Estimation of Permanent Magnet Synchronous Generator Performance with Artificial Neural Network Models

Abstract

The interest in renewable energy sources has grown with the increase of environmental pollution and the decrease of fossil fuels. It is possible to provide energy supply security and diversity by using renewable energy sources. In this regard, wind energy, which is one of the renewable energy sources whose share in energy production increases day by day, emerges as a local and environmentally friendly solution. Many different types of generators are used in wind turbines and these have advantages and disadvantages according to each other. Permanent magnet synchronous generators (PMSG) are preferred because of their advantages such as high efficiency, high power density and being used directly in wind turbines without the need for gear system. In this study, the performance of the 2,5 kW PMSG, with a 14-pole surface placement, internal rotor, suitable for use in wind turbines, has been examined by changing the physical structure of the magnet. For this purpose, performance parameters such as total magnet consumption, efficiency, power loss have been successfully estimated using single and double hidden layered multi layer neural network (MLNN), elman neural network (ENN) and radial basis function neural network (RBFNN).

Keywords

• multilayer neural network
• elman neural network
• radial basis function neural network
• permanent magnet synchronous generator
• çok katmanlı sinir ağı
• elman sinir ağı
• radyal tabanlı fonksiyon sinir ağı
• sabit mıknatıslı senkron generatör