Nonlinear Modeling and Estimation of Voltage to Current Angle Derivative for Accurate Inductance Identification of PMSMs

For permanent magnet synchronous machines (PMSMs), accurate inductances are critical to achieve high performance control, but inductance estimation can be affected by machine nonlinearities and the coupling with other machine parameters. Hence, this article presents the inductance estimation from a new perspective by using the derivative of the voltage to current angle (VCA) to improve the estimation accuracy especially for the d-axis inductance. The idea is to derive the derivative of VCA based inductance estimation model from the machine model, which is decoupled from the machine parameters such as permanent magnet (PM) flux linkage and improves the estimation performance. To consider magnetic saturation, nonlinear inductance variation is modeled in a local region of interest to improve the estimation performance. To avoid the influence from winding resistance and inverter distortion, the voltage variation under two speeds is employed to construct the estimation model. To reduce the influence from the measurement noise, the derivatives of VCA are modeled and estimated from the measurements using nonlinear polynomials. Hence, one can use measurements under low speeds for inductance estimation, which can reduce the influence of nonlinearities such as core loss in using high-speed data. The proposed approach is validated with extensive simulations, experiments, and comparisons on a test PMSM drive.