Multivariate polynomials are often used to model nonlinear behavior, e.g., in parallel Wiener models. These multivariate polynomials are mostly hard to interpret due to the presence of cross terms. These polynomials also have a high amount of coefficients, and the calculation of an inverse of a multivariate polynomial with cross terms is cumbersome. This paper proposes a method to eliminate the cross terms of a multivariate polynomial using a linear input transformation. It is shown how every homogeneous polynomial described using tensors can be transformed to a canonical form using multilinear algebraic decomposition methods. Such tensor decomposition methods have already been used in nonlinear system modeling to reduce the complexity of Volterra models. Since every polynomial can be written as a sum of homogeneous polynomials, this method results in a decoupled description of any multivariate polynomial, allowing a model description that is easier to interpret, easier to use in a design, and easier to invert. This paper first describes a method to represent and decouple multivariate polynomials using tensor representation and tensor decomposition techniques. This method is applied to a parallel Wiener model structure, where a multiple-input-single-output polynomial is used to describe the static nonlinearity of the system. A numerical example shows the performance of the proposed method.
[1]
Gene H. Golub,et al.
Matrix computations (3rd ed.)
,
1996
.
[2]
Pierre Comon,et al.
Symmetric tensor decomposition
,
2009,
2009 17th European Signal Processing Conference.
[3]
E. Bai,et al.
Block Oriented Nonlinear System Identification
,
2010
.
[4]
Richard A. Harshman,et al.
Foundations of the PARAFAC procedure: Models and conditions for an "explanatory" multi-model factor analysis
,
1970
.
[5]
Gérard Favier,et al.
Parametric complexity reduction of Volterra models using tensor decompositions
,
2009,
2009 17th European Signal Processing Conference.
[6]
Leon O. Chua,et al.
Fading memory and the problem of approximating nonlinear operators with volterra series
,
1985
.
[7]
Robert D. Nowak,et al.
Tensor product basis approximations for Volterra filters
,
1996,
IEEE Trans. Signal Process..
[8]
J. Schoukens,et al.
Parametric and nonparametric identification of linear systems in the presence of nonlinear distortions-a frequency domain approach
,
1998,
IEEE Trans. Autom. Control..
[9]
Rasmus Bro,et al.
The N-way Toolbox for MATLAB
,
2000
.
[10]
M. Schetzen.
The Volterra and Wiener Theories of Nonlinear Systems
,
1980
.