研究半监督学习的模糊核聚类方法用于变速箱早期故障诊断的方法。故障特征不明显、样本差异小是机械故障早期检测的

难点， 基于半监督学习的核聚类方法利用少量已知模式的样本， 结合大量未知模式的样本进行半监督学习， 得到较好的识别效果。-In this paper, motion control method of semi-closed CNC method is presented for gear-box fault early detection. The difficulty

machine tools is introduced. First, regard backlash compensation in mechanical fault early detection is to detect the weakly fault

pitch error compensation and the ideal speed from interpolation information immerged in noises. The semi-supervised kernel

calculating as the velocity feed-forward.