因受下游岷江、横江涨水顶托的影响,向家坝水库出库流量与下游水位的关系较为复杂,对水文计算结果和水库调度决策产生了直接影响。以2014年7月~2015年9月向家坝水库下游水位、出库流量、横江水文站和高场水文站流量为研究数据,利用顶托法和BP神经网络模型对横江、岷江的顶托影响进行了分析。结果表明,BP神经网络法检验期最大误差0.32 m,平均误差0.11 m；顶托法最大误差0.74 m,平均误差0.40 m；顶托量大于1m时,BP神经网络法平均误差0.09 m,顶托法平均误差0.42 m。 BP神经网络模型计算精度较高,尤其是在顶托量较大时计算结果较好,可用于生产实际。%The outflow of Xiangjiaba Reservoir is affected by the backwatering effect of Minjiang River and Henjiang River at downstream, so the relationship between discharge and water level is complex, which impacts the hydrological computation and reservoir operation directly. On the basis of the observed downstream level, discharge of Xiangjiaba Reservoir and discharges of Hengjiang Hydrological Station on Hengjiang River and Gaochang Hydrological Station on Minjiang River from July 2014 to Sep-tember 2015, the backwatering effect was analyzed by the backwater method and BP network. The results show that, during the test period, the maximum error and average error of BP network are 0. 32m and 0. 11m;the maximum error and average error of backwater method are 0. 74m and 0. 40m;when the backwatered level is larger than 1 m, the average error of BP network and backwater method are 0. 09m and 0. 42m respectively;BP network method has a higher accuracy, especially when the backwa-tered level is large, which can be applied in practice.
展开▼
