Intelligent control
Code
10993
Academic unit
Faculdade de Ciências e Tecnologia
Department
Departamento de Engenharia Electrotécnica
Credits
6.0
Teacher in charge
Paulo José Carrilho de Sousa Gil
Weekly hours
4
Total hours
62
Teaching language
Português
Objectives
Study of inteligent control techniques, both theoreticaly and applied. Starting by introducing concepts on linear dynamic systems identification and adaptive pole placement control, followed by the use of artifitial neural networks and fuzzy logics as means to approximate nonlinear dynamics. The core of this course is the development of neural and fuzzy controllers.
Prerequisites
Control theory and Computer control systems.
Subject matter
Linear Systems Identification: Problem; The identification process; Linear time invariant models; Parameter estimation: least squares; Model validation; RLS.
Adaptive Control: Functional models; Pole placement.
Artificial Neural Networks: The neuron; Activation functions; Proactive multi-layer networks; Approximation properties; Supervised training in multi-layer networks ; Generalization and validation; Neural control architectures.
Fuzzy Control: Fuzzy systems fundamentals; Defuzzification of variables; Inference with linguistic variables; Defuzzification of linguistic variables; Fuzzy control design.
Bibliography
Main references
- Identification and Adaptive Control, Paulo Gil, 2002 (in portuguese)
- Neural Control, Hermínio Duarte-Ramos, 2002 (in portuguese)
- Fuzzy Control, Hermínio Duarte-Ramos, 2002 (in portuguese)
Additional references
- System Identification, Lennart Ljung, 1987
- System Identification and Control Design, I. Landau, 1990
- Neural Network Design, M. Hagan, 1996
- Neuro-Fuzzy and Soft Computing, Jang, Sun e Mizutani, 1995
Evaluation method
The course assessment comprises two components: Lab. work ((TG1.1 + TG1.2 + TG1.3)) + two quizzes.
The CI grade is given by:
Grade = (TG1.1 + TG1.2 + TG1.3)*0.5/3 + (MT 1 + MT 2)*0.25.