【摘要】 现有的基于仿射不变矩的轴心轨迹分类方法仅支持对二维轴心轨迹的形状进行分类,不能对三维轴心轨迹进行分类。本文首先引入了一种基于特征点之间距离的均值和方差来进行分类性能评价的矩阵,验证了给出的三维不变线矩对同一形状的轴心轨迹保持仿射变换的不变性,并使用Quadratic SVM实现了对三维轴心轨迹和二维轴心轨迹的分类。数值实验表明,所提出的三维不变矩和SVM结合的方法能够有效对不同形状的轴心轨迹故障进行分类。更多还原

【Abstract】 The existing axial trajectory classification method based on affine invariant moment only supports the classification of the two-dimensional axis trajectory, cannot classify the threedimensional axis trajectory. We firstly introduced a classify performance evaluation matrix based on mean and variance of the distance between feature points of different types, which was then used to verify the invariance of the given three-dimensional moment invariants while applying 3 D affine transformation to the same shape of 3 D axis trajectory. The following numerical experiments support the validation in classifying the 3 D axis trajectory by SVM based on the given 3 D invariant moments.更多还原