Objectives and competences
The objective of this course is that student will understand principles, design and use controllers based on AI techniques.
Content (Syllabus outline)
• Introduction to the theory of artificial techniques (AI): definition, usage in control.
• Deep neural network based control: artificial neuron, activation function, construction of multi-layer neural networks, different types of neural networks, BPG learning method, examples of robot control application.
• Control with genetic algorithms: biological basis of genetics algorithms, offspring, mutation, evolution, selection, Darwin algorithm, effect of cross-bread, effect of mutation, intersection of both effects (mutation and cross-bread), algorithm of genetic calculus, an application of optimisation problem in control of mechatronics system.
• Fuzzy logic control: fuzzy logic, fuzzification, defuzzification, fuzzy membership function, decision table, rule base. Application of fuzzy logic controller for mechatronics systems.
• Particle swarm optimisation (PSO) control: presentation of the PSO algorithm, comparison between PSO and genetic algorithms, aplication of PSO algorithm for syntesis of PID type of contoller for simple robot system, aplication of PSO algorithm for indentification of asynchronous AC motor parameters.
• Combining different AI techniques with each other and AI techniques and classical techniques with each other.
Learning and teaching methods
• lectures,
• tutorial,
• lab work.
Intended learning outcomes - knowledge and understanding
• demonstrate knowledge and understanding from the topics of a design of controllers based on AI techniques
• design, implement and use controller based on AI techniques
Intended learning outcomes - transferable/key skills and other attributes
Communication skills: oral lab work defence, manner of expression at written examination.
Use of information technology: use of software tools for design and programming of soft-computing techniques.
Readings
• R. Šafarič, A. Rojko; Inteligentne regulacijske tehnike v mehatroniki, Univerza v Mariboru,FERI, Maribor 2007
• A. Dobnikar: Nevronske mreže, teorija in aplikacije, Didakta, Radovljica 1990. S. Y. Kung: Digital neural networks, PTR Prentice Hall, Engelwood Cliffs, New Yersey,1993.
• Miran Brezočnik: Uporaba genetskega programiranja v inteligentnih proizvodnih sistemih, FS, Maribor, 2000.
• Elektronska skripta: Riko Šafarič, Andreja Rojko: Inteligentne regulacijske tehnike v mehatroniki, FERI, Maribor, 2005 (spletni naslov https://estudij.um.si/pluginfile.php/437200/mod_resource/content/1/Inteligentne%20regulacijske%20tehnike%20v%20mehatroniki%20V2.pdf).
Additional information on implementation and assessment Both midterm exams may be replaced by a written exam.
