2016/09/26 by Philippe Dreesen, Dreesen, Philippe, David T. Westwick +5
Computer Science · Engineering · Mathematics · #Advanced Adaptive Filtering Techniques #Digital Filter Design and Implementation #FOS: Electrical engineering #FOS: Mathematics #Numerical Analysis (math.NA) #Real-time simulation and control systems #Systems and Control (eess.SY) #Tensor decomposition and applications #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.1609.08063
openalex publication_date 2016/09/26 · openalex created_date 2022/10/01 · openalex updated_date 2026/08/01
Providing flexibility and user-interpretability in nonlinear system\nidentification can be achieved by means of block-oriented methods. One of such\nblock-oriented system structures is the parallel Wiener-Hammerstein system,\nwhich is a sum of Wiener-Hammerstein branches, consisting of static\nnonlinearities sandwiched between linear dynamical blocks. Parallel\nWiener-Hammerstein models have more descriptive power than their single-branch\ncounterparts, but their identification is a non-trivial task that requires\ntailored system identification methods. In this work, we will tackle the\nidentification problem by performing a tensor decomposition of the Volterra\nkernels obtained from the nonlinear system. We illustrate how the parallel\nWiener-Hammerstein block-structure gives rise to a joint tensor decomposition\nof the Volterra kernels with block-circulant structured factors. The\ncombination of Volterra kernels and tensor methods is a fruitful way to tackle\nthe parallel Wiener-Hammerstein system identification task. In simulation\nexperiments, we were able to reconstruct very accurately the underlying blocks\nunder noisy conditions.\n