Discussion on: “An Adaptive Variable Structure Control Law for Sensorless Induction Motors”

Abstract

During recent years, the research efforts in the fieldof induction motor control have been mainly focusedon ‘sensorless’ solutions. Sensorless control couplesthe advantages given by the use of induction motorscompared to the other kind of electric machines (see,for instance, [1]) with the possibility of reducing therealization costs of the control system, thanks to theelimination of the sensors relevant to the mechanicalvariables [2]. In induction motors, with or withoutsensors to measure the rotor flux and speed, the prob-lem of controlling the speed (torque) and the flux isquite difficult to solve, because of the system non-linearities, the strict coupling between the state vari-ables, and the presence of critical parameters suchas the rotor resistance, the values of which may sig-nificantly change with respect to their nominal valueduring operations. As a consequence, high perform-ance and high robustness properties are required bythe control and the observer algorithms.A great number of valid proposals of sensorlesscontrol schemes for induction motors has appeared inthe literature recently (see, for instance, [3], [4], [5], [6],[7], [8] and the references cited therein). Among these,many proposals rely on the possibility of solving theinduction motor control problem, even in the sensor-less case, via a variable structure control approach,forcingthecontrolledsystemtooperateintheso-called‘sliding mode’ [9]. The sliding mode control designmethodology, capable of guaranteeing high levels ofrobustness against matched disturbances and para-meter variations, seems quite appropriate to designboth the observers and the controllers for inductionmotors (see, for instance, [10], [11], [12], [13], [14], [15],[16], [17], [18]).The paper by Barambones and Garrido presentsa further proposal in this context. It is based on theuse of field oriented control [1], but by virtue of theuse of a suitably designed speed observer, can beexploited even in sensorless applications. The paperby Barambones and Garrido exploits the potential-ities of the sliding mode approach only in the designof the control loop. As for the design of the fluxand speed observer, it relies on a Luenberger observer(for rotor flux and stator currents) coupled with anadaptive mechanism to estimate the rotor speed. Thecontrol variable becomes discontinuous, as the slid-ing variable, depending on the speed tracking erroranditsintegral,vanishes.Thecontrolamplitudeisnotselected on the basis of known upperbounds of theuncertainty terms, as is common in sliding modecontrol design [19], but it is adapted through a tuningmechanismwhichisswitchedoffonlywhentheslidingmanifold is reached. This, eliminates the necessity ofthe knowledge of the upperbounds, which is clearly abenefit. Yet, the proposed gain adaptation mechan-ism can be conservative, in the sense that the steady-state value reached by the switching gain is not thesmaller value which allows to overcome the actualsystems uncertainties. This simply depends on the factthattheproposedadaptationmechanismalwaystendsto increase the switching gain unless the sliding vari-able is equal to zero. Indeed, since the system uncer-