With the applications of permanent magnet synchronous motor (PMSM) widely, the PMSM fault detection strategies are getting more important. One of the most common faults is partial permanent magnet (PM) demagnetization. Although the FFT of the stator current signal has been used to detect permanent magnet (PM) demagnetization faults, it’s only useful in the stationary conditions. For purpose of achieve the partial demagnetization detection on the nonstationary conditions, this paper proposes a novel partial demagnetization fault detection algorithm based on the FFT analysis. Firstly, this paper introduces the mechanism of partial demagnetization of a PMSM. Then, by transforming the current signals in the time-domain to the ones in the position-domain with a constant amplitude and frequency, they become stationary so that the FFT can be applied even in the non-stationary conditions. In the end, the simulation results verify that the proposed method is effective.