WO1997011524A1

WO1997011524A1 - Method and apparatus for control of a switched reluctance motor

Info

Publication number: WO1997011524A1
Application number: PCT/US1996/015037
Authority: WO
Inventors: David G. Taylor; Saeed Ur Rehman
Original assignee: Georgia Tech Research Corporation
Priority date: 1995-09-20
Filing date: 1996-09-19
Publication date: 1997-03-27
Also published as: EP0792537A4; JPH10509859A; US5982117A; CA2205588A1; EP0792537A1; AU7365096A

Abstract

A method for estimating and controlling the rotor position of a SR motor comprises the steps of sampling phase current(s) (53) and estimating phase flux(es) (54) from at least one conducting phase of the SR motor, sampling phase current(s) (52) and estimating phase inductance(s) (54) from at least one unexcited phase of the SR motor, and determining the rotor position (55) by combining the estimated phase flux and the estimated phase inductance (54). Once the rotor position has been determined/estimated by the above method, the position of the rotor can be controlled so as to provide, overall, and effective control of the SR motor. In particular, once the position of the rotor is determined/estimated, the excitation provided to the motor is then adjusted (56) to urge the rotor from its current position towards its desired position. In this way, position control of the motor can be obtained. Of course, this can be extended to obtain velocity control as well.

Description

METHOD AND APPARATUS FOR CONTROL OF A SWITCHED RELUCTANCE MOTOR

BACKGROUND OF THE INVENTION Field of the Invention

The present invention relates, in general, to a method and apparatus for controlling a switched reluctance motor and, more specifically, to sensorless position and velocity control using rotor position estimation based on magnetic flux and inductance signals. Description of Related Art

A switched reluctance (SR) motor is a brushless, salient pole machine with multiple poles on both the rotor and stator. The stator has phase windings, unlike the rotor which is unexcited having no windings or permanent magnets mounted thereon. Rather, the rotor ofa SR motor is formed of a magnetically permeable material, typically iron, which causes the rotor to rotate when current to the stator phase windings is switched on and off in a sequential fashion in correspondence to rotor position. However, the rotor position information required for accurate and precise control is a double-valued function (within each electrical cycle) of either phase flux or phase inductance, for any fixed value of phase current. Thus, given a measurement of current and a calculation of either flux or inductance alone, a pair of possible rotor positions can be determined.

Traditionally, the rotor position information required for phase current switching has been provided by such devices as Hall-effect sensors, optical switches, optical encoders, or resolvers. More recently, efforts have been made to dispense with the use of such mechanical position sensors in favor of position estimators based on quantities determinable from the motor terminals. For example, U.S. Patent No. 4,772,839 discloses a rotor position estimator for SR motors which simultaneously measures changes in current in two idle (unexcited) phases, processes the measurements to provide a pair of possible rotor angles for each such phase, and combines the angles in a fashion which yields a unique estimate of instantaneous rotor position. If two phases ofthe SR motor do not remain unenergized throughout the sampling period or if any phase ofthe motor experiences a change of state during the sampling period, the '839 estimator provides an extrapolated rotor position in place of the estimated instantaneous rotor position. However, the estimator disclosed in the '839 patent estimates rotor position based on the assumption that the IR drop (voltage drop due to electrical resistance) and the back EMF (an electromagnetic force caused by induction) term ofthe phase voltage are so small as to be negligible when estimating phase inductance. Such an assumption is not always valid.

U.S. Patent No. 4,959,596 discloses a SR motor drive system in which an "impedance sensing" technique is used to implicitly estimate certain rotor positions. This technique involves applying voltage sensing pulses to an idle phase, which results in a change in phase current. The change in phase current is inversely proportional to the instantaneous phase inductance. Commutation time is determined by comparing the change in phase current to a threshold current, thus synchronizing phase excitation to the rotor position. Since no explicit determination of rotor position is obtained, the excitation to the phases cannot be shaped with respect to the position, limiting the control performance that can be achieved (e.g., position control is not possible).

