Position Observer for Electrical Machines

Abstract

A control system for an electrical motor comprises a rotor, a stator having a plurality of phase windings, and an inverter for applying voltage to the plurality of phase windings by connecting individual phase windings to a first or second voltage level. The control system is configured to measure a first rate of change of current in a first phase winding, of said plurality of phase windings, connected to the first voltage level, to measure a second rate of change of current in a second, different phase winding connected to the first voltage level, and to calculate a difference between the first and second rate of change of current. The control system is further configured to use the calculated difference to obtain data related to a position of the rotor.

Claims

1. A control system for an electrical motor comprising a rotor, a stator having a plurality of phase windings, and an inverter for applying voltage to the plurality of phase windings by connecting individual phase windings to a first or second voltage level, the control system configured: to measure a first rate of change of current in a first phase winding, of said plurality of phase windings, connected to the first voltage level; to measure a second rate of change of current in a second, different phase winding connected to the first voltage level; to calculate a difference between the first and second rate of change of current; and to use the calculated difference to obtain data related to a position of the rotor.

2. The control system of claim 1, wherein the control system is further configured to use said data related to the position of the rotor to determine the position of the rotor relative to the stator, and to use the determined position to control operation of the electrical motor.

3. The control system of claim 1, wherein the control system is configured to measure the first and second rates of change of current substantially simultaneously.

4. The control system of claim 1, comprising at least one sensor coil located in proximity to said first and second phase windings, wherein the control system is configured to measure a voltage in the sensor coil to determine the first and second rates of change of current.

5. The control system of claim 4, wherein the sensor coil is a Rogowski coil arranged such that a current in said first phase winding passes through the Rogowski coil in a first direction and a current in said second phase winding passes through the Rogowski coil in a second direction opposite to the first direction, in order to calculate the difference between the first and second rate of change of current.

6. The control system of claim 1, comprising: a first Rogowski coil arranged such that the current in the first phase winding passes through the first Rogowski coil; a second Rogowski coil arranged such that the current in the second phase winding passes through the second Rogowski coil; and a differential amplifier connected to the first and second Rogowski coils, in order to calculate the difference between the first and second rate of change of current.

7. The control system of claim 6, wherein the control system is configured to measure a third rate of change of current in a third phase winding of said plurality of phase windings, the control system further comprising: a third Rogowski coil arranged such that the current in the third phase winding, passes through the third Rogowski coil; and a second differential amplifier connected to the first and third Rogowski coils, in order to calculate a difference between the first and third rate of change of current, wherein the control system is further configured to use the difference between the first and third rate of change to obtain further data related to a position of the rotor.

8. The control system of claim 7, wherein the control system is configured to measure the third rate of change of current when the third phase winding is connected to the first voltage level.

9. The control system of claim 7, comprising a third differential amplifier connected to the second and third Rogowski coils, in order to calculate a difference between the second and third rate of change of current, wherein the control system is further configured to use the difference between the second and third rate of change to obtain further data related to a position of the rotor.

10. The control system of claim 6, wherein each differential amplifier is an operational amplifier.

11. The control system of claim 1, wherein the control system is further configured to: measure a third rate of change of current in the third phase winding and a fourth rate of change of current in the first phase winding when the first and third phase windings are connected to the same first or second voltage level; calculate a difference between the third and fourth rate of change of current; and use the calculated difference between the third and fourth rate of change of current to obtain second data related to a position of the rotor.

12. The control system of claim 11, wherein the control system is further configured to: measure a fifth rate of change of current in the second phase winding and a sixth rate of change of current in the third phase winding when the second and third phase windings are connected to the same first or second voltage level; calculate a difference between the fifth and sixth rate of change of current; and use the calculated difference between the fifth and sixth rate of change of current to obtain third data related to a position of the rotor.

13. The control system of claim 11, wherein the control system is further configured to use the data and the second data and/or the third data to determine a position of the rotor.

14. The control system of claim 13, wherein the control system is further configured to control the electrical motor by synchronizing the current in the plurality of phase windings based on the determined position.

15. A control system for an electrical motor comprising a rotor, a stator having a plurality of phase windings, and an inverter for applying voltage to the plurality of phase windings by connecting individual phase windings to a first or second voltage level, the control system configured: to measure a first rate of change of current in a first phase winding, of said plurality of phase windings, connected to the first voltage level; to measure a second rate of change of current in a second, different phase winding connected to the second voltage level; to measure a third rate of change of current in a third, different phase winding connected to the second voltage level; to calculate a difference between the first rate of change of current and the sum of the second and third rates of change of current; and to use the calculated difference to obtain data related to a position of the rotor.

16. A control system for an electrical motor comprising a rotor, a stator having a plurality of phase windings, and an inverter for applying voltage to the plurality of phase windings by connecting individual phase windings to a first or second voltage level, the control system configured to measure a rate of change of current in a positive voltage supply rail of the inverter or a zero voltage supply rail of the inverter, and to obtain data related to a position of the rotor from the measured rate of change of current.

17. A control system for an electrical motor comprising a rotor, a stator having a plurality of phase windings, and an inverter for applying varying voltages to the plurality of phase windings so as to cause rotation of the rotor, the control system configured: to operate the motor in a low speed regime; and while operating in the low speed regime: to measure the stator current in the phase windings of the stator arising from the application of stator voltage; to determine position of the rotor in relation to the stator; and to calculate a motor characteristic of the electrical motor from the positions of the rotor, the stator voltages and stator current, and a resistance of the phase windings.

18. The control system of claim 17, wherein the data related to the resistance of the phase windings comprises resistance as a function of stator current.

19. The control system of claim 17, wherein the motor characteristic of the motor is flux linkage as a function of stator current.

