Аннотация: Рассматривается применение спектрального метода к синтезу систем управления электроприводами. Предлагается использовать отличную от традиционной систему базисных функций. Показано, что диадное вейвлет-преобразование сигналов удовлетворяет основным требованиям к системам базисных функций.

The paper demonstrates a method for optimizing a fuzzy controller through a genetic algorithm that uses an elitism strategy. A simplified model of a regulated four-way intersection was used as the object of regulation, describing an intersection with one lane of traffic in each direction. A series of computer experiments on optimization of the fuzzy controller was performed for the set values of the traffic intensity. The results of several runs, each with the duration of the simulation was limited to ten generations, demonstrated repeatability of the settings, on the basis of which a conclusion about the achievement of the maximum efficiency control method was made. The results of modeling traffic light regulation in three variants are compared: 1) fixed duration of traffic light phases, 2) the rule base for the fuzzy controller is compiled by an expert method, and 3) the rule base for the fuzzy controller is optimized by a special evolutionary algorithm. It is shown that the average tube length for the optimized regulator in the experiments sought to the minimum value (about 11 vehicles, which is almost twice less than in the preparation of the expert rule base of a fuzzy controller and about four times less than for a fixed cycle traffic light).