U.S. Patent No. 5,097,190 discloses a rotor position estimator for a SR machine based on instantaneous phase flux and phase current measurements. Phase current and flux sensing are performed in a predetermined sequence that depends on the particular quadrant of operation, i.e., forward motoring, reverse motoring, forward generating, or reverse generating. For each phase in the predetermined sequence of sensing, phase flux and phase current measurements are made during operation in a pair of predetermined sensing regions, each defined over a range ofthe rotor angles. The rotor angle estimates are derived from the phase flux and phase current measurements for each respective phase during the respective sensing regions thereof. The rotor angle estimates for each phase are normalized with respect to a common reference phase, and a rotor position estimate for the SR machine is computed according to an equation which accounts for the fact that for any given rotor angle determined, the rotor poles ofthe SR motor may be approaching alignment or unalignment. The use of two simultaneously conducting phases is not disclosed. The above-described prior art generally fails to provide an unambiguous determination of rotor position (i.e., the '596 patent does not produce explicit position estimates and only provides timing signals for commutation purposes, and the '839 and ' 190 patents each lead to two possible rotor position estimates at least some ofthe time). Thus, a need still exists for a means of completely eliminating ambiguity in rotor position estimations for a SR motor. It is to the fulfillment of this need that the present invention primarily is directed. SUMMARY OF THE INVENTION

Briefly described, a preferred form ofthe present invention comprises a method for estimating the rotor position of a SR motor. The method comprises the steps of sampling phase current(s) and estimating phase flux(es) from at least one conducting phase ofthe SR motor, sampling phase current(s) and estimating phase inductance(s) from at least one unexcited phase ofthe SR motor, and determimng the rotor position by combining the estimated phase flux and the estimated phase inductance.

Preferably, the step of estimating the phase flux of at least one conducting phase includes sampling a phase voltage and taking an integration ofthe sampled phase voltage and phase current to obtain a phase flux estimate. Also preferably, the step of estimating the phase inductance includes applying a voltage pulse to at least one unexcited phase and measuring a change in current due to the voltage pulse.

Preferably, either a flux estimate or inductance estimate is determined for all phases at all times. Also preferably, the rotor position is determined from a least- squares solution ofthe flux and inductance characteristics. Preferably, the least- squares solution is evaluated using different weights for each phase equation.

Once the rotor position has been determined/estimated by the above method, the position ofthe rotor can be controlled so as to provide, overall, an effective control ofthe SR motor. In particular, once the position ofthe rotor is determined/estimated, the excitation provided to the motor is then adjusted to urge the rotor from its current position towards its desired position. In this way, position control ofthe motor can be obtained. Of course, it will be clear to those skilled in the art that the method ofthe present invention can be extended to obtain velocity control as well. A major advantage or benefit ofthe present invention is that the determination or estimation ofthe rotor position results in unambiguous rotor position. This is in contrast to the known prior art in which for a given inductance or flux level, two different possible rotor positions exist. By eliminating this ambiguity in the position information, much more effective control ofthe motor can be effected without needing to know a prior history of the motor's position.

Accordingly, it is the primary object ofthe present invention to provide a method for estimating rotor position ofa SR motor which is simple, effective, and reliable in operation.

It is another object ofthe present invention to provide a method for estimating rotor position of a SR motor which results in an unambiguous estimate of the position of the rotor.

It is another object ofthe present invention to provide accurate position estimations of the rotor of a SR motor so as to provide or enable effective control.

These and other features, advantages, and objects ofthe present invention will become more apparent upon reading the following specification in conjunction with the accompanying drawing figures.

BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a schematic illustration ofa mechanical configuration for carrying out the method ofthe present invention. Fig. 2 is a schematic, functional block diagram of a simple control scheme implemented using the present invention to control a SR motor.

Fig. 3 is a plot of phase flux versus rotor position characteristics, parameterized by phase current.

