Inductive Air Gap Estimation Method for Active Magnetic Bearings

This application is related to commonly assigned U.S. patent application Ser. No. 07/970,197 of J. P. Lyons, M. A. Preston and G. B. Kliman and to commonly assigned U.S. patent application Ser. No. 07/970,196 of J. P. Lyons and M. A. Preston, both filed concurrently herewith and incorporated by reference herein.

The present invention relates to a method for determining rotor shaft position and additional state information (i.e., air gap velocity and magnetic flux) in an active magnetic bearing system, without requiring air gap position sensors.

Active magnetic bearings have great potential for improving the efficiency of rotating machinery by: reducing or eliminating the complexity of bearing lubrication systems; enabling high rotational speeds; and providing active means of controlling vibrations in complex rotating masses. Recent advances in power electronics and control microprocessors have made active magnetic bearing systems feasible in, for example, gas turbine engine applications.

Presently, however, active magnetic bearings are unsuitable for high-reliability applications, such as aircraft engine rotor support and vibration control. A major problem is lack of accuracy and reliability of air gap position sensors used for determining rotor shaft position. Hence, it is desirable to provide a method for determining rotor gap position in an active magnetic bearing system, without requiring separate air gap position sensors.

In an active magnetic bearing system employing radial or thrust bearings, rotor shaft position is determined without requiring air gap position sensors, resulting in a high reliability system. The flux paths for each respective electromagnetic force actuator of the active magnetic bearing (e.g., comprising a stator pole-pair in a radial magnetic bearing) are isolated in order to minimize the magnetic coupling therebetween. In this way, the current (or magnetomotive force, mmf) in each respective actuator pole-pair is functionally related to the flux therethrough, independent of the currents (or mmf's) in adjacent actuator pole-pairs. Application of a digital pulse width modulated (PWM) voltage waveform to a stator pole during normal bearing operation results in a ripple current waveform at the PWM switching frequency. The resulting change in stator pole current and pole voltage are measured, and the average differential inductance is determined therefrom. Then, with knowledge of the differential inductance and the average pole current, the effective air gap length, air gap velocity, and magnetic flux are determined according to the inherent magnetization characteristic of the magnetic bearing.

The features and advantages of the present invention will become apparent from the following detailed description of the invention when read with the accompanying drawings in which:

FIG. 1a is a front, partially cross sectional, view of an active radial magnetic bearing having isolated magnetic actuators in accordance with the present invention;

FIG. 1b graphically represents magnetically isolated control axes for the active magnetic bearing of FIG. 1a;

FIG. 2 is a graphical illustration of the magnetic characteristic of a force actuator pole-pair of an active radial magnetic bearing such as that of FIG. 1;

FIG. 3 is a graphical representation of a typical digital PWM voltage applied during normal bearing operation to a stator pole-pair and the resulting ripple pole current for the active radial magnetic bearing of FIG. 1;

FIG. 4 is a graphical illustration of the magnetic characteristic used for lookup of air gap length given differential inductance ##EQU1## and current for a magnetic force actuator of an active radial magnetic bearing such as that of FIG. 1;

FIG. 5 is a graphical illustration of the magnetic actuator characteristic of ##EQU2## given air gap length and current useful for air gap velocity calculation of an active radial magnetic bearing such as that of FIG. 1; and

FIG. 6 is a block diagram of one implementation of an axis controller for controlling an active magnetic bearing in accordance with the present invention.

FIG. 1 illustrates a fault-tolerant active radial magnetic bearing 10, such as described in commonly assigned U.S. patent application Ser. No. 07/970,197 of J. P Lyons, M. A. Preston and G. B. Kliman, cited hereinabove.

Magnetic bearing 10 includes a stator 12 and a rotor 14 mounted for rotation within stator 12 and coupled to a shaft 16. Each stator pole has a winding 18 wound thereon in well-known manner. By way of illustration, magnetic bearing 10 of FIG. 1a has three radial control axes, as indicated in FIG. 1b. In particular, stator 12 has twelve electromagnetic stator poles configured as six force-producing pole pairs A1, B1, C1, A2, B2, and C2, with a 60.degree. radial spacing between pole-pairs. Two pole-pairs at 180.degree. radial separation operate as a respective one of the control axes.

Stator 12 is segmented by situating a non-magnetic stator flux barrier 40 between adjacent pole-pairs. The stator flux barriers 40 provide magnetic isolation for each pole-pair and thus allow functioning force-actuation pole-pairs to continue operating even in such close proximity to faulted magnetic poles. Advantageously, any two of the three (or more) control axes are sufficient to maintain rotor suspension in the bearing system. Hence, bearing operation continues even in the presence of faults, such as, for example, faulted magnetic poles, power electronic shorts, and phase power loss. Even greater fault tolerance may be achieved using more than three control axes.

In operation, the magnetic flux produced by the forcing current or magnetomotive force (mmf) in each pole winding 18 circulates through the stator poles of each respective pole pair and through the rotor and stator cores, crossing two air gap lengths g. An exemplary magnetic flux path 50 for pole-pair A1 is illustrated in FIG. 1.

According to the present invention, air gap position is electronically estimated, or inferred, by digitally controlling the pole currents in an active magnetic bearing having magnetically isolated force actuators (i.e., pole-pairs), such as that of FIG. 1.

FIG. 2 illustrates magnetization characteristics (flux versus current) at various air gap lengths, from a minimum gap length g.sub.min to a maximum gap length g.sub.max, for a typical electromagnetic actuator, i.e., pole-pair. At relatively low current levels, the magnetization characteristic is dominated by the reluctance of the air gap. Lower flux per unit current is achieved at larger air gaps. The curves become non-linear at relatively high excitation current levels due to effects of magnetic saturation.

