Nonlinearity tests (Q2753633)

The paper is devoted to the system identification problem and particularly to the nonlinearity assessment algorithms. Two tests are under consideration. Some interesting results are obtained. One of the tests considered is based on the Gilbert transform. It is applied and investigated in the frame of a numerical example. After some investigations it is shown that a type of nonlinearity of a complex system can be successfully identified. In comparison, the nonlinear prediction error test is considered and evaluated. It is shown that input-nonlinearities cannot be detected in such a way. The acquired results could be of interest for all specialists working in the area of system identification.