Noise source recognition of permanent magnet synchronous motor based on order analysis

Abstract: The noise source of permanent magnet synchronous motor (PMSM) was identified in this paper. Firstly, the noise signal in the acceleration process was tested and the short time Fourier transform (STFT) was used to perform the time-frequency analysis. Main noise characteristics including frequency order and resonance region were obtained. Then, the order feature of rolling bearing was studied and the fractional order noise produced by ball pass frequency was recognized. Based on the Maxwell stress tensor method, the frequency order of radial electromagnetic force acting on the teeth surface was derived. The characteristics of the ideal radial force under sinusoidal current with no rotor eccentricity were derived. The influence of current harmonics on the radial force was discussed. Current in the acceleration process was monitored and STFT was used to analyze the type of current harmonics. It was found current harmonics could be divided into 2 types: 1) current harmonics which were multiples of the fundamental current frequency and 2) current harmonics which were close to the switching frequency. The force order produced by the first type of current harmonics was the same as that in the ideal condition and the second type of current harmonics produced force harmonics around the switching frequency. By introducing the eccentricity correction factor the force harmonic due to rotor dynamic eccentricity was analyzed. It was found that rotor dynamic eccentricity induced extra space and frequency harmonics. Especially, low space harmonic which contributed most to the overall noise level was produced in the eccentricity case. In order to obtain the modal parameters in the resonance region, the finite element (FE) model of stator was established. The material anisotropy of stator core and winding was considered according to the actual structure of stator and modal shape. Modal tests of stator core and stator assembly were conducted to validate the equivalent model of stator core and winding, respectively. The relative errors of modal frequency of the stator core and stator assembly model were all below 5% and satisfied the requirement of engineering error. Furthermore, constraint modal analysis of stator was conducted by applying the actual constraints in the test bench to the equivalent model, and modal parameters of the stator under actual installation conditions were obtained. Finally, with the excitation feature information of rolling bearing and electromagnetic force accounting for current harmonics and rotor dynamic eccentricity, the source of main tested noise order was identified. The modal parameters in the resonance region were also recognized through the modal analysis of stator. It is concluded from the recognition research in this paper that the fractional order noise comes from the ball frequency which mainly depends on the number of balls. In the ideal condition, the frequency order of electromagnetic force is the even multiple of fundamental current frequency. The influence of current harmonics on electromagnetic noise depends on the harmonic type. Current harmonics close to the switching frequency play a more important role than those which are multiples of the fundamental current frequency. The noise order produced by rotor dynamic eccentricity mainly comes from the resonance when the force harmonics due to the eccentricity pass through the modal frequency of stator. The contribution of each source to the overall noise was also analyzed. And it is found that the noise produced by the force due to the interaction of permanent magnet field and armature reaction field from current harmonics contributes the most remarkable part, the noise produced by rotor eccentricity takes the second place and the fractional order noise by rolling bearings comes last. The work in this paper can be used to identify and diagnose the noise source of PMSM and lay the foundation for the vibration and noise reduction further.