Publication: Subspace identification of deterministic system by using orthogonal decomposition method
Date
2010-04-01
Authors
Abdullah, Norulhuda
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The main objective of this research project is to perform a system identification procedure to identify linear dynamical systems by using the orthogonal decomposition
method. The idea is to project the input output data onto the space of exogenous input. The LQ decomposition is used to obtain the deterministic components. The Orthogonal method
is then applied to deterministic components in order to derive state-space models of the plant. The scope of the subspace identification method was to build single-input singleoutput
(SISO) models from observed input-output data sequences. This procedure is done by using MATLAB software. There are three numerical examples that are used to
illustrate the performance of the orthogonal decomposition method. In this project work, the numerical examples for simulation data include simple data, heating system data, and
electrical servo motor data. Those three examples are simulated for both noise free system and noise system. Next, the validation process is performed by using three methods; Mean
Square Error (MSE), Best Fit (BF), and Variance Accounted for (VAF). Based on simulation results, the orthogonal method is able to identify the system even for a plant
model who heavily corrupted by noise. Overall, the objective of this research as to investigate the reliability and capability of the orthogonal decomposition method in
identifying the deterministic system is achieved successfully.