Asymmetric Volterra Models Based on Ladder-Structured Generalized Orthonormal Basis Functions

Abstract

Abstract-In this paper, an improved method to construct and estimate Volterra models using Generalized Orthonormal Basis Functions (GOBF) is presented. The proposed method extends results obtained in previous works, where an exact technique for optimizing the GOBF parameters (poles) for symmetric Volterra models of any order was presented. The proposed extensions take place in two different ways: (i) the new formulation is derived in such a way that each multidimensional kernel of the model is decomposed into a set of independent orthonormal bases (rather than a single, common basis), each of which is parameterized by an individual set of poles responsible for representing the dominant dynamic of the kernel along a particular dimension; and (ii) the new formulation is based on a ladder-structured GOBF architecture that is characterized by having only real-valued parameters to be estimated, regardless of whether the GOBF poles encoded by these parameters are real-or complex-valued. The exact gradients of an error functional with respect to the parameters to be optimized are computed analytically and provide exact search directions for an optimization process that uses only input-output data measured from the dynamic system to be modeled. Computational experiments are presented to illustrate the benefits of the proposed approach when modeling nonlinear systems.Index Terms-Generalized orthonormal basis functions (GOBF), nonlinear systems, system identification, Volterra models.

The modified extended Kalman filter based recursive estimation for Wiener nonlinear systems with process noise and measurement noise

“…From (4), t given the dataset {Y t − 1 } follows the Gaussian distribution with mean − t and unknown variance D − t . Using (10) and (11) gives…”

“…7 A Wiener system is composed of a dynamic linear block followed by a nonlinear block and has been widely applied to the description of the actual physical plants such as biological processes and chemical processes. 8,9 The reason mainly lies in two-folds: (a) the structure of Wiener systems is highly simple in contrast to other nonlinear models such as the NARMAX models and Volterra models; 10,11 and (b) the Wiener systems are judged to have the ability to represent nonlinear systems with arbitrary precision. 12,13 When the nonlinear block of the Wiener system has static property, that is, the output nonlinearity is expressed as a linear combination of known bases, the Wiener system can be parameterized into a linear regressive model and the conventional identification algorithms for the linear systems can be used.…”

A state-space description for perfect-reconstruction wavelet FIR filter banks with special orthonormal basis functions

“…The modeling of real systems is of great importance to engineers, since models are usually needed for the design of new processes and the analysis of existing ones. Generally, advanced techniques of controller design, optimization and supervision are based on process models, and the quality of the model directly influences the quality of the final solution to the problem [2][3][4].…”

“…There are models obtained by using orthogonal functions that form a complete basis for the Lebesgue L 2 [0, ∞) space and orthonormal basis functions [2,3]. Such models have some characteristics that make them attractive for dynamic systems modeling: absence of output recursion, not requiring prior knowledge of the exact structure of the vector of regression; possibility to increase the capacity of representation of the model by increasing the number of orthonormal functions employed; natural uncoupling of the outputs in multivariable models; tolerance to unmodeled dynamics, and others [3].…”