Efficient Kernel Computation for Volterra Filter Structure Evaluation

Abstract

Despite their generality, conventional Volterra filters are inadequate for some applications, due to the huge number of parameters that may be needed for accurate modelling. When a state-space model of the target system is known, this can be assessed by computing its kernels, which also provides valuable information for choosing an adequate alternate Volterra filter structure, if necessary, and is useful for validating parameter estimation procedures. In this paper, we derive expressions for the kernels by using the Carleman bilinearization method, for which an efficient algorithm is given. Simulation results are presented, which confirm the usefulness of the proposed approach.

Integration of Volterra model with artificial neural networks for rainfall-runoff simulation in forested catchment of northern Iran

“…The y 1k and y 2k state the Volterra model output 151 obtained at the linear and nonlinear parts of the Volterra model, respectively, and the ɛ k represent 152 the random noise (with zero mean) of the model output. 153 The major drawback of the Volterra model is the exponential growth of the model parameters 154 with the memory and order of the model, which causes problems on identification of the model 155 particularly when the system is strongly nonlinear (Goulart and Burt, 2012 Then y 2k can be stated as:…”

Recursive Even Mirror Fourier Nonlinear Filters and Simplified Structures

“…Using Assumption 2, it can be now proved that, with basis functions as those given in Table I, we can arbitrarily well approximate the system in (1). In fact, for any , according to the Stone-Weierstrass theorem, there is a linear combination of basis functions, shortly noted as , such that for any in it is (7) Let us now consider the system (8) According to (7) it results (9) with . Therefore, according to the Assumption 2 and the Stone-Weierstrass theorem, for any there is a sufficiently small and a linear combination of basis functions as those in Table I such that the error between the output of (1) and (8) is…”

A novel nonlinear equalizer for extending the dynamic range of analog-to-digital converters

“…Volterra filters are a general technique for representing and compensating arbitrary nonlinearities with memory [3]. Nevertheless, a major drawback of this technique is that the number of coefficients in the Volterra filter grows exponentially with the order and the memory of the ADC nonlinearity behavior model [4]. To overcome the highly computational complexity, the memory polynomial (MP) [5] and horizontal coordinate system (HCS) [6,7,8] architectures for truncated Volterra filters have been proposed.…”