20. The control system of claim 19, configured to calculate the flux linkage by using an equation to multiply an imaginary part of the stator current and the resistance of the phase windings to get a result, subtract the result from an imaginary part of the stator voltage to get a difference, and divide the difference by a radial speed of the rotor.

21. The control system of claim 20, wherein the equation is $\lambda \left(I_{\mathrm{s\_real}}\right)=\frac{V_{\mathrm{s\_imag}}-I_{\mathrm{s\_imag}}{R}_s}{\omega },$ where λ(I.sub.s_real) is the flux linkage as a function of a real part of the stator current, V.sub.s_imag is the imaginary part of the stator voltage, I.sub.s_imag is the imaginary part of the stator current, R.sub.s is the resistance of the phase windings, and ω is the radial speed of the rotor.

22. The control system of claim 19, further configured to use the position of the rotor in calculating the flux linkage by locating a rotating reference frame for an electromotive force, emf, generated by the rotation of the rotor such that an imaginary part of the emf is equal to zero in that reference frame, and wherein the inverter output voltage and current are evaluated in that reference frame.

23. The control system of claim 17, further configured to extrapolate and/or interpolate the flux linkage to determine the flux linkage at values of stator current which have not been measured, for example when the motor is operating in a high speed regime.

24. The control system of claim 23, configured to extrapolate and/or interpolate the flux linkage using a non-linear saturating function.

25. The control system of claim 24, wherein the non-linear saturating function is a Hill equation.

26. The control system of claim 17, for determining the position of the rotor relative to the stator, being further configured: to measure a first rate of change of current in a first phase winding, of said plurality of phase windings, connected to a first voltage level; to measure a second rate of change of current in a second, different phase winding connected to the first voltage level; to calculate a difference between the first and second rate of change of current; and to use the calculated difference to obtain data related to a position of the rotor.

27. A control system for an electrical motor comprising a rotor and a stator having a plurality of phase windings and an inverter for applying voltage to the plurality of phase windings so as to cause rotation of the rotor; wherein the control system is configured to operate in a first control mode when the motor is in a low speed regime, and to operate in a second control mode when the motor is in a high speed regime.

28. The control system of claim 27, further is configured, when operating in the first control mode, to calculate a relationship between flux linkage and current of the stator, and, when operating in the second control mode, to use the relationship to estimate an electromotive force, emf, generated by the rotation of the rotor.

29. The control system of claim 27, wherein, when operating in the first control mode, the control system is configured to measure a variation in the rate of change of current to determine a position of the rotor in relation to the stator, and when operating in the second control mode, the control system is configured to estimate the emf in the stator to determine the position of the rotor.

30. The control system of claim 27, wherein the control system is configured to calculate further data related to flux linkage so as to update the flux linkage relationship when operating in the first control mode.

31. The control system of claim 30, wherein the control system is configured to compare the further data to previously calculated data related to the flux linkage relationship to detect changes in motor condition.

32. The control system of claim 27 wherein, when the motor transitions from the low speed regime to the high speed regime, or from the high speed regime to the low speed regime, the control system is configured to operate in both the first control mode and the second control mode and to generate a weighted control signal, being a weighted combination of control signals produced in the first and second control modes.

33. The control system of claim 27, wherein the motor is in the high speed regime when a speed of the motor is greater than 10% of a rated speed of the motor, and in the low speed regime otherwise.

34. The control system of claim 27 being further configured: to operate the motor in the low speed regime; and while operating in the low speed regime: to measure a stator current in phase windings of a stator arising from an application of a stator voltage; to determine position of the rotor in relation to the stator; and to calculate a motor characteristic of the electrical motor from the positions of the rotor, the stator voltage and the stator current, and a resistance of the phase windings.

35. An electrical motor comprising the control system of claim 1.

36. An electrical motor comprising the control system of claim 17.

37. An electrical motor comprising the control system of claim 27.

Description

[0017] In order that the invention may be more fully understood reference will now be made, by way of example, to the accompanying drawings, in which:

[0018] FIGS. 1, 2, 3 and 4 are sections through examples of machines to which the controller of the invention may be applied;

[0019] FIG. 5 shows an equivalent circuit of a machine according to the invention in the rotating reference frame;

[0020] FIG. 6 shows a vector diagram of a motor running with emf in phase with current;

[0021] FIG. 7 shows a vector diagram of a motor running in an overexcited way with current lagging the emf vector;

[0022] FIG. 8a shows a schematic diagram of an inverter with a three phase motor;

[0023] FIG. 8b shows a schematic diagram of the inverter in an inverter state V(1,0,0);

[0024] FIG. 9 shows a schematic diagram of another inverter with a three phase motor;

[0025] FIG. 10 shows a simplified diagram of the inverter in the V(1,0,0) inverter state;

[0026] FIG. 11 shows a schematic diagram of three Rogowski coils in a three phase motor;

[0027] FIG. 12 shows a non-linear flux linkage to current characteristic calculated from data collected according to the invention; and

[0028] FIG. 13 shows a flow chart of a decision process occurring in an electronic control system according to the invention.

[0029] FIG. 1 shows a permanent magnet synchronous motor with magnets mounted on the surface of the rotor. The stator 241 is made from laminated steel. In this example the stator has 9 teeth interspersed between the 9 slots 242. The stator winding would typically be wound with a short pitch with one coil wound around each tooth and then the three coils spaced at 120° to each other connected together to create each of three phase windings. The rotor has a steel core 243 surrounded by a layer of permanent magnet material, 244. In this example the permanent magnet layer 244 is magnetized radially to create a six pole magnetic field. Rotation of the rotor induces three emfs in the stator windings which are displaced in phase by 120° (electrical degrees). Operation as a motor requires an inverter to deliver three alternating armature currents in synchronism with the internal emf induced in each winding. The inductance of the phase windings of this motor does not change significantly with rotor position.