Fig. 4 is a plot of phase inductance versus rotor position characteristics, parameterized by phase current. Fig. 5 is a diagram illustrating the method ofthe present invention, which demonstrates how a flux estimate of at least one excited phase and an inductance estimate of at least one unexcited phase are sufficient to resolve the ambiguity regarding rotor position.

DETAILED DESCRIPTION OF THE DRAWINGS Referring now to Fig. 1 , an illustrative mechanical configuration for carrying out the method ofthe present invention is depicted. In this illustrative mechanical configuration 10 a SR motor 12 is controlled without traditional mechanical sensors connected to it. The SR motor 12 contains internal (unshown) phase or stator windings which are electrically connected to an amplifier 32 via electrical cabling 38. In this way, the amplifier 32 commands phase input voltages to be applied to the stator windings. A current sensor 24 measures stator currents which are sensed as phase current measurements by a microprocessor 26 by way of analog-to-digital (A/D) converter (not shown). Such A/D converter is of conventional design and is well known in the art. While separate hardware components are shown for the amplifier 32 and the microprocessor 26, obviously these functions can be provided in a single unit (an integrated component).

Referring now to Fig. 2, a schematic, functional block diagram is depicted. As depicted in Fig. 2, a start-up sequence is performed to find the initial rotor position (see Block 51). This may be accomplished by probing all the motor phases, estimating their phase inductances, and then estimating the initial rotor position from these inductance estimates. Alternatively, one motor phase may be energized to bring the rotor to a known position. Once the initial position is known, the initial excitation is calculated and applied to the motor (see Block 52). After this initial excitation to the motor, the phase current and phase voltage are sampled from at least one conducting phase, and the probing current due to application of a probing voltage in at least one unexcited phase is sampled (see Block 53). Next, the phase flux from at least one conducting phase and the phase inductance from at least one unexcited phase are estimated (see Block 54). Next, the rotor position is determined estimated by combining the phase flux and the phase inductance estimates (see Block 55). The new rotor position estimate is then used to calculate a new excitation to command the motor (see Block 56). This new excitation for the motor is then applied to the motor (see Block 57). This cycle continues as long as the motor is being operated. Having provided a brief overview of the method according to the present invention, a more in depth presentation ofthe method follows:

The method ofthe present invention combines inductance estimates and flux estimates in the position estimation process and provides a unique estimate at every sampling instant, generated using the latest phase measurements, and hence never needs to extrapolate previous estimates. The method is capable of providing continuous estimates of rotor position and hence can be used either to simply commutate the motor at proper instants for average torque control, or can be used as part of a sensorless control scheme for instantaneous torque control (required for any servo system). The method uses a piecewise representation ofthe pertinent motor characteristics, i.e. phase inductance L(θ, ) and phase flux-linkages Ψ(θ, ). Each of these functions is dependent upon two variables, i.e. rotor position, θ, and phase current, /^', and represents a surface in a two-dimensional plane. This surface can be divided into a number of rectangular domains, and within each domain these functions can be modeled using a simple bilinear equation, such as L(θ, ) = A₁(/) + A₂( )θ (1)

Ψ(θ, ) = B,( + B₂(/)θ (2)

The coefficients used to evaluate A(i) or B( ), in the above equations, vary from domain to domain and may be determined off-line from various unconstrained linear least-squares problems. This particular representation ensures continuity of the functions even at the domain boundaries. In addition, as will be shown later, this particular representation makes it possible to estimate position using a linear least- squares formulation, thus adding a degree of noise immunity.

Although phase inductance varies with angle, this variation is more pronounced at low values of current than at higher values because saturation reduces the phase inductance (See Fig. 4). This suggests the use of an unexcited phase. instead of an excited phase, for the estimation of phase inductance, which in turn could be used to estimate rotor position.

Phase flux, on the other hand, is also a function of phase current and rotor position, but its variation with respect to position is greater for larger currents than for smaller currents (See Fig. 3). This suggests that any method which estimates rotor position from phase flux estimates should make use of excited phase measurements, instead of obtaining the flux estimates by probing an unexcited phase.

