Investigation of effect on input order selection in system identification

Nurfa, Fahar (2016) Investigation of effect on input order selection in system identification. Project Report. UTeM, Melaka,Malaysia. (Submitted)

Text (24 Pages) Investigation of effect on input order selection in system identification.pdf - Submitted Version Download (2MB)

Abstract

System identification techniques are discussed and applied in investigation on effect of input order selection.The performance of each identified input order was investigated to determine its effect and to identify the best model structure.The design of the experiments and the data processing will be discussed using MATLAB's system identification graphical user interface (GUI) named 'IDENT' found in System Identification Toolbox.Basically,system identification is the technique of developing models of dynamic systems from experimental input and output data.To perform system identification,one or more test input data sequences and the measured output data sequences are required for the system being modelled.Input order selection is an essential pre-processing stage in problems such as machine learning.Model to be used is autoregressive with exogenous input (ARX) model.In this study,3 different equations were used throughout the investigation.Every equation was evaluated based on several indicators.The procedure of discovering a decent model can be isolated into four assignments which are analysis configuration, model type selection,parameter estimation and model validation.In conclusion,the effectiveness of all indicators help user to do input order selection in system identification.The indicators that were used throughout this investigation allows the user to identify the most accurate model and select a complexity which leads to a good reproduction of input properties of true system.

Actions (login required)

View Item