[0030] The hybrid stepping motor is a further example of a synchronous motor. Hybrid stepping motors usually have two, three or five phase windings. FIG. 2 illustrates a transverse cross-section through one end of a typical two-phase hybrid stepping motor.

[0031] The two-phase hybrid stepping motor of FIG. 2 consists of a stator 221 and a rotor 222, both of which are constructed from stacked laminations of electrical grade steel. The stator has eight poles 223, four associated with each phase, which widen at the tips to a group of teeth 224. Each stator pole 223 has a winding in the form of a short-pitched phase coil 225. These coils 225 are connected, usually in series, to form two electrically independent motor phase windings A and B. The rotor has a large number of radially spaced teeth 226 (typically 50) at each of its ends, and incorporates a permanent magnet which produces static flux along the axis of the machine. This makes the rotor teeth at one end of the machine all North poles and the rotor teeth at the other end of the rotor all South poles. The teeth 226 at one end of the rotor 222 are offset by half a tooth pitch with respect to the teeth 226 at the other end of the rotor 222. The rotor tooth pitch is the same as the tooth pitch of the stator 221. The emf induced in the two stator phase windings arises from the rotation of the rotor. The emf in phase A is 90° phase shifted from the emf in phase B. The inductance variation in the stator phase windings is relatively small because two of the 4 stator poles have the opposite alignment compared to the other two stator poles. Sensorless operation of the hybrid stepping motor is improved significantly by the use of this invention. The weak inductance signal means that sensorless operation with back emf estimation methods has to cover a greater range of operating speeds and an accurate non-linear motor model is vital to allow the hybrid stepping motor to operate down to speeds as low as 1 rps (60 rpm).

[0032] FIG. 3 shows a synchronous reluctance motor with stator 231 and rotor 233. The stator shown has 24 slots 232, though 12 and 36 stator slots are also common. The stator slots contain a distributed three phase winding. The rotor 233 is made from laminated steel and contains slotting to act as flux barriers. This creates axes of high and low reluctance such that if the stator winding creates a 4 pole magnetic field the rotor will align the flux barriers so that the stator flux follows a low reluctance path. As the stator field rotates the rotor will follow the field in synchronism. Many different rotor constructions can be used, the aim being to deliver the maximum ratio of reluctance variation. In this motor there is no rotating magnet to induce an emf in the stator windings; the torque is created by variations in the self inductances of the stator windings and variations in the mutual inductance between phase windings as the rotor rotates.

[0033] As an enhancement to the synchronous reluctance motor, permanent magnets can be inserted within the rotor slotting to increase the magnetic saliency further and increase torque production. Such a machine is known as a permanent magnet assisted synchronous reluctance motor. In a permanent magnet synchronous reluctance motor torque is produced by the stator current interacting with both magnet induced emf and emf induced from variable inductance.

[0034] FIG. 4 shows a permanent magnet motor in which the permanent magnets 253 are located within the steel lamination of the rotor. This form of rotor is referred to as interior permanent magnet (IPM) and provides permanent magnet flux linking the stator phase windings while the steel sections between the magnets 254 provide a path of variable inductance for the stator flux. Typically the inductance Ld of the direct axis in line with the magnet axis is lower than the quadrature inductance Lq in line with the axis between the magnets where there is a greater content of steel. The stator slots 252 in the stator 251 would typically contain a distributed three phase winding though it is also possible to have an interior permanent magnet motor with a stator with fewer slots and a concentrated coil winding.

[0035] All of these machine types have the common feature that they can be modelled as having an internally induced, rotational emf in the stator windings. The rotational emf will include components induced from the rotation of magnets in motors containing permanent magnets (FIG. 1, FIG. 2 and FIG. 4) and will contain additional emf components if there is a variation between the direct axis inductance Ld and the quadrature axis inductance Lq as the rotor rotates. In synchronous reluctance motors the variation in reluctance as the rotor rotates leads to variations in both the self inductance of each phase winding and variations in mutual inductance between windings. The term “rotational emf” as used in this invention, can therefore be any time varying combination of changing fluxes, due to a phase winding's own current (self inductance), the currents in other windings (mutual inductance) or a rotating permanent magnet. In certain interior permanent magnet motors the rotational emf will comprise components from variations in both permanent magnet flux and inductance as the rotor rotates.

[0036] In the mathematical analysis of electrical machines of these types it is common to analyze the machine totally in electrical cycles and electrical degrees. When this is done the analysis is similar for all synchronous machines of any type and pole numbers. In such analysis it is common to transform the statically orientated, time varying quantities of each stator phase winding into a single system of rotating vectors, rotating at the same frequency as the electrical frequency of the rotor.

[0037] Using the electrical quantities on a synchronously rotating reference frame containing components of all the phase windings of the motor an equivalent circuit is shown in FIG. 5. The stator voltage equation, in any synchronous rotating reference frame, is given by:

V.sub.s=E.sub.s+jI.sub.sX.sub.s+I.sub.sR.sub.s,  (1)

where V.sub.s, E.sub.s, I.sub.s are the vector quantities of the stator voltage, rotational emf and stator current with respect to the rotating reference frame; Xs is the reactance of the stator phase winding and Rs is the resistance of the stator phase windings. If there is a difference between the direct axis inductance Ld and the quadrature axis inductance Lq, Xs will be a complex vector with different in-phase and out-of-phase components. This is common in motors with significant variation in reluctance (and hence stator inductance) at different rotor orientations. An example of such a motor where reactance Xs could have different quadrature components would be an interior permanent magnet synchronous motor or a synchronous reluctance motor. However, since the rotating reference frame rotates with the rotor, the components of Xs do not vary significantly with time and the term jI.sub.sX.sub.s does not contribute to torque production. The rotational emf is the extended emf containing all emf terms contributing to motor torque production.

