Method for rotor current estimation

Abstract

The present invention relates to a method of rotor current estimation of an electric machine (1), comprising: - a determination step (131) in which a current estimate in one PWM half-period is calculated, - an identification step (132) in which in a corresponding PWM half-period of a voltage waveform (20), a first sampling center location (134) and a second sampling center location (136) are identified at zero-voltage at a PWM / 2 interval; - a sampling step (133) in which current values (140, 142) are taken in the sampling center locations (134, 136); - a correction step (139) in which the current estimate is corrected based on the current values (140, 142); - an estimation step (144) in which a current value for the rotating part is estimated, such current value estimation taking into account a transformer transformation ratio.

Description

[0001] DESCRIPTION

[0002] Title: Method for rotor current estimation

[0003] The present invention relates to the fields of electronics and electrical engineering, and more particularly to the field of electric machines.

[0004] Electric machines may be used in electric and hybrid vehicles in order to provide torque to the wheels of the vehicle. The motor can for instance be an electrically excited synchronous machine or EESM. The EESM is an alternative to permanent magnet synchronous machines as the latter require the usage of rare-earth materials.

[0005] In an EESM, energy must be transferred from a non-rotating part, i.e. the stator of the electric machine, to a rotating part, i.e. the rotor. Such power transfer can either be conductive or inductive.

[0006] In the case of a conductive power transfer, electric contact is achieved by pushing brushes of the non-rotating part against slip rings of the rotating part. While the slip ring system of the conductive power transfer allows a direct measurement of the rotor current, which is required for current and torque control, it however causes both mechanical and electric losses and it requires regular maintenance.

[0007] In the case of an inductive power transfer, the power transfer is contactless and uses a rotating high-frequency transformer, which is fed by a phase shift full bridge converter or PSFBC. In this case, a rectifier is connected between the transformer and the rotor winding of the electric machine. There is no mechanical loss in an inductive power transfer and the process is maintenance-free. However, it is highly undesired to place a current sensor on the rotating part for complexity and cost reasons, as this would require a sensor power supply on the rotating part of the machine and a contactless signal transfer to the non-rotating part. As a result, direct measurement of the rotor current is, with today’s existing technology, not possible at acceptable cost.

[0008] In order to accurately determine the rotor current, which is required for efficient operation of the EESM, sensors can be used on the non-rotating part of the inductive power transfer system and a predictive model is provided to estimate the rotor current. In other words, the sensors are positioned on a primary side on the non-rotating part to further estimate what current provided on a secondary side, i.e. on the rotating part. It is for instance known to sample the primary side transformer current at a time corresponding to the middle of a positive pulse of the transformer voltage, and then apply a transformer transformation ratio. It is also known to use the converter input current, input voltage, duty cycle, and then apply a transformer transformation ratio.

[0009] However, while traditional methods relying on sensors positioned on the primary side of the non-rotating part usually work in steady state conditions, they might be inaccurate in dynamic or transient conditions.

[0010] The present invention fits into this context by providing a method capable of accurately estimating the output current on the rotating part of the electric machine both in steady state conditions and in transient conditions, such method relying on data acquired from the non-rotating part.

[0011] In this context, the present invention is directed to a method of rotor current estimation for a rotating part of an electric or hybrid vehicle electric machine, in which the electric machine comprises a converter which is operated with phase shifts, with a duty cycle ratio associated with a phase shift, the converter providing voltage to a transformer of the electric machine, such voltage being provided to a primary side of the transformer.

[0012] According to the invention, the current estimation method comprises a determination step in which a current estimate in one pulse width modulation (PWM) half-period is calculated; an identification step in which in a corresponding PWM half-period of a voltage waveform generated by the converter, a first sampling center location and a second sampling center location are identified at zero-voltage at a PWM / 2 interval; a sampling step in which a first current value and a second current value corresponding respectively to the first sampling center location and the second sampling center location are taken in a primary transformer current waveform; a correction step in which the current estimate calculated in the determination step is corrected based on the current values taken during the sampling step; and an estimation step in which a current value for the rotating part is estimated according to the corrected calculated current estimate from the correction step, such current value estimation taking into account a transformer transformation ratio.

[0013] The current estimation method according to the invention is destined to be used to determine the output current of a rotating part of an electric machine, for instance a rotor of a motor of an electric or hybrid vehicle. The electric machine is more precisely an inductive electrically excited synchronous machine or iEESM. It comprises, among other components, a DC / AC converter, a transformer and a rectifier.

[0014] The converter is a phase shift full bridge converter or PSFBC, which is operated with phase shifts with which a duty cycle ratio is associated. The converter provides bipolar rectangular voltages which are applied to the transformer, more precisely to a primary side of this transformer. This primary side of the transformer is on a non-rotating part of the electric machine, whereas a secondary side of the transformer is on the rotating part of the electric machine.

[0015] In order to estimate the rotor current, the current estimation method comprises a plurality of steps which are performed consecutively, among which a determination step, an identification step, a sampling step, a correction step and an estimation step. The determination step corresponds to a calculation of a current estimate via a power balance approach which is accurate in steady state conditions.

[0016] Using a calculation of a current estimate during a determination step rather than a direct measurement of primary transformer current is advantageous in that it is more robust in terms of timing errors.

