A Practical Estimation Method of Inland Ship Speed Under Complex and Changeful Navigation Environment

The complexity and changefulness of inland navigation environment in space time makes it hard to guarantee the applicability accuracy existing ship speed models. In this paper, a novel method for modelling under complex changeful is proposed. Firstly, an unsupervised machine learning algorithm, Density-Based Spatial Clustering Application with Noise (DBSCAN), utilized cluster environmental data including water level, speed, wind direction, get segment division information, which greatly helps reduce influence other uncertain factors on model. Then, Generalized Regression Neural Network (GRNN) tailored employed build estimation model multiple input variables. Finally, detailed case study sailing Yangtze River trunk line conducted validate proposed methods. results show that established based methods works effectively analysis. Moreover, compared regression neural networks, GRNN has best performance modelling.