基于宽度学习系统的内河航道船舶轨迹分类算法

Classification algorithm of ship trajectory in inland waterways based on broad learning system

为解决内河航道中具有不同运动模式的船舶轨迹识别问题,提出一种基于宽度学习系统(broad learning system,BLS)的船舶轨迹分类算法。对通航区域进行划分并制定轨迹筛选规则以构建标签矩阵。利用分段三次Hermite插值法分别从轨迹点记录时间上等时距和轨迹点空间分布上等间距两个角度,从原轨迹数据中进行特征点坐标的提取以构建轨迹特征矩阵。将标签矩阵和轨迹特征矩阵代入BLS以实现分类算法的训练与测试。以京杭运河淮安段交叉航道AIS数据为实例,进行轨迹分类实验。结果表明,基于BLS的船舶轨迹分类算法在分类精度和训练耗时上均优于基于反向传播神经网络和支持向量机的轨迹分类算法。

In order to solve the problem of identifying ship trajectories with different motion patterns in inland waterways,a ship trajectory classification algorithm based on the broad learning system(BLS)is proposed.The navigation area is divided and the trajectory selection rules are formulated to construct the label matrix.The piecewise cubic Hermite interpolation algorithm is used to extract the characteristic point coordinates from the original trajectory data from the perspectives of the equal time interval in the recording time and the equal space interval in the spatial distribution to construct the trajectory characteristic matrices.In order to train and test the classification algorithm,the label matrix and trajectory characteristic matrices are put into the BLS.The AIS trajectory data of the Huai’an section of the Beijing-Hangzhou Canal are selected for the trajectory classification experiment.The result shows that the ship trajectory classification algorithm based on the BLS is superior to those based on the back propagation neural network and the support vector machine in classification accuracy and training time.