[0017] During the identification step, sampling center locations for the primary transformer current acquisition are identified in each pulse width modulation (PWM) half-period based on the voltage waveform generated by the converter, more precisely at zerovoltage points within the voltage waveform. In the sampling step the primary transformer current corresponding to the sampling center locations from the identification step is sampled.

[0018] During this sampling step, at least a first sample corresponding to the first sampling center location and a second sample corresponding to the second center location are taken.

[0019] In the correction step, the primary transformer current values corresponding to the same PWM half-period as in the determination step are used to correct the calculated primary transformer current estimate from the determination step. In other words, the primary transformer current estimate of the determination step is corrected based on current values taken during the sampling step, these current values themselves corresponding to the sampling center locations identified during the identification step. The corrected current estimate thus obtained is then used in the estimation step to determine the rotor current.

[0020] When all the steps of the current estimation method are completed, the current value for the rotor is known and such current value can be used to control torque provided to the wheels of the electric or hybrid vehicle. Unlike other methods using a primary transformer current, in which parasitic elements may induce detrimental oscillatory behavior, the present method is accurate in both steady state conditions and dynamic conditions.

[0021] According to an optional characteristic of the invention, during the determination step two consecutive primary transformer current values corresponding respectively to a positive voltage pulse and a negative voltage pulse in the voltage waveform are taken, and the correction step comprises a substep wherein corrected consecutive current values which are calculated based on the current values taken during the sampling step and an additional averaging step in which the absolute values of the corrected current values are averaged.

[0022] As a result, the correction of the current estimate which has been previously calculated either relies directly on the current values corresponding to the sampling center locations at zero-voltage, or it relies on a further calculation based on consecutive current values which correspond to positive and negative pulses. During this further calculation, the current values corresponding to the sampling center locations at zerovoltage are used to determine the consecutive current values.

[0023] According to an optional characteristic of the invention, the current values taken during the sampling step correspond to a single value.

[0024] According to an optional characteristic of the invention, the current values taken during the sampling step correspond to a plurality of values.

[0025] Using a single value for the current value can notably correspond to a single- shot sampling method, whereas using a plurality of values to an oversampling method. The single-shot sampling method is easier to implement while the oversampling method improves accuracy as well as signal quality.

[0026] According to an optional characteristic of the invention, during the identification step the first sampling center location is identified before a given positive or negative pulse and the second sampling center location is identified after the given pulse. In other words, the sampling center locations are chosen so that they surround a given pulse of the voltage waveform.

[0027] According to an optional characteristic of the invention, the first sampling center location and the second sampling center location are identified at the start and at the end of the PWM half-period.

[0028] Such choice of the sampling center locations at the start and at the end of the PWM half-period is carried out in case of a center-aligned pulse generation. Alternatively, the the sampling center locations can be shifted from the start and the end of a given PWM half-period, as long as there is a distance of one PWM half-period between the first sampling center location and the second sampling center location which are both identified in a zero-voltage phase.

[0029] According to an optional characteristic of the invention, during the determination step, the current estimate corresponding to the PWM half-period is calculated from an input power of the converter, the duty cycle ratio and an input voltage of the converter.

[0030] The input power can be calculated from sensor values, for instance the converter input current and the converter input voltage.

[0031] According to an optional characteristic of the invention, the input power of the converter is obtained from at least an input current of the converter, an average value of the input current of one PWM half-period corresponding to the duty cycle of said PWM half-period.

[0032] According to an optional characteristic of the invention, the transformer transformation ratio is a relation between inductance and mutual inductance of the transformer.

[0033] In addition, it is possible to use a variable corresponding to an efficiency parameter which is implemented for example by a look-up table in dependency of the relevant quantities. Adding such variable improves the accuracy of the current estimation method. Alternatively, the transformer ratio is an effective transformation ratio. Such effective transformation ratio is necessary if inductance values are not constant but rather current dependent, for instance due to magnetic saturation. According to an optional characteristic of the invention, the method of rotor current estimation is combined with a machine model by embedding the machine model in an observer structure.

[0034] The combination of the current estimation method with a machine model improves the accuracy of the rotor current estimation, particularly so in dynamic conditions. The machine model can for instance help reduce deviations generally occurring in dynamic conditions. The machine model can be driven by an output voltage estimate of the inductive power transmission system and by a stator axis current derivative. The output voltage estimate of the inductive power transmission system corresponds to the voltage which is supplied to the rotor winding of the electric machine. The current derivative of the stator d-axis current can be used to further improve the estimation in dynamic conditions.

[0035] The observer structure is for instance a Luenberger observer, an extended Luen- berger observer, or a Kalman filter.

[0036] According to an optional characteristic of the invention, a corrective value for the observer structure is calculated from an observer error signal based on the rotor current estimate from the estimation step.

[0037] In other words, the observer error signal may be calculated based on the results of the estimation step. Other methods for rotor current calculation could also be used for calculating the observer error signal. When the observer error signal is calculated using a combination of multiple methods or embodiments, weighting factors may be implemented.

[0038] According to an optional characteristic of the invention, the corrective value is set to zero when the observer error signal is unreliable.

[0039] For instance, a near zero duty cycle ratio would render the observer error signal unreliable, as it is then impossible to measure a primary transformer current which really corresponds to the rotor current.

[0040] Other characteristics, details and advantages of the invention will become clearer on reading the following description, on the one hand, and several examples of realisation given as an indication and without limitation with reference to the schematic drawings annexed, on the other hand, on which: [Fig. 1] is a schematic representation