부산대학교 도서관

The use of signal injection methods in the estimation of magnet temperature is an innovative and non-intrusive approach used in permanent magnet motors. One of the biggest challenges in these approaches however involves the determination of the optimal frequency of the injected signal in order to accommodate its detection, accuracy, sensitivity and implementation. This paper presents an innovative strategy which will dynamically determine the bandwidth of suitable frequencies at which these signals may be injected in a five-phase Permanent Magnet Assisted Synchronous Reluctance Motor (PMa-SynRM). The process involves the elimination of the known fundamental, harmonic and fault dependent frequencies from the current frequency spectrum to resolve a range of remaining frequencies at which the signal may be injected. A randomization process is then adopted to determine a frequency from the dynamically determined range of available frequencies to effectively estimate the magnet temperature. The experimental analysis has been conducted on a five-phase 3.7kW PMa-SynRM and the values obtained have been verified using an innovative wireless temperature measurement setup.