As shown in Fig. 3 and Fig. 4, rotor position is a double-valued function of either phase inductance or phase flux, for any fixed value of phase current. This means that any attempt to invert the flux (inductance) functions, given a measurement of current and an estimate of flux (inductance), would result in a pair of possible rotor positions instead of one unique value. As shown in Fig. 5, the applicants have overcome this hurdle by the use of at least two pieces of data from at least two different phases, e.g. either two or more phase inductances, two or more phase fluxes, or one or more phase inductance and one or more phase flux, to solve for a unique rotor position. For example, the use of a single inductance estimate would provide two rotor position estimates (Θ_A1 and θ^), and use of a single flux estimate would provide two rotor position estimates (Θ_B] and Θ_B2), but the combined use of inductance and flux estimates would indicate a unique position estimate (the angle indicated by both Θ_A] and Θ_B1). For a 3-phase SR motor, however, it is never possible to find either two excited or two unexcited phases at all times. Thus, prior art techniques cannot resolve the double-valued ambiguity in rotor position; indeed, ambiguity will exist for at least 50% ofthe time assuming an optimally controlled 3-phase motor using known prior art techniques. The existing techniques make use of heuristics, in one way or another, to produce an estimate of rotor position when only one piece of data is available (i.e. for 50% ofthe time, for a 3-phase motor).

The method ofthe present invention solves the ambiguity problem by combining the use of flux and inductance estimates. In the normal mode of operation for a SR motor, there will always be at least one phase which is excited and at least one phase which is unexcited, even for 3-phase motors, and hence the two pieces of data needed to resolve position ambiguity are always available. Each excited phase can be used to produce an estimate of phase flux Ψ, while each unexcited phase can be probed to get an estimate of phase inductance L. Each of these estimates, together with their respective phase currents, generates a set of possible domain values which can be matched to determine a common domain. Once this common domain is known, a system of linear equations can be written to identify the unknown rotor position. For each unexcited phase an equation ofthe form

A,(z) + A₂(i) θ = L (3) is considered, and for each excited phase an equation ofthe form B,( + B₂( θ = Ψ (4) is considered. Once the system of constraint equations is formulated, a least-squares solution for θ can be obtained to estimate the rotor position. Moreover, weighted least-squares methods can also be used to take advantage of any knowledge regarding which phases are the best indicators of position at any given point in time. Means for computing estimates of flux and inductance, given measurements of phase voltage and phase current, are disclosed in prior art. Phase flux can be estimated by evaluating the right-hand side of the equation

Ψ( = Ψ(t_o) + J (v(τ) - Jϋ(τ)) * (5) t₀ where v is the phase voltage, is the phase current, R is the phase resistance, and t₀ is the instant at which time-integration beings. Various means for approximating this fundamental expression for phase flux are readily apparent. For example, if the integration is synchronized with the excitation ofthe phase, then the initial phase flux at t₀ can be assumed to be zero. Phase inductance can be estimated by evaluating the right-hand side ofthe equation

L(t) = vTΛ - Ri(ή - ed. i^'(t) (6) where v is the phase voltage, i is the phase current, R is the phase resistance, e is the back-emf, and the ' denotes time-differentiation. Various means for approximating this fundamental expression for phase inductance are readily apparent. For example, if the motor is operating in a relatively low-speed regime, then the back-emf can be assumed to be zero.

The present invention combines both flux and inductance estimates in the rotor position estimation procedure, and addresses the measurement noise issue through the use of least-squares data fitting. Unlike prior art methods, the method ofthe present invention is not confined to deriving rotor position estimates from inductance measurements alone or from flux measurements alone. The estimates of phase inductance from at least one phase and phase flux from at least one other phase are processed together to define the (unique) rotor angle. The present invention does not require use of extrapolation, nor does it need information about the particular mode of operation (e.g., forward or reverse motoring or generating). By combining flux estimates and inductance estimates, the present invention eliminates rotor position ambiguity 100% ofthe time for 3-phase SR motors.

