Web application for simple system identification from experimental data

Abstract

This paper presents a project for simple web-based system identification from experimental data. A user can upload input/output data from a process to be identified in three common file formats, the data can be filtered simply using two approaches and a structure of the identified model is user-controlled. The resultant discrete-time ARX model is obtained in a common form together with the possibility to assess quality of identification and to obtain also a continuous-time model. The developed application is built on the interconnection of the latest web technologies with the MATLAB computing system where the obtained results can be easily exported. The paper explains motivation for development of this site and gives also detailed description of the whole process including the Web—MATLAB interconnection. The results are presented using selected screen-shots of the application and discussed further.