Han H.,Staff Office 101 of the Second Artillery Engineering University |
Ma H.,Staff Office 101 of the Second Artillery Engineering University |
Cao J.,Xi'an Jiaotong University |
Zhang J.,Xi'an Jiaotong University
Hsi-An Chiao Tung Ta Hsueh/Journal of Xi'an Jiaotong University | Year: 2012
A novel non-parametric identification method is proposed to solve the hard problem in identifying Volterra frequency domain kernels of multi-input multi-output (MIMO) nonlinear systems. Firstly, the single tone stimuluses with different frequency components are added to the input ports of the system under test. Secondly, the output signals of the system are sampled and every order kernel output is separated by using the Vandermonder method. Thirdly, a formula is derived in terms of output properties of Volterra kernels of MIMO nonlinear system stimulated by single tone. Since the multitone signal is applied to stimulate MIMO nonlinear system, multi-point of Volterra frequency domain kernels can be simultaneously identified in a single-round test, and hence, the efficiency of the proposed method is improved. The proposed method is more practical because it only relies on input-output data of systems. The proposed method is applied to identify the Volterra frequency domain kernels of a two-input two-output nonlinear system. The results show that the maximum deviation in identification is less than 10-5, and hence, the proposed method is effective.