Sensorless self-commissioning of synchronous reluctance motors at standstill

Abstract

This paper proposes a standstill method for identification of the magnetic model of synchronous reluctance motors (SyRMs). The saturation and cross-saturation effects are properly taken into account. The motor is fed by an inverter with a short sequence of bipolar voltage pulses that are first applied on the rotor d-and q-axes separately and then simultaneously on both the axes. The stator flux linkages are computed by integrating the induced voltages. Using the current and flux samples, the parameters of an algebraic magnetic model are estimated by means of linear least squares. The proposed method is robust against the stator resistance variations and inverter nonlinearities due to the high test voltages (of the order of the rated voltage). The fitted model matches very well with the reference saturation characteristics, measured using a constantspeed method, and enables extrapolation outside the sample range. The method was tested with a 2.2-kW SyRM, whose shaft was uncoupled from any mechanical load, which is the most demanding condition for this method. The proposed method can be used for automatic self-commissioning of sensorless SyRM drives at standstill.

Sensorless commissioning of synchronous reluctance machines augmented with high frequency voltage injection

“…The amplitude of the test square wave voltage in d and q axes are called , and , , respectively. Details can be found in [11].…”

“…For practical implementation though, any rotor misalignment must be taken into account. Therefore, test #1 will be performed as in [11], using the initial off-line estimation of the rotor position. Test #2 and Test #3 will be assisted by online sensorless position estimation to extend the feasible current limit , extend the measurement domain in the current plane , .…”

“…For example, the flux maps can be implemented in form of look up tables or, alternatively, using analytical functions. The polynomial magnetic model presented in [11] is based on few parameters and was successfully tested on SyRMs of different size. For this reason it will be used also in this work.…”

“…In particular, a and a are obtained from test #1, a and a from test #2 and a from test #3. Details can be found in [11].…”

“…In [10] [11], the rotor position is off-line estimated before the test through high frequency (HF) signal injection and demodulation, or determined via initial parking through DC excitation, and then considered constant during the test. In turn, any unwilled movement of the rotor provoked by the machine response to the excitation of [10] [11] would make the initial position estimation inaccurate, to the point that the rotor would even start spinning and the test would fail. This shortcoming limits the feasible measurement area in the dq current plane.…”