数控系统连接相关故障的核PCA监测方法

Monitoring Method on Connection Dependent Faults in Numerical Control System Using Kernel PCA

针对半闭环数控系统中可能出现的联轴器及连接螺栓松脱、连接杆件裂损等一类故障,利用系统中电动机驱动装置提供的冗余力矩监视信号和编码器反馈信号,通过将系统的动力学模型作为非线性核函数,提出一种基于核的主元分析故障监测方法。该方法解决了数控系统中由于位置和转矩间的非线性关系而带来的分析困难,进而使得故障发生时的特征能够得到准确及时提取。通过对X-Y数控平台中联轴器松脱故障监测实验,验证了该方法对于上述一类故障的有效性和实用性。

Focusing on a class of faults including the release of conjunctions or bolts and rupture of poles or manipulators in semi close-loop numerical control system, a kernel principle component analysis method was developed to monitor them. It used the signals drivers of motors and treated the dynamic of redundant torque monitoring and feedback encoders in the model of the system as the nonlinear kernel function. The method solved the non-linear problem between the position and torque in NC system, which made the process of extracting faults characters exact and timely when failure occurred. An experiment was employed to monitoring the fault of conjunction release on an X-Y NC platform. The results show that this method is validity and practicability for the above class of faults.