaries) and the water level from in front of the dam to the target station. This method does not use terrain data to avoid the effect of systematic error, thereby reducing the requirement for input boundary accuracy on one hand and improve prediction accuracy on the other. In the experimental part, we compute the water level of Changshou Station of Three Gorges Reservoir. The data sets (including training, validation and test sets) consist of the operation data from 2009 to 2019. When the water level is close to the land requisition line, the prediction error is within±0.4 m, which is better than the result of water dynamic model. Therefore, the proposed method can meet the accuracy requirement of the real-time scheduling of water level.

长距离调水工程闸前水位受诸多水力控制因素影响，其波动趋势具有很强的非线性和随机性特征，难以用水动力机理模型高精度模拟，成为长距离输水调度方案制定的一大障碍。提出了一种基于深度学习网络的闸前水位预测新方法，建立了一个三层的LSTM水位预测模型，并应用于南水北调中线京石段的闸前水位预测，与深度神经网络（DNN）预测结果进行了对比。结果显示LSTM预测结果具有很高的精度，纳什系数高达0.99，均方根误差最高为0.029 m，能很好地预测水位波动趋势，预测效果比DNN更好。总结在LSTM模型构建时应考虑最大迭代次数对计算效率影响以及LSTM隐藏单元数目和学习率对精度的影响。本研究可为长距离调水工程水位预判、调度预警、水资源调度决策以及闸门智能控制提供重要参考。

The water level immediately upstream the gate of long-distance water diversion project is affected by many hydraulic control factors, and its fluctuation trend has strong non-linear and stochastic characteristics, making it difficult to simulate with high accuracy by hydrodynamic mechanism models, which remains a challenge to the diversion scheme. A new method of water level immediately upstream the gate prediction is proposed in this study based on deep learning network, and a three-layer LSTM water level prediction model is established and applied to water level immediately upstream the gate prediction which is compared with the deep neural network (DNN) in Jingshi section of MRP. The results show that the proposed model predict the trend of water level fluctuation better than DNN with high accuracy of Nash coefficient up to 0.99 and root mean square error up to 0.029 m. In conclusion, the influence of iterations on the calculation efficiency and the influence of the number of LSTM hidden units and the learning rate on accuracy should be considered in construction of LSTM model. Important reference can be provided for water level prediction, scheduling warning, water resource scheduling decision and intelligent gate control in long-distance water diversion project.