长江航道主缓流判别算法研究

船舶航行时充分地利用主缓流航道可以节约能源。为了更加准确地划分长江航道的主缓流航道,提出了利用 k-近邻算法和 P 分位数法的划分方法。改进后的 P 分位数法采用动态选择 P 值的方法避免了主流区域可能过小的问题,k-近邻算法利用了测点的水深、流速等多种航道水文信息来划分主缓流。利用长江航道金口处和岳阳处的11个截面的实测数据对 k-近邻算法和 P 分位数法进行了对比分析,其中有10个截面的主缓流区域相近。进一步对比长江航道中典型的龙口水道和杨林岩水道,发现根据算法得出的主流区宽度和实际的相比,误差在12%以内,可以满足船舶航行的需要。 

An Algorithm for Identifying Main and Slow Flow Channels of Yangtze River

Vessels which navigate on the main flow channel can save energy.In order to more accurately divide the main and slow channels of the Yangtze River,this paper proposed methods of k-nearest neighbor algorithm and P-tile method.The improved P-tile method can avoid the main stream being too small by dynamic selection of P values.The k-nearest neighbor algorithm partitions the main and slow streams with channel hydrology information,such as depths and flow velocities.With the 1 1 sections'data of Yueyang and Jinkou measuring stations in Yangtze River,this study com-pared the partition results of these two algorithms.The results showed that areas of main and slow channels at the 10 out of 1 1 sections were close.A further comparison with the waterways at Longkou and Yanglinyan of the Yangtze River indi-cated the difference of the mainstream zone width estimated from the proposed algorithms was within 12% comparing to the actual,which can meet the requirements of the ship navigation.