| Abstract |
This paper proposes apply fuzzy learning techniques for modeling electromechanical systems, incorporating learning capabilities into the control system to adjust itself to the different operating conditions. Compensation of non - linear terms affecting system dynamics and implementation of "intelligent" systems with self-learning capabilities are the main applications for this new association. Three fuzzy algorithms are presented, learning from examples the rules expressing system behavior regardless of having analytical dependency or not. To investigate the association between electromechanical systems and fuzzy modeling, an experimental system composed by an electro-hydraulic position system is used. The performance of the algorithms will be compared and studied for different cases like systems modeling in a direct and indirect way, the number of need examples to build a reasonable model, the different learning capabilities of each algorithm and their use of the accessible information. |
