Iterative selection of GOB poles in the context of system modeling
This paper is concerned with the problem of system identification using expansions on generalized orthonormal bases (GOB). Three algorithms are proposed to optimize the poles of such a basis. The first two algorithms determine a GOB with optimal real poles while the third one determines a GOB with optimal real and complex poles. These algorithms are based on the estimation of the dominant mode associated with a residual signal obtained by iteratively filtering the output of the process to be modelled. These algorithms are iterative and based on the quadratic error between the linear process output and the GOB based model output. They present the advantage to be very simple to implement. No numerical optimization technique is needed, and in consequence there is no problem of local minima as is the case for other algorithms in the literature. The convergence of the proposed algorithms is proved by demonstrating that the modeling quadratic error between the process output and the GOB based model is decreasing at each iteration of the algorithm. The performance of the proposed pole selection algorithms are based on the quadratic error criteria and illustrated by means of simulation results.
KeywordsGeneralized orthonormal bases (GOB) Laguerre functions Kautz functions pole estimation modelling identification
Unable to display preview. Download preview PDF.
- N. Saidi, H. Messaoud. Supervised equalization of a linear communication channel using generalized orthogonal basis (GOB), In Proceedings of the 6th International Multi-Conference on Systems, Signals and Devices, Jerba, Tunisia, 2009.Google Scholar
- A. C. den Brinker, H. J. Belt. Model reduction by orthogonalized exponential sequences. In Proceedings of PRORIS/IEEE Workshop on Circuits, System and Signal Processing, Mierlo, Netherlands, pp. 77–82, 1996.Google Scholar
- R. Malti, D. Maquin, J. Ragot. Optimality conditions for the truncated network of the generalized discrete orthonormal basis having real poles, In Proceedings of IEEE Conference on Decision and Control, Tampa, Florida, USA, pp. 2189–2194, 1998.Google Scholar
- A. Kibangou, G. Favier, M. M. Hassani. A growing approach for selecting generalized orthonormal basis functions in the context of system modeling, In Proceedings of IEEEEURASIP Workshop on Nonlinear Signal and Image Processing, IEEE, Grado, Italy, 2003.Google Scholar
- A. Mbarek, K. Bouzrara, H. Messaoud, Optimal expansion of linear system using generalized orthogonal basis. In Proceedings of International Conference on Electrical Engineering and Software Applications, Hammamet, Tunisia, 2013Google Scholar
- A. Kibangou, G. Favier, M. M. Hassani. Iterative optimization method of GOB-volterra filters. In Proceedings of the 16th IFAC World Congress, Prague, Czech Republic, vol. 38, no. 1, pp. 773–778, 2005.Google Scholar
- A. da Rosa et al. Robust expansion of uncertain Volterra kernels into orthonormal series, In Proceedings of the American Control Conference, Marriott Waterfront, Baltimore, MD, USA, pp. 5465–5470, 2010.Google Scholar
- M. Guglielmi. Signaux Aleatoires, Modelisation, Estimation, Detection. Paris, France: Editions Hermes Science, 2004.Google Scholar