物件導向軟體之架構(Object-Oriented Software Construction)探討

System Identification I

Projection-based model reduction: Formulations for physics-based machine learning

In this communication, necessary and sufficient conditions are presented for the unique blind identification of possibly nonminimum phase channels driven by cyclostationary processes. Using a frequencydomain formulation, it is first shown that a channel can be identified by the second-order statistics of the observation if and only if the channel transfer function does not have special uniformly spaced zeros. This condition leads to several necessary and sufficient conditions on the observation spectra and the channel impulse response. Based on the frequency-domain formulation, a new identification algorithm is proposed

Blind channel identification based on second-order statistics: a frequency-domain approach

This chapter presents a summary of the issues discussed during the one day workshop on “Support Vector Machines (SVM) Theory and Applications” organized as part of the Advanced Course on Artificial Intelligence (ACAI ’99) in Chania, Greece [19]. The goal of the chapter is twofold: to present an overview of the background theory and current understanding of SVM, and to discuss the papers presented as well as the issues that arose during the workshop.

Support vector machines: Theory and applications

A new approach for simultaneous online identification of unknown time delay and dynamic parameters of discrete-time delay systems is proposed in this paper. The proposed algorithm involves constructing a new generalized regression vector and defining the time delay and the rational dynamic parameters in the same vector. The gradient algorithm is used to deal with the identification problem. The effectiveness of this method is illustrated through simulation.

New results on discrete-time delay systems identification

This article presents a new approach of systematic determination of models base for the multimodel approach. The application of this approach requires, first, to classify a numeric data by exploiting the self-adapting artificial Kohonen neural-networks. The obtained data relative to the clusters are then exploited for both structural and parametric estimation of base models. To resolve the estimation problem of the validity of the elementary models, used we proposed a new technique, based on the minimization of a quadratic criterion. This criterion exploits the centers of clusters obtained in the determination of the models base step. A comparative study with the residues approach showed the contribution in precision of the proposed validities computation technique. The satisfactory results obtained in numerical simulation, incited us to validate experimentally the contributions already mentioned. Indeed, an on-line experimental validation of the new proposed multimodel representation was carried out on an olive oil esterification reactor. The obtained results are very satisfactory in terms of precision and robustness.

Multimodel Approach using Neural Networks for Complex Systems Modeling and Identification

In this paper, we apply a hierarchical identification principle to identify simultaneously the unknown time delay and dynamic parameters of discrete-time delay systems. In our approach, we separate the nonlinear cost function into two simple cost functions and present a gradient iterative algorithm for estimating directly the time delay and the parameters. The cost function used to estimate the time delay is an integer nonlinear programming problem (IP) and the solution of the integer version of a nonlinear optimization problem is obtained by rounding the solution for its continuous relaxation (CP). Furthermore, we give an appropriate choice of the convergence factor. Finally, the effectiveness of this method has been illustrated through simulation.

Hierarchical gradient based identification of discrete-time delay systems

This paper deals with the problem of identification of Wiener systems with unknown integer time delay. This problem presents several difficulties, since it consists in estimating simultaneously the integer time delay, the parameters of the linear dynamic subsystem and the parameters of the non-linear static block. To overcome these difficulties, we propose a new approach to identify the Wiener time delay systems based on the Hierarchical gradient identification principle. This proposed method consists, first, in decoupling the original optimization problem into three simple separate criteria jointly optimized and second, in using the gradient algorithm to minimize each function. The analysis gives us insight into the properties of this algorithm and develops condition that must be satisfied to ensure the convergence and to guarantee the use of rounding property. Simulation results are presented to illustrate the performance of the proposed method.

Kamel Abderrahim

Papers

New results on discrete-time delay systems identification

Multimodel Approach using Neural Networks for Complex Systems Modeling and Identification

Hierarchical gradient based identification of discrete-time delay systems

Multistage for identification of Wiener time delay systems based on hierarchical gradient approach

Adaptive control design using stability analysis and tracking errors dynamics for nonlinear square MIMO systems