[0038] The torque produced by the motor is given by

[00001]$\begin{array}{cc}T=\frac{\stackrel{\_}{I_s}\star \stackrel{\_}{E_s}\cos \left(\gamma \right)}{\omega },& \left(2\right)\end{array}$

where
γ is the angle between the rotational emf vector and the stator current;
ω is the rotational velocity of the rotor in mechanical radians per second.

[0039] Operation of the motor with maximum torque per ampere is achieved if γ=0 and the E.sub.s and I.sub.s vectors are coincident. When, γ=0, the rotational emf vector E.sub.s is in phase with the stator current vector and the torque is again given by equation (2) which is now maximum and equal to

[00002]$\begin{array}{cc}T=\frac{\stackrel{\_}{I_s}\star \stackrel{\_}{E_s}}{\omega }.& \left(3\right)\end{array}$

[0040] The angular position of the rotational emf vector with respect to the stator current vector therefore determines the torque production of the motor. FIG. 6 shows a vector diagram showing the relationship between stator voltage vectors calculated on the rotating reference frame.

[0041] In the example illustrated by FIG. 6, the E.sub.s and I.sub.s vectors are coincident; γ=0, the rotational emf vector E.sub.s is in phase with the stator current vector and the torque is maximum for the given excitation current level and is given by equation (3).

[0042] In a further example illustrated by FIG. 7 the rotational emf vector, E.sub.s leads the I.sub.s vector by an angle γ. The motor is therefore producing less than maximum torque per ampere and the torque is now given by equation (2).

[0043] The primary purpose of a sensorless position estimation algorithm is to estimate the position of the rotor in real time while the rotor is rotating and while the load is changing. The secondary purpose is to adjust the operation of the motor to ensure that the torque production mechanism of the motor is optimized to its highest efficiency or to any other criteria determined by a particular application.

[0044] Since V.sub.s, E.sub.s, I.sub.s all have real and imaginary parts calculated on a rotating reference frame, the reference frame can be chosen on any particular axis appropriate for the mathematical analysis. In most prior art analysis the d-axis is chosen as the rotating reference frame. However, in a motor with emf components induced from both inductive and magnet fluxes it is convenient to use the axis of the sum of all rotational emf components as the estimated rotational emf vector position. Equation (1) can be rewritten into a real part (in phase with the estimated rotational emf vector) and an imaginary part (orthogonal to the estimated rotational emf vector):

V.sub.s_real=E.sub.s_real+I.sub.s_real*R.sub.s−I.sub.s_imag*X.sub.s  (4)

and,

V.sub.s_imag=E.sub.s_imag+I.sub.s_imag*R.sub.s+I.sub.s_real*X.sub.s  (5).

[0045] As described earlier the values of Xs in equations (4) and (5) may be a function of current and may also have components in both the real and imaginary complex planes.

[0046] These equations can be rearranged to give equations for the real and imaginary components of the rotational emf vector, E.sub.s,

E.sub.s_real=V.sub.s_real−I.sub.s_real*R.sub.s+I.sub.s_imag*X.sub.s  (6)

E.sub.s_imag=V.sub.s_imag−I.sub.s_imag*R.sub.s−I.sub.s_real*X.sub.s  (7)

[0047] If values for Rs and Xs can be determined including any non-linear variation in Xs with current and knowing the magnitude of the voltage vector delivered through the inverter, these equations can be evaluated in real time during rotation of the motor. The magnitude of the voltage vector can either be estimated from the commanded pulse width modulation duty cycles in the inverter or be monitored at the outputs of the inverter. The calculated values from equations (6) and (7) can be used to update the estimate of the rotating reference frame of the rotational emf vector such that the value of Es_imag calculated by equation (7) is kept close to zero.

[0048] The above equations are similar to those disclosed in Prior Art documents such as WO 2010055168. Furthermore, documents such as “Sensorless Control of Interior Permanent Magnet Synchronous Motor by Estimation of an Extended Electromotive Force” (Chen et al., IEEE, 2002) and “Sensorless Control Strategy for Salient-Pole PMSM Based on Extended EMF in Rotating Reference Frame) (Morimoto et al., IEEE Transactions on industry applications, vol. 38, No. 4, 2002) describe the combination of all the torque producing emf terms, (whether the emf comes from reluctance variations (inductance variations) or from permanent magnet flux variation) as an extended emf, Ex. Using the terminology in equation (6) and (7), the extended emf, Ex, can be written as:

E.sub.x=E.sub.s_real+j E.sub.s_imag.  (8)

[0049] In WO 2010055168 a method was described to use these equations and particularly equation (7) to adjust the operating point of a motor until Es_imag is converging on zero. At that point the torque producing or extended emf has no imaginary component and the rotating reference frame has been determined to track the position of the rotating emf.

[0050] However implementation of a sensorless control scheme as described in WO 2010055168 relies on a non-linear model of the inductance of the motor in each axis being determined and implemented over the whole operating range of a motor and generator to correctly determine the reactance term Xs. The reactance Xs will vary independently in the real and imaginary axes with non-linear functions of the current in those axes. There will also be cross-coupling effects since current in one axis will cause saturation in certain parts of the magnetic circuit which are shared by the other axes. In the paper “Sensorless Control of Interior Permanent Magnet Synchronous Motor by Estimation of an Extended Electromotive Force” (Chen et al., IEEE, 2002) and “Sensorless Control Strategy for Salient-Pole PMSM Based on Extended EMF in Rotating Reference Frame) (Morimoto et al., IEEE Transactions on industry applications, vol. 38, No. 4, 2002) the authors ignore the non-linear dependence of the inductance terms with current and refer to constant values for Ld and Lq.

