À¶Ý®ÊÓƵ

The increasing need for manufacturing speed and accuracy leads to the desire of more accurate 
dynamic models of machine tools. Usually, parameter identification of multi-body systems is 
limited to the least-squares solution of over-determined linear equations. To enhance the 
modeling and identification accuracy, a new identification method is proposed by using a non-
linear optimization method with a normalized weight objective function to estimate the rigid-
body parameters of a MIMO system, which is coupled and nonlinear by nature. This method 
normalizes the prediction sensitivity of each axis. The proposed method is applied to a three-axis 
feed drive system, which contains one linear drive and two rotary drives, stacked in a series 
configuration. The results show that the total relative prediction error decreases significantly 
compared to the conventional least square methods. Moreover, the identified dynamic model can 
be used as a virtual machine to predict the tracking error more accurately compared to using 
separate SISO models.