Adaptive Fuzzy Control of Synchronous Compensator by Neural Technology

Abstract

Improving the efficiency of power supply systems of the industrial enterprise is associated with the local sources of reactive power: synchronous compensators, and static condensers which eliminate the transfer of reactive power from the network by the power line. Application of synchronous compensator requires high quality control of the driving current, which depends on the nature of the load (regulated object). Lack of mathematical description of the regulated object precludes the synthesis of the classical PI-, PID- regulators’ settings. The most effective method in that case is to use fuzzy controllers. However, fuzzy controllers tend to err in static conditions. Another drawback of the fuzzy control is the lack of adaptability. The article addresses the issue of building of fuzzy voltage regulator for the synchronous compensator with the aid of neural technology. Voltage regulator contains an adaptive fuzzificator and an activation functions block. Neuron adaptation is carried out by the sequential education method in one iteration step. Activation functions block provides control for the driving current, which stabilizes the voltage on the load node buses. Development of the static adaptive fuzzy controller determines the relevance of this proposal.