水库来水量长期综合预报方法研究

水库生产计划的制订、调度、汛末蓄水等活动需要了解未来的水情,因此对水库来水量做出长期预报非常重要。本文利用神经网络灵活多变的拓扑结构,建立了基于物理成因的水库来水量定性预报和定量预报模型,最后综合二者结果得出预报结论。将模型应用于我国东北地区的丰满、云峰水库年来水量的长期预测,取得了较好的效果。实践表明,定性预报和定量预报相结合的综合预报方法,克服了传统使用单一定性或定量预报模型进行预报的盲目性,可以提高预报结果的精度和稳定性,值得应用和推广。

Research on long-term synthesis prediction method of inflow to reservoir

Long-term prediction of inflow to reservoir is important for decision making of production planning,reservoir operation and storage at the end of flood period.A long-term qualitative and quantitative synthesis prediction model of inflow to reservoir based on physical factors is built by aid of the variable topological structure of neural network.The model is demonstrated by application for Fengman and Yunfeng reservoirs in the northeast of China and the results indicate that the synthesis model can be an effective way to improve the forecasting accuracy and stability.The model is worthy for popularization and application on reservoir operation and storage of water at the end of flood period.