Graduate study programme

Back   Schedule   Hrvatski

Data based modeling DRb3-03

ECTS 7 | P 30 | A 0 | L 30 | K 0 | ISVU 149759 | Academic year: 2019./2020.

Course groups

Prikaži sve grupe na predmetu

Course lecturers



Introduce students to the basics of methodology for extracting knowledge about a process from the available measured data, and teach them how to build a process model with required properties based on these pieces of information. Present relevant skills required for handling available software tools for analysis and processing of measured data, as well as software tools for building process models based on these data. Acquaint students with the way of introducing intelligence into automatic control systems.

Conditions for enrollment

Requirements met for enrolling in the second year of the study programme

Course description

Modelling of processes and other functional relationships in data, based on measured data. Measured data obtained by a separate experiment and plant (operating) data. Measured data informativness. Sample time selection. Preprocessing of measured data and forming data sets for process model building. Building a static and dynamic model. Selection of input and output variables and model structure selection. Methods for model parameter estimation. Regression modelling. Non-recursive and recursive methods for model parameter estimation. Methods based on projection of input space into a latent subspace. Evaluation of the built process model. Application of artificial neural networks in data based modelling. Application of the Matlab software package based on data modelling. Virtual (soft) sensor and difficult-to-measure process variable estimation. Program implementation of built mathematical models into the industrial information system.

Student requirements

Defined by the Student evaluation criteria of the Faculty of Electrical Engineering, Computer Science and Information Technology Osijek and paragraph 1.9

Monitoring of students

Defined by the Student evaluation criteria of the Faculty of Electrical Engineering, Computer Science and Information Technology Osijek and paragraph 1.9

Obligatory literature

1. 1 Perić, N., I. Petrović Identifikacija procesa FER, Zagreb, 2000.,

2. 2 Fortuna, L., S. Graziani, A. Rizzo, M.G. Xibilia Soft sensors for Monitoring and Control of Industrial Processes Springer-Verlag London Limited 2007.

Pretraži literaturu na:

Recommended additional literature

1. 1 Ljung, L. System Identification - Theory for the User Prentice-Hall, Eaglewood Cliffs, 1987.,

2. 2 Haykin, S. Neural Networks – A Comprehensive Foundation, 2nd edition Prentice Hall, 1999.,

3. 3 Martens, H., T. Naes, Multivariate Calibration, 2nd edition John Wiley & Sons, New York, 1991.

Course assessment

Conducting university questionnaires on teachers (student-teacher relationship, transparency of assessment criteria, motivation for teaching, teaching clarity, etc.). Conducting Faculty surveys on courses (upon passing the exam, student self-assessment of the adopted learning outcomes and student workload in relation to the number of ECTS credits allocated to activities and courses as a whole).

Overview of course assesment

Learning outcomes
Upon successful completion of the course, students will be able to:

1. carry out the collection, analysis and preprocessing of measured data

2. highlight the advantages and disadvantages of a given process identification method

3. develop a dynamic mathematical model for a given problem by selecting an appropriate process identification method and implement it in Matlab

4. explain problems in process monitoring and in realization of the control system with the existence of difficult-to-measure process variables, and problem solving using estimators

5. evaluate the suitability of a particular modelling method based on the projection of the input data space into the latent space for a given problem

6. build a process model based on data by using analyzed methods and the Matlab program package

Aktivnosti studenta: Vidi tablicu aktivnosti