Analysis on evolution of Hengsha Passage in the Yangtze River estuary with BP artificial neural network

In this paper, based on the analysis of hydrology and sediment data, a BP artificial neural network model which has been applied maturely and widely is established to study the relationship between the middle-section width of the Hengsha Passage and the three other factors - the runoff, the sediment discharge of the South Branch and the split ebb flow ratio of the North Channel. With a structure on 3-1-7-1 and appropriate parameters, the BP artificial neural network is well trained and tested. The model can perform well to predict the evolution of the Hengsha Passage. The proper regulation in the Hengsha Passage, which may benefit to the operation of the deep water channel in North Passage, is suggested. © 2011 IEEE.