Preferably, either a flux estimate or inductance estimate is determined for all phases at all times. Also preferably, the rotor position is determined from a least- squares solution ofthe flux and inductance characteristics. Preferably, the least- squares solution is evaluated using different weights for each phase equation. Although the invention has been described with reference to preferred embodiments thereof, it is to be understood that these embodiments are merely illustrative ofthe application ofthe principles ofthe invention. Numerous modifications may be made therein without departing from the scope and spirit ofthe invention as set forth in the following claims.

Claims

WHAT IS CLAIMED IS:

1. A method for estimating the rotor position of a switched reluctance motor, comprising the steps of: a. estimating phase flux from at least one excited phase; b. estimating phase inductance from at least one unexcited phase; c. determining the rotor position by combining the estimated phase flux(es) and the estimated phase inductance(s).

2. The method of Claim 1 wherein said step of estimating phase flux comprises time integration ofa signal determined from phase voltage and phase current from said at least one excited phase.

3. The method of Claim 1 wherein said step of estimating phase inductance comprises applying a voltage pulse to said at least one unexcited phase and measuring a change in current.

4. The method of Claim 1 wherein one of either a flux estimate or an inductance estimate is determined for all phases at all times.

5. The method of Claim 1 wherein the rotor position is determined from a least-squares solution of flux and inductance characteristics.

6. The method of Claim 5 wherein said least-squares solution is evaluated using different weights for each phase equation.

7. A method for controlling the rotor position of a switched reluctance motor, comprising the steps of: a. estimating phase flux from at least one excited phase; b. estimating phase inductance from at least one unexcited phase; c. determining the rotor position by combining the estimated phase flux(es) and the estimated phase inductance(s); d. determining the rotor velocity by processing the estimated rotor position; e. applying excitation to the motor in response to the determined rotor position and rotor velocity, to urge the rotor position toward a desired rotor position setpoint or desired rotor position trajectory.

8. The method of Claim 7 wherein said step of estimating phase flux comprises time integration of a signal determined from phase voltage and phase current from said at least one excited phase.

9. The method of Claim 7 wherein said step of estimating phase inductance comprises applying a voltage pulse to said at least one unexcited phase and measuring a change in current.

10. The method of Claim 7 wherein one of either a flux estimate or an inductance estimate is determined for all phases at all times.

11. The method of Claim 7 wherein the rotor position is determined from a least-squares solution of flux and inductance characteristics.

12. The method of Claim 11 wherein said least-squares solution is evaluated using different weights for each phase equation.

13. The method of Claim 7 wherein said step of determining rotor velocity comprises approximate time-differentiation ofthe estimated rotor position.

14. A method for controlling the rotor velocity of a switched reluctance motor, comprising the steps of: a. estimating phase flux from at least one excited phase; b. estimating phase inductance from at least one unexcited phase; c. determining the rotor position by combining the estimated phase flux(es) and the estimated phase inductance(s); and d. determining the rotor velocity by processing the estimated rotor position; e. applying excitation to the motor in response to the determined rotor position and rotor velocity, to urge the rotor velocity toward a desired rotor velocity setpoint or desired rotor velocity trajectory.

15. The method of Claim 14 wherein said step of estimating phase flux comprises time integration of a signal determined from phase voltage and phase current from said at least one excited phase.

16. The method of Claim 14 wherein said step of estimating phase inductance comprises applying a voltage pulse to said at least one unexcited phase and measuring a change in current.

17. The method of Claim 14 wherein one of either a flux estimate or an inductance estimate is determined for all phases at all times.

18. The method of Claim 14 wherein the rotor position is determined from a least-squares solution of flux and inductance characteristics.

19. The method of Claim 18 wherein said least-squares solution is evaluated using different weights for each phase equation.

20. The method of Claim 14 wherein said step of determining rotor velocity comprises approximate time-differentiation ofthe estimated rotor position.