Corresponding author: Anisimov Dmitrii N., Ph. D., Associate Professor, National Research University "Moscow Power Engineering Institute", Moscow, 111250, Russian Federation, e-mail: AnisimovDN@mpei.ru

Received on October 03, 2016
Accepted on November 11, 2016

The paper is devoted to the analysis of the influence of various factors on the dynamic characteristics of the fuzzy logic controller (FLC). In particular, the following factors are considered: the choice of the input and output variables; the number of terms, the turndown and the form of the membership functions of the fuzzy variables, the character of the relation between the space of the antecedents and the space of the consequents (the rule base), the mode of defuzzification and the measures of the sub-conditions significance. A research was carried out in two directions. The first direction consisted in determination of the dependence of FLC static and frequency characteristics on its settings. The second one was an analysis of the frequency characteristics in the fuzzy control systems, the construction of FLC linear model in a form of PD controller and determination of the dependence of its parameters on FLC settings. The conducted research allows us to draw the following conclusions. 1. Selection of the input and output variables and the number of terms should be based on the structure of the control system and on the requirements to its quality, avoiding their unreasonable increase. 2. The logical basis should be chosen at the stage of the system design and should not be changed in the process of its functioning. Thus, the algebraic basis ensures the least nonlinear distortion of FLC characteristics, making its behavior quite predictable. On the other hand, the control surface of FLC, when using the maxmin basis, has a small slope near the origin of the coordinates, which enhances the stabilizing properties of the controller. 3. The turndown and dilatation-concentration degree of the membership functions of the fuzzy variables make it possible to change the system's behavior smoothly in a wide range and can be used for an automatic tuning of FLC during a normal operation. 4. The fuzzy relation between the space of the antecedents and the space of the consequents (the rule base) should be tuned in a test mode before the system operation and should not be changed hereafter. 5. It is advisable to carry out a procedure of defuzzification by the method of the center of gravity. 6. Measures of the sub-condition significance can be used for an automatic tuning of FLC in cases, when the parameters of the control object in the process of operation do not change by more than one order.
Keywords: fuzzy logic inference; dynamic characteristics; membership function; relational model; aggregation of the sub-conditions; truth degrees; measure of significance; fuzzy controller; approximating model; PD-controller

For citation:

Anisimov D. N., Mai Tkhe Anh. Dynamic Properties of the Fuzzy Control Systems Based on the Relational Models, Mekhatronika, Avtomatizatsiya, Upravlenie, 2017, vol. 18, no. 5, pp. 298—307.

To the contents