[0051] It is possible to make approximations for the model of the reactance Xs and its components arising from Ld and Lq, or to model it off-line using computational calculations of the electromagnetic circuit. The following discussion illustrates the provision of a simple method to create this non-linear relationship within the motor controller even while the motor is rotating or when it is stationary.

[0052] In one embodiment, the calculation works as follows. Firstly at zero speed the resistance of the motor phases can be determined by applying an inverter voltage and measuring the dc current which flows. If this dc current vector is orientated in line with an axis of zero torque the rotor will be held in a stationary position during this measurement. If the voltage vector and the current measurements are mapped onto the rotating reference frame (with zero speed) the resulting DC equation can be solved to find resistance. Under dc conditions with no rotation of the vector the reactance terms are zero and the rotational emf terms are zero. It is advantageous to perform the resistance measurement over the whole range of inverter output current as non-linear voltage drops, such as the on-state voltage drops, in the inverter switches can be included in the resistance function as a function of the current flowing. The non-linear resistance function can be stored within memory inside the microprocessor or in an external non-volatile memory circuit.

[0053] Secondly at low speed the motor can be operated using inductance sensing methods (b) as described, for example, in publications such as “Sensorless Control of AC Machines at Low Speed and Standstill Based on the “INFORM” Method” (Schroedl, IEEE, 1996) and “Sensorless control of Surface Mounted Permanent Magnetic Machine Using the Standard Space Vector PWM” (Hua et al., IEEE, 2007). In addition, a further calculation may be implemented which provides a significant improvement in the position calculation over the prior art methods. The inverter can be placed in one of six active states referred to as V(1,0,0), V(0,1,0), V(0,0,1), V(0,1,1), V(1,0,1), V(1,1,0), such that only one inverter switching leg is connected to +V (State=1) or only one inverter switching leg is connected to 0V (State=0) while the other two switching legs have the opposite state. Two final non-active or zero vectors states are V(0,0,0) and V(1,1,1). These non-active states apply zero voltage across the motor windings. In each of the six active states there is always one inverter leg which has a unique state and two inverter legs which share a common state.

[0054] Prior art controllers measure the rate of change of current in the phase winding with the unique state. It has been discovered that additional information is available by using the rate of change of current data from the two phase windings which share a common state. This is illustrated in FIG. 8a, which shows an inverter with a three phase motor (i.e. an electrical motor having three phase windings). The inverter comprises a positive supply rail (+Vdc), a return path (0V) and six diodes connected in parallel with switches. Three inductors, representing the three phase windings, are connected to the inverter. A di/dt sensor (for example a Rogowski coil) is closely coupled to a conductor carrying a current of each phase winding to measure the rate of change of current (di/dt) in that phase winding.

[0055] A single measurement in the phase winding with the unique state can be enhanced by additional measurement of the rate of change of current in the other phase windings, as can be understood from FIG. 8b, which shows the inverter in the state V(1,0,0). The switching leg connected to phase U is uniquely in state (1) with the upper switch conducting current from the positive supply rail. The switching legs connected to phases V and W are in the shared state (0) with the lower switches conducting current back to the 0V rail.

[0056] In this state, V(1,0,0) the signal for di/dt(U)=−1*(di/dt(V)+di/dt(W))

[0057] Therefore di/dt(U) −1*(di/dt(V)+di/dt(W))=2*di/dt(U)

[0058] By using all three di/dt measurements and summing the signals in this way the signal of di/dt(U) is doubled and signal to noise ratio is improved. This implementation illustrates that the difference between a first rate of change of current and the sum of second and third rates of change of current can double the resolution of the reading of first rate of change of current.

[0059] In a further implementation only two di/dt sensors may be required since the di/dt in the third phase of the inverter can always be calculated knowing that the three phase currents and therefore the three phase di/dt signals add up to zero.

[0060] FIG. 9 shows an alternative sensor arrangement for measuring the rate of change of current. Instead of requiring a di/dt sensor in at least two phase outputs of the inverter, it is possible to use a single sensor 401 on the positive dc rail or a single sensor 402 in the 0V rail of the inverter. In each of the active inverter states the di/dt of the current measured by a di/dt sensor in either the +Vdc supply rail or the 0V supply rail will be the same as the current through the phase which has a unique conducting path and will be the sum of the current in the phases sharing the common conducting state.

[0061] In the above arrangements, it will be appreciated that at least two temporally spaced measurements will be required to determine any change in di/dt. It has now been discovered that an additional position dependent signal can be measured by subtracting the rate of change of current in the phase windings which have the common switching state (i.e. the phase windings that are connected to the same voltage level). It has been found that measuring the split in the rate of change of current in the two parallel connected common phases provides instantaneous position dependent data without having to use two successive readings in two inverter voltage states. This can be understood with reference to FIG. 10, which shows a simplified diagram of the inverter in state V(1,0,0). In this state the signals di/dt(V) and di/dt(W) will not necessarily be equal. The split in current and the different values of the di/dt between Phase V and Phase W is position dependent and is therefore another measure of the difference in inductance between phases V and W. One of the di/dt signals from the two phases connected to a common supply rail can be subtracted from the other. Using this method some mutual constituent parts of the signals not containing position data, such as amplifier offsets, are automatically cancelled. It is therefore possible to deduce position without needing to wait for an alternative state to isolate the position dependent data. The improvement offered by this approach allows position dependent data to be calculated from an effectively simultaneous set of di/dt measurements, taken close together during a single operating state of the inverter. This approach provides additional position dependent data not accessed by prior art implementations.

[0062] The subtraction of the two di/dt signals from phase windings sharing a common state can be implemented in a microprocessor or other control circuit after the two di/dt signals from the phase windings attached to the common state are converted to digital signals in an analog to digital converter.

