高端数控装备多源信息融合状态识别模型

Condition recognition model based on multi-source information fusion for high-end CNC equipment

为了对高档数控装备的运行状态进行实时监测和有效感知,实现对状态的有效识别和判断,提出了一种基于运行状态多源多域空间信息融合的状态识别模型,采用增殖流形的相似度进行状态识别。对高端装备的电流信号和振动信号进行信息融合,对融合信号的时域、频域、时频域进行特征获取,重构初始特征的多域高维相空间,采用局部线性嵌入结构进行降维,优化本征维数,采用距离判据获得低维敏感特征,构建低维流形特征的增殖相似度参数实现对设备不同状态的识别。最后将该模型用于主轴试验台和某加工中心进行了试验验证,快速有效方便地识别出主轴的不同状态,验证了模型的有效性。

In order to perform running condition real time monitoring and effective perception of high-end CNC equipment and realize effective identification and judgement of the working condition,a condition recognition model based on running condition multi-source multi-domain space information fusion is proposed. The similarity of the proliferation manifold is adopted to carry out condition recognition.Firstly,the vibration and current signals of the high-end equipment are fused. Then,the features of the fused signal are acquired in time domain,frequency domain and time-frequency domain; the multi-domain high-dimensional phase space of the initial characteristics is reconstructed. The local linear embedded structure is adopted to perform dimension reduction,the intrinsic dimensionality is optimized;and the distance criterion is adopted to obtain the low dimension sensitive features. The proliferation similarity parameters of lowdimensional manifold characteristics are constructed and used to recognize different working conditions of the equipment. Finally,the model was applied in the spindle test platform as well as a certain vertical machining center to conduct test verification,and different spindle conditions were recognized fast,effectively and conveniently. The results verify the effectiveness of the model.