Finite-Element Analysis of Electrical Machines for Sensorless Drives With High-Frequency Signal Injection

Abstract

Abstract-A challenge during the design process of an electrical machine is the characterization of the various parameters in a computational time as short as possible. Frequently it is required the computation of the electrical machine parameters which are necessary for the tuning of the drive control algorithm. This paper deals with a strategy to compute the high frequency signal injection response of a sensorless controlled electrical machine. It allows to determine the self-sensing capability of the machine directly during the design process. Such a capability can be defined in any given operating point, for example along the Maximum Torque Per Amps trajectory. Then, also the high frequency machine parameters can be computed.In addition, the strategy proposed here requires a very short computational time to get such a data. After a magnetostatic field analysis, carried out so as to get the torque for a given current, the flux density distribution is stored and the differential reluctivity tensor is evaluated in each element of the mesh. Then, a time-harmonic analysis is carried out in a linearized structure so as to compute the d-q parameters at the injection frequency.In order to validate the proposed procedure, experimental results on different machine type are included in the paper. This allows to prove the reliability of the procedure as a valued tool for the characterization of the machine.

Finite Element Small-Signal Simulation of Electromagnetic Devices Considering Eddy Currents in the Laminations

“…where I 1 and I 2 are the bias current components which produce the main magnetization of the system,î 1 and i 2 the sinusoidal small-signal currents superimposed to the bias components, while -1 and -2 are the shift displacement of the small-signal waveforms. Since theî 1 andî 2 are small-signals the ratio between the injection signals and the bias components is assumed less than 5%, otherwise, nonlinear phenomena can introduce error in the computation of the parameters described in (7). More details on this hypothesis are reported in the next sections.…”

“…Often the system parameters are required in order to tune the controller and to fulfill the design requirements such as dynamic response. As an example, the inclusion of eddy currents in the lamination for the computation of the incremental parameters of an electrical machine in order to determine the rotor position has been recently proposed [7], [8]. The inclusion of the parasitic e ects during the computation of the system parameters, allows to evaluate their impact on the parameters themselves.…”

“…In the latter case, the system is supplied with two frequencies: the main excitation (for example the torqueproducing current of the motor) and a high frequency signal (a carrier signal to identify the rotor position). In such a case the system can be profitably simulated using a small-signal approach [7] that relies on the frozen permeability method. Such a method assumes that the two frequencies do not interact significantly for the parasitic e ects.…”

Fast calculation of carrier harmonic iron losses caused by pulse width modulation in interior permanent magnet synchronous motors

“…Recently, a method for fast calculating the PM eddy current losses caused by PWM voltage harmonics in the IPMSM over the entire working range was proposed in [13], which is based on the combination of the frozen differential reluctivity tensor method (FDRTM) [14, 15] and the linear time‐harmonic finite‐element analysis (THFEA) [16]. The key idea of this method is that the flux density variations generated by the harmonic voltages are too small to affect the saturation of the ESSs.…”

“…The key idea of this method is that the flux density variations generated by the harmonic voltages are too small to affect the saturation of the ESSs. Hence, the FDRTM can be applied to obtain the locally linearised model of the IPMSM, which is also named as the small‐signal model of the IPMSM [14]. Based on this model, the relationship between harmonic voltages and the corresponding flux density variations can be conveniently studied.…”