An automatic identification of phase inductance for operation of switched reluctance machines without position sensor

Abstract

This paper describes a method to operate a switched reluctance machine (SRM) sensorless with a precise position estimation but without the necessity of a prior known machine characteristic. When a machine is connected to the inverter for the first time, an initialization sequence is started to determine the inductance profile of the machine automatically. Within this sequence the machine is started and the inductance of all phases is recorded over time while the machine is rotating. From the data the rotor position is extracted and linked to the recorded inductance values. The data is finally merged to one inductance profile, which is used for controlling the machine. Only information about the machine configuration, the maximum allowable machine speed and phase currents must be provided. The proposed method was implemented on a test bench and verified with a 2 kW four-phase SRM using direct average torque control (DATC).

A novel and precise on-line digital measurement of unsaturated phase inductance of a switched reluctance motor

“…In a broad perspective, the measurement methods can be categorised as off-line and on-line methods. In case of off-line measurements, the recorded data of applied phase voltage, using either a single-phase AC supply [13,14] or a pulsed DC source [15][16][17][18][19], and phase currents are obtained from the repeated experiments performed by locking the rotor of an SRM at different angular positions. They are further processed either analytically or by means of a software subroutine in a computer [13][14][15][16][17][18][19].…”

“…In case of off-line measurements, the recorded data of applied phase voltage, using either a single-phase AC supply [13,14] or a pulsed DC source [15][16][17][18][19], and phase currents are obtained from the repeated experiments performed by locking the rotor of an SRM at different angular positions. They are further processed either analytically or by means of a software subroutine in a computer [13][14][15][16][17][18][19]. The precision of the methods [13][14][15][16][17][18][19] depends on the locking mechanism and the resolution of angular rotation of the rotor in each step of measurement.…”

“…They are further processed either analytically or by means of a software subroutine in a computer [13][14][15][16][17][18][19]. The precision of the methods [13][14][15][16][17][18][19] depends on the locking mechanism and the resolution of angular rotation of the rotor in each step of measurement. Moreover, the methods also require extra mechanical arrangements and hardware to rotate and lock the rotor precisely at planned discrete and numerous positions throughout the rotor tooth pitch.…”

“…The authors also claim novelty of the present work in terms of absence of any additional computational platform over and above that required to operate the SRM-based drives. Additional mechanical hardware is also not used for locking the rotor precisely at planned discrete positions, unlike [13][14][15][16][17][18][19]. Apart from this, the proposed method being online in nature does not utilise any soft computing/regression techniques for prior training/learning of the model unlike [20,21].…”

Technology, Research and Applications of Switched Reluctance Drives

“…Today, often a LUT-based approach is used, whereby the tables are provided by either FEA or measurements. To reduce the initialization effort and to account for production uncertainties, the amount of auto-parameterizing [41], genetic algorithms, self-learning and neural network tuning algorithms, that adjust the characteristic stored in the LUT to the actual machine, has increased steadily. Alternatively, to model saturation and phase coupling in an SRM, Fleming proposes to use MEC-based models [42], [43].…”