Study and detection of demagnetization in line start permanent magnet synchronous machines using artificial neural network

This paper makes an effort to study the causes and effects of permanent magnet demagnetization in permanent magnet machines and hence proposes an exclusive artificial neural network (ANN) based permanent magnet demagnetization detection scheme. A laboratory 2.8 kW line-start permanent magnet synchronous machine (LSPMSM) is used in the numerical investigations for initiating permanent magnet demagnetization and detecting the fault. Firstly, experiments were performed on the machine to determine it parameters and understand its steady-state and dynamic performance using a developed position sensor and an experimental setup. A mathematical model of the machine was then developed using the d-q axis theory to analyze the behavior of the machine under healthy and demagnetization conditions. Later, an ANN based detection scheme is proposed and verified through numerical investigations. The results obtained from the investigations are thus analyzed.