防控三峡水库支流水华的生态约束型优化调度

三峡水库蓄水后支流水华频发，严重影响了库区支流的水环境，开展水库生态调度，改变干支流水体之间的水动力条件，能有效的防控水库支流水华的发生。提出了嵌套水华预测模型的水库生态约束型调度模型，以水位（H）、水位变频(HF)、水位变幅(HR)为水华预测模型的输入量，水体中的叶绿素浓度为预测输出量，采用BP神经网络（BNN）构建了水华预测模型；将水华预测模型嵌入到水库调度模型中，以水体中的叶绿素值为约束构建了防控支流水华的生态调度模型，并以离散型动态规划法（DDDP）对调度模型进行了求解。结果表明：构建模型的预测验证相关性为R=0.8536，能较好的预测水体叶绿素含量；在极端水华100%控制保证率的生态约束条件下，生态调度发电量较实际调度发电量损失了0.89%，三峡水库开展生态调度在保证整体经济效益不亏损的情况下能有效的控制支流极端水华的暴发。

Research on Optimal Operation at Ecological Constraint for Controlling Algal Blooms in the Tributaries of Three Gorges Reservoir

After the impoundment of Three Gorges Reservoir, aquatic environment of the tributaries in this reservoir has been subject to algal blooms. The ecological operation can change the hydrodynamic condition of the reservoir tributaries, and help to control algal blooms. A nested prediction model of reservoir operation with ecological constraint was proposed. The back propagation neural network algorithm(BNN) was used to construct a model for forecasting the algal biomass in the tributaries. The input variables of this prediction model are water level(H), water level frequency conversion(HF) and water level Range(HR).The output variable is chlorophyll concentration. The prediction model is embedded into the reservoir operation model and constructing the ecological operation model which considering the water blooms in the tributaries, then using the discrete dynamic programming(DDDP) to solve the ecological operation model. The results show that the correlation coefficient between the observation and simulation calculated by the prediction model is R = 0.853 6. That means this model can predict the chlorophyll concentration well. Under the condition of 100% control guarantee rate for extreme water bloom. The calculation shows that the ecological operation power generation is 0.89% less than the real power operation. Carrying out the ecological operation in the Three Gorges Reservoir can effectively control algal blooms in the tributary without losing the overall economic benefits.