[0063] Alternatively, to avoid some numerical errors and loss in accuracy which can arise in the numerical subtraction of the two digital signals, the subtraction of the rate of change of two phase currents can be achieved in a single Rogowski coil by passing two phase currents through the coil in opposite directions. The subtraction is therefore achieved by cancellation of the magnetic fields and the difference signal is created by the Rogowski coil.

[0064] FIG. 11 shows a further possible connection method for three Rogowski coils, 551, 552 and 553 used to measure the rate of change of currents in a three phase motor. Rogowski coil 551 is placed in close proximity to a conductor carrying the current, Iu, to motor phase U. The voltage induced across the terminals of the Rogowski coil 551 is

[00003]$v_{551}=M\frac{d\mathrm{Iu}}{dt},$

where M is the mutual inductance between the conductor and the Rogowski coil.

[0065] In a similar way Rogowski coils 552 and 553 create voltages v552 and v553 which are proportional to the rate of change of current in conductors carrying the current to motor phases V and W.

[0066] As explained above a significant advantage can be obtained by measuring the difference between the rate of change of current in two phases, particularly in any of the inverter states where two phases are connected to the same or common supply rail. FIG. 11 illustrates how this can be achieved with minimum error. One end of each of the Rogowski coils 551, 552, 553 is connected to a central node 550. This central node acts like the star point of a star connected three phase system. In this case it is the star point of the three Rogowski coils measuring the di/dt in the three phase windings. Since the sum of three phase currents adds up to zero, the sum of the differential of three phase currents also adds up to zero. The central node of 551, 552 and 553 will therefore have a relatively stable voltage. For ease of measurement it may also be useful to connect an optional voltage 555 to this central node or star point which shifts all the signals near to the centre of the measurement voltage range.

[0067] Using this arrangement, it is now very straightforward to measure the difference between any two rates of change of phase current. An operational amplifier 511 can be arranged to have the signal from Rogowski coil 551 connected to its positive input and the signal from Rogowski coil 552 connected to its negative input. The voltage across the differential input pins of the operational amplifier 511 is therefore

[00004]$M\frac{d\mathrm{Iu}}{dt}-M\frac{d\mathrm{Iv}}{dt}.$

[0068] The value of M is dependent on the coupling between the conductor carrying the phase current and the Rogowski coil. M also depends on factors such as the number of turns of the Rogowski coil and its cross-sectional area. To obtain consistent measurements it would be expected that the construction and positioning of multiple Rogowski coils in a multi-phase motor would be controlled within reasonable manufacturing tolerances. If this is not the case different values of M can be assigned to each measurement to allow variations in M to be included in the calculation.

[0069] Using well known methods of feedback the operational amplifier gain, k, will act on the difference between the two signals and create an amplified signal which is directly related to the difference between the rate of change of currents in the two conductors. This amplified signal would be

[00005]$kM\left(\frac{d\mathrm{Iu}}{dt}-\frac{d\mathrm{Iv}}{dt}\right).$

[0070] This method offers lower noise and distortion than separately amplifying each signal and then making the subtraction after the amplifier. In such a case there can be slight differences in the amplifier gains, k1 and k2 and the output of the subtraction would be

[00006]$M\left(k_1\frac{d\mathrm{Iu}}{dt}-k_2\frac{d\mathrm{Iv}}{dt}\right).$

[0071] Amplifier 512 can be used to measure and amplify the difference between the outputs of Rogowski coils 552 and 553 to give

[00007]$kM\left(\frac{d\mathrm{Iv}}{dt}-\frac{d\mathrm{Iw}}{dt}\right).$

[0072] Amplifier 513 can be used to measure and amplify the difference between the outputs of Rogowski coils 553 and 551 to give

[00008]$kM\left(\frac{d\mathrm{Iw}}{dt}-\frac{d\mathrm{Iu}}{dt}\right).$

[0073] In this way the star connection of the three Rogowski coils provides a means to create a signal containing small differences between the rates of change of current in any two motor phase windings and to then amplify that small difference without introducing errors caused by mis-matched gains in two separate amplifiers. In this way the circuit can detect very small differences in the rates of change of currents in two phase windings. Using this approach it has been possible to estimate the position in a permanent magnet synchronous motor with very small variations in inductance with position of the rotor.

[0074] The embodiment shown in FIG. 11 has operational amplifiers 511, 512 and 513, but other types of differential amplifiers can be used instead.

[0075] The three output signals from amplifiers 511, 512 and 513 can be sampled by a microprocessor and converted to digital data signals. The sampling should coincide with inverter states when the two di/dt signals at the input to the chosen amplifier are taken from the rates of change of current in phase windings sharing a common state within the inverter.

[0076] Irrespective of the method used to subtract the di/dt signals the result will be three new signals, where K is a constant containing all the multiplication factors in the signal chain.

[00009]$\mathrm{Signal}A=K\left(\frac{d\mathrm{Iu}}{dt}-\frac{d\mathrm{Iv}}{dt}\right).\mathrm{Signal}B=K\left(\frac{d\mathrm{Iv}}{dt}-\frac{d\mathrm{Iw}}{dt}\right).\mathrm{Signal}C=K\left(\frac{d\mathrm{Iw}}{dt}-\frac{d\mathrm{Iu}}{dt}\right).$

[0077] Calculation of position from these signals now proceeds with vector mathematics to convert to two phase signals, Alpha and Beta

[00010]$\begin{array}{c}\mathrm{Alpha}=\mathrm{Signal}A-\left(\frac{\mathrm{Signal}B+\mathrm{Signal}C}{2}\right)\\ \mathrm{Beta}=\sqrt{3}\left(\frac{\mathrm{Signal}B-\mathrm{Signal}C}{2}\right)\end{array}$

[0078] The rotor angle is then calculated by the Arc Tangent of Beta/Alpha.

