Context:

This work applies an experimental methodology to the design of a control system based on a non-conventional Mamdani fuzzy controller that regulates the speed of an encoder-based DC motor.

Method:

The proposed methodology consists of four steps: i) fuzzy controller input/output selection, ii) fuzzy controller design, iii) controller hardware implementation, and iv) membership function parameterization. This methodology generates seven pairs of unique error and control signals that are differentiated by experimentally adjusting the parameters of the triangular membership functions used for the fuzzy controller design, which was implemented in an Atmega328P micro-controller. For each of the seven approaches defined, an experiment was performed, performing a control action to obtain the transient response of the DC motor speed when the reference was a step-type signal.

Results:

The motor response and the reference signal were used to calculate the error, whose squared error integral was estimated to determine which experimental approach yielded the best fuzzy control results, i.e., with the lowest possible error.

Conclusions:

The proposed methodology ensures the minimization of the squared error integral between the signal to be controlled and the reference signal. For fitting 6, the performance index obtained was J = 0.0002, which represents a decrease of ≈ 99.99% with respect to the worst case (fitting 1), whose performance index was J = 4.10.

        · resumo em Espanhol     · texto em Inglês     · Inglês ( pdf )