FIG. 3 illustrates a digital pulse width modulated (PWM) voltage waveform applied to a pole-pair during normal bearing operation. The PWM voltage results in a ripple current waveform at the PWM switching frequency, also shown in FIG. 3. According to the present invention, the change in pole current (.DELTA.I=I.sub.pmax -I.sub.pmin) and the applied voltage V.sub.pole are measured using suitable current and voltage sensors, respectively. The average differential inductance is then computed over the switching period.

In particular, when a phase is on and V.sub.pole =V.sub.1, ##EQU3## where R represents the stator resistance, .PSI. represents the stator flux linkage, and L represents the stator inductance. Equation (1) can be approximated by: ##EQU4## where ##EQU5## and .DELTA.t.sub.1 represents the time for which V.sub.pole =V.sub.1.

When the phase is off and V.sub.pole =V.sub.2, (during a time .DELTA.t.sub.2), ##EQU6## which can be approximated by: ##EQU7## where ##EQU8## Assuming that ##EQU9## (i.e., gap velocity remains constant throughout the PWM interval and I.sub.avg1 =I.sub.avg2), then subtracting equation (5) from equation (2) to eliminate the velocity and IR terms term gives from which the average differential inductance L.sub.avg over the switching cycle .DELTA.t.sub.1 +.DELTA.t.sub.2 can be calculated as ##EQU10##

The effective air gap length is then determined from the inherent magnetization characteristic (FIG. 4) using L.sub.avg and I.sub.avg, where ##EQU11## According to one embodiment, two-dimensional interpolation of magnetization curves stored in a microprocessor is performed. Alternatively, a functional model of the magnetic bearing flux path can be employed. Accuracy is dependent on the speed at which the measurements, interpolations, and calculations can be made since the approximations of equations (2) and (5) assume constant air gap and rate of change of air gap. Note that FIG. 4 yields unique answers for restricted ranges of current--principally in the unsaturated or linear magnetic operating range.

Advantageously, the effects of changing air gap (represented by the term ##EQU12## in equations (1) and (4)) and the resistive drop IR are avoided by the air gap estimation method of the present invention, such that inductance for each stator phase is isolated and additional sensors (e.g., flux sensors) are not required.

This technique for position estimation can be applied at either high or low current levels; however, it is not effective just beyond the onset of magnetic saturation, due to non-unique air gap estimates (See FIG. 4). Thus, the specific active magnetic bearing application must be organized to have an available actuation pole in an appropriate current range. This is generally possible in most systems, where a single actuator at a time is active in each axis. It is also possible in actively biased bearings where the two magnetic actuators per axis are controlled by decreasing current in one magnetic actuator while simultaneously increasing current in the second magnetic actuator. In such case, estimates from the actuator having the lower current is used. Given the air gap position, the air gap velocity and magnetic flux level can be determined at any bearing current level.

Given the value of the air gap position g, however determined, this inference method can be extended according to the present invention to determine both the air gap velocity and magnetic flux levels. These measurements are useful in feedback control of the magnetic bearing. The flux-linkage, which is directly related to actuator force, can be determined by two-dimensional interpolation of the magnetic characteristic of FIG. 2 (.PSI. as a function of g and i) stored in a microprocessor. The air gap velocity component can be determined by summing equations (2) and (5), looking up ##EQU13## from FIG. 5 ##EQU14## as a function of .PSI. and i), then computing ##EQU15##

One implementation of a controller for an active magnetic bearing according to the present invention is illustrated in FIG. 6. A digital signal processor (DSP) 60 provides desired PWM timing to a gating logic block 62. Gating logic block 62 provides gating signals, via optical couplers 64, to an inverter 66 for providing drive signals to an active magnetic bearing such as that of FIG. 1. DC bus voltage measurements are taken by a voltage sensor 68 and provided to a voltage feedback block 70. Current measurements are taken by sensors 72 and provided to current feedback blocks 74-75. The voltage and current measurements used in making the air gap position, velocity and flux measurements according to the present invention are provided via respective analog-to-digital (A/D) converters 76-80 to DSP 60.

Effectively, by the air gap estimation method of the present invention, each magnetic pole-pair is employed both for controlling air gap position and as an inductive feedback sensor, thereby eliminating the need for separate sensors. As a result, the parts count is reduced, and the robustness and reliability of the active magnetic bearing system are improved. The sensing process is non-intrusive in that it simply involves monitoring the electrical variables as the active magnetic bearing operates.

As a further advantage, all force actuation pole-pairs can be used to measure the rotor position, if desired. The result would be the equivalent of two position sensors per control axis, providing inherent redundancy because only one air gap measurement per control axis is generally required to locate the rotor shaft.

Application of the air gap estimation method of the present invention to radial magnetic bearings has been illustrated herein by way of example only. The air gap estimation method is applicable to other types of active magnetic bearings, such as thrust bearing applications. The technique is the same, involving isolation of the flux paths for the different electromagnetic force actuators, digital control of the pole current, determination of the incremental pole inductance by elimination of velocity and IR components, and determination of effective air gap length from the calculated incremental pole inductance.

While the preferred embodiments of the present invention have been shown and described herein, it will be obvious that such embodiments are provided by way of example only. Numerous variations, changes and substitutions will occur to those of skill in the art without departing from the invention herein. Accordingly, it is intended that the invention be limited only by the spirit and scope of the appended claims.