Implementação em FPGA de controladores nebulosos sintonizados por algoritmos genéticos

Abstract

This master thesis proposes a systematic design for fuzzy controllers in FPGA hardware, and proposes a methodology to tune these controllers using genetic algorithms. The steps of simulation and experimental results of the implementation were obtained over the control of a nonlinear plant: the damped pendulum. The system consists of a FPGA project development board and interface cards for drives and data acquisition. As main features of this project can be cited: the digital control algorithm implemented by using a hardware description language, the parallel processing architecture to provide improved system performance, and implementing a methodology based on genetic algorithms as search tool to tune the fuzzy controller with function based performance deviations on the desired response. From a process model simulations were carried out and tuning fuzzy controllers through the application of genetic algorithms. With the results obtained in simulation is performed the implementation of the fuzzy controller in VHDL associated with an interface for data acquisition trials. Experimental results validate the theoretical analysis and confirm system performance, and offer satisfactory results for several operating points for a linear model obtained from a nonlinear system.