中文摘要：

针对多支流河流汇流计算问题,提出了一种灰色-人工神经网络组合汇流计算方法（简称 GACC）,即先运用灰色系统理论中灰色关联分析法识别出各支流的汇流时间,然后根据汇流时间组合人工神经网络样本,建立BP网络,用BP网络进行训练,最后用得到的BP网络模型进行汇流计算.以塔里木河3源流汇流计算为应用实例,结果表明,这种汇流计算方法合理、有效,值得推广.

英文摘要：

Located in south Xinjiang, the Tarim River is the largest continental river in China, and its catchment area is 1.02×10^6 km^2. There is no runoff formation along the mainstream of the Tarim River, and the mainstream is completely recharged by its tributaries, especially its three main head streams, i. e. , the Aksu River, Yarkant River and Hotan River. Most of the residual stream water of the Yarkant River is diverted into the Xiaohaizi Reservoir after it is diverted into the irrigated areas, and the Yarkant River recharges the mainstream of the Tarim River only in flood season sometimes. Stream flow of the Hotan River, recharging the mainstream of the Tarim River, is sharply reduced after it is diverted into the irrigated areas and lost along its watercourse passing through the Hexi Corridor and the Taklimakan Desert. The recharge of the Aksu River to the Tarim River is the most in the Tarim River Basin. In view of convergence computation of multi-tributary rivers, in this paper, a Grey-ANN convergence computation method is put forward. Firstly, the grey correlation analysis in the grey systematic theory is used to calculate the converging time of each tributary. Secondly, the artificial neural networks and the BP networks are developed based on the convergence time. Finally, the convergence computation is carried out using the BP network model. A case study on the convergence computation of the three head streams of the Tarim River is carried out. The results show that the computation method is reasonable, effective and worth to be extensively applied.