Online Identification of Six-Phase IPMSM Parameters Using Prediction-Error Sensitivities to Model Parameters

Six-phase Interior Permanent Magnet Synchronous Machines (IPMSM) drives with dual three-phase configuration offer unique merits that make them attractive for reliability-critical applications. Online identification of machine parameters, i.e., permanent magnet flux linkage
$\Psi_{m}$
, stator resistance
$R_{s}$
and inductances can enhance the drive's performance. A full-order model based open-loop predictor yields predicted currents which are sensitive to model parameters. This sensitivity, known as the prediction gradient
$\Psi^{T}$
, is aimed to be exploited in the proposed method to track parameters. Single Synchronous Reference Frame based modeling and current control is adopted for robust performance. Stochastic Gradient Algorithm (SGA) is used to compute the gain-matrix that computes the parameter-updates when discrepancies exist between the physical and model parameters. The concept is validated by demonstrating
$\Psi_{m}, R_{s}$
online-tracking and the influences from wrong inductances are analyzed analytically and with the aid of an Embedded Real-Time Simulator. Results show satisfactory convergence speeds and asymptotic behavior.