[0079] The rotor angle calculated in this way will be have two cycles per electrical cycle of the motor. A number of known methods can use this signal to create a single angle which has one angular rotation cycle for each whole electrical cycle of the motor.

[0080] In a synchronous reluctance motor with no magnets the two repeating cycles of inductance which make up each electrical cycle are identical and will produce the same torque independent of the direction of current. In a permanent magnet motor the two cycles are associated with the opposite magnetic polarity and must be distinguished to ensure the direction of the torque is correct.

[0081] When the motor is first excited known pulsing methods can be used to determine the polarity of the magnet and start the angle estimation with the correct initial half of the cycle.

[0082] In a three phase motor signals Signal A, Signal B and Signal C are also balanced three phase signals and two of them can be used to estimate the third. In such a case it is also possible to use the three Signals to cross-check each other. The coil used to create one of the di/dt measurements may have a higher current than the other two and the coil used to calculate that di/dt may itself become saturated and its signal may be lower than expected. In these conditions the three di/dt signals may no longer add up to zero and a correction algorithm can be implemented so that the robustness of the position estimation is not compromised.

[0083] As only two phase windings will share a common state in a given inverter state, the signals Signal A, Signal B and Signal C will be calculated at different times in the inverter switching sequence. However, it is always possible to configure the inverter to include the required common states within its normal operating sequence and to ensure that the signals Signal A, Signal B and Signal C do not occur too far apart in time so that they can be combined together in a single calculation of the angle of the rotor.

[0084] It will be understood that a combination of at least two of the three signals (which can be considered as first, second and third data related to position of the rotor) from three difference measurements can be used to derive a continuous estimate of the rotor position which can be used to synchronize the current in the motor phase windings to always produce maximum torque for a given current in a chosen direction even at very low speeds or zero speeds when EMF methods do not work.

[0085] Embodiments described herein, using the difference of rate of change of current (i.e. signals A, B and/or C), can have the significant advantage of being directly applicable to many different electrical motors, without requiring prior knowledge of those motors. Known methods, on the other hand, require various motor parameters as inputs in order to provide accurate position readings when operating at low speeds.

[0086] As a result of these improvements to rate of change of current measurements leading to more accurate inductive sensing methods for sensorless control of synchronous motors, the position of the rotor can be determined quite accurately and robustly. It is possible to run the motor accurately with inductive sensing methods and this can be used to make measurements and calculations of the flux linkage to current relationship of the motor phase windings at known rotor positions. Flux linkage is the product of flux and coil turns and is always equal to the integral of terminal voltage less any internal voltage drops due to resistance. In a linear inductance case it is the product of inductance and current but since inductance changes as the magnetic circuit becomes saturated it is more useful to use the concept of flux linkage as a function of current to represent the non-linearity of a magnetic circuit. Such data is required for sensorless control methods based on rotational emf and extended emf estimation. Prior art methods based on rotational emf estimation have had to rely on pre-calculated flux linkage to current relationships which contain inaccuracies or approximations or were often assumed to be linear with no change in inductance from low current to high current. With the improvements above, whilst the motor is rotating under the control of the inductive di/dt sensing methods, the current vector can be accurately positioned to a position of maximum torque and the position of the rotating reference frame will be locked to and will track the position of the extended emf (even if the motor may be rotating so slowly that the rotational emf could not otherwise be detected) such that Es_imag can be assumed to be zero. Under these conditions equation (5) simplifies to

I.sub.s_realX.sub.s=V.sub.s_imag−I.sub.s_imagR.sub.s  (9)

[0087] The left hand side of equation (9) is known to be non-linear and is the product of flux linkage and rotational velocity (rather than the product of current, inductance and velocity in the case of a constant inductance). The flux linkage as a function of current can therefore be calculated from measured values by dividing equation (9) by angular velocity and using equation (10)

[00011]$\begin{array}{cc}\lambda \left(I_{\mathrm{s\_real}}\right)=\frac{V_{\mathrm{s\_imag}}-I_{\mathrm{s\_imag}}{R^s}_{{}^{}}}{\omega }.& \left(10\right)\end{array}$

[0088] Where Vs_imag and Is_imag are the components of the applied inverter voltage and the inverter current on the imaginary axis, orthogonal to the real axis, of the rotating reference frame, Rs is the resistance which may be taken from the stored data obtained in the initial step and ω is the angular velocity of the rotating reference frame at the time of the measurements.

[0089] Whilst the motor is running under the control of the di/dt signals, aligning the stator current vector to the rotor reference frame, the method performs repetitive calculations according to equation (10) at different currents and speeds building up the non-linear data set relating flux linkage to current with the rotor running at low speeds. Once a few data points have been collected, the controller can use the data directly with linear interpolation between points but the quality of the control is dramatically improved and extended by using the data set to fit a non-linear curve through the data points. Since the characteristics of a magnetic circuit are known to be that of a saturating function it is possible to fit the data to a non-linear saturating equation such as the Hill equation (https://en.wikipedia.org/wiki/Hill_equation_(biochemistry)) which is a non-linear saturating function used in medical applications. One useful form of the Hill equation is

[00012]$\begin{array}{cc}\lambda \left(I_{\mathrm{s\_real}}\right)=\frac{a{I}_{\mathrm{s\_real}}^n}{b+I_{\mathrm{s\_real}}^n},& \left(11\right)\end{array}$

[0090] where a, b and n are constants to be determined to fit the calculated data from at least three measurement points. The constant n is usually between 0.5 and 1.5 and determines shape of the saturation curve. If n is greater than 1, the saturation follows a sigmoid shape where the gradient starts low at low current, then increases and then drops again at high current. When n<=1 the gradient starts steeply and reduces as the magnetic material starts to saturate. An example of a fitted Hill equation is shown in FIG. 12. Since the Hill equation naturally follows a saturating magnetic characteristic only a few data points are needed to create the whole characteristic. This effectively enables the motor to be calibrated for sensorless control using EMF estimation, by first running at low speeds under the synchronous and closed loop control of di/dt sensing methods while making measurements at a few current levels and fitting the flux linkage data points to a saturating non-linear function which provides data over the whole torque range of the motor, thus removing the need for complex calculations to determine the magnetic characteristic.

[0091] Once the non-linear magnetic characteristic of the motor has been identified the controller can proceed to use Equation (7) within the real time motor control. In each control loop execution the equation can be calculated on the estimated rotor reference frame to create a new value of

E.sub.s.sub.imag=V.sub.s_imag−I.sub.s_imagR.sub.s−ωλ(I.sub.s_real).  (12)

[0092] Any non-zero result of this calculation is an error in the estimated position of the rotor reference frame and of the rotor emf axis. If the error is positive the error is added to the estimated velocity through a suitable controller such as proportional and integral to create a new speed estimate and to predict the position of the rotor reference frame at the next time step. In this way the controller can track the position of the rotational emf axis by continuously ensuring that errors in the imaginary rotational emf are driven to zero.

[0093] The accuracy of the estimation using equation (12) is dramatically improved by the use of the non-linear flux linkage function determined within the inverter while the motor is rotating as this automatically includes all the scaling gains and measurement errors of a real system. As has been described the equation 12 works best at higher motor speeds where the rotational emf terms are larger. If the motor returns to low speed the controller can gradually return to closed loop synchronous control using instantaneous rate of change of current measurements. The transition between the two methods can be instantaneous or be gradual using a graduated weighting function which may be based on speed or based on speed and load.

[0094] Additionally when the motor returns to the instantaneous rate of change of current measurements at low speed the measurement of the flux linkage to current characteristics can be restarted if required, to add further detail to the non-linear function. In other words, re-calibration can take place every time the motor returns to low speeds (or once a week, once a month etc.) This approach can therefore be used on a continual basis or just used during commissioning.

[0095] An additional calculation can be performed in the real axis according to equation (13) since while under the control of the instantaneous rate of change of current methods the imaginary current is zero and equation (6) simplifies to (13)

E.sub.s.sub.real=V.sub.s_real−I.sub.s_realR.sub.s.  (13)

[0096] This equation (13) allows the value of the rotational emf to be calculated from the inverter measurements and after dividing by speed its value can also be stored as a function of current in the real axis. This is a record of the permanent magnet flux or torque producing flux in the real axis.

[0097] The data stored under these approaches can also be used as a condition monitoring feature in a motor control. Once stable imaginary axis flux linkage characteristic and/or real axis flux linkage characteristic have been stored, the motor can be monitored over its life for changes in those characteristics while the motor is running at low speeds when the calculation of the flux linkage characteristics are available. Any change in this characteristic could be a sign of deterioration in winding insulation or damage to one of the coils in the motor or demagnetisation of the permanent magnet. Such an approach allows early detection of demagnetisation of the magnets to be detected. The controller could be configured to limit the maximum current in the motor to avoid complete demagnetisation until the magnet condition is checked.

[0098] FIG. 13 shows a flow diagram illustrating the steps of a method performed by a control system according to an embodiment. The method comprises the following steps: Measure resistance at zero or very low speeds or import resistance data (S1). Run motor using measurements of rate of change of current to calculate position and correctly orientate the stator current with respect to the rotor (S2). While the motor is running with the correct orientation of stator current with respect to the rotor at low to medium speeds use equation 10 to calculate points on the non-linear flux linkage to current characteristic (S3), and optionally use a Hill equation to create a full non-linear flux linkage characteristic (S4). Run the motor at higher speeds using the non-linear flux linkage characteristic obtained at lower speeds to estimate and update the position of the rotating reference frame to track the position of the emf vector (S5).

[0099] Implementation of this invention has been described with reference to electric motors with rotors rotating relative to a stator. The invention can be applied directly to linear electric motors to control linear travel velocities under conditions of changing loads. Such motors still have electrical cycles which can be modelled as systems of rotating vectors.

[0100] The control schemes according to this invention can also be used to measure the non-linear magnetic characteristics of a generator and then control the generator using the extended emf method with much greater accuracy that would be possible with off-line calculations of the magnetic characteristics.

[0101] The Figures have illustrated electric motors with the phase windings connected in star configuration. All the described approaches will work in an identical manner if the motor is connected in delta.

[0102] In any of the techniques described relying on the measurement of rate of change of current, the techniques can be implemented using any method of creating a rate of change of current. Examples could include the analogue differentiation of a measurement of the current; the measurement of two instantaneous current readings a short time apart and then numerically calculating the rate of change of current.

[0103] FIG. 11 shows how three Rogowkski coils can be connected to three operational amplifiers to calculate the difference and amplify the result to create Signal A, Signal B and Signal C. As Signal A, Signal B and Signal C can be measured in specific inverter states, a multiplexer can be used to select the inputs to one or more amplifiers to perform the required subtraction and amplification at each known inverter state. In this way the number of operational amplifiers and number of analog to digital converters can be reduced. This also allows the new difference calculations to be supplemented with measurements of di/dt in a single phase winding.

[0104] In safety critical applications it is often highly desirable to have redundancy of rotor position detection. The methods described can be used to supplement a rotor position sensor such that a low cost sensor can be enhanced to give high resolution and robustness. That is, the motor control system described herein may comprise one or more sensors to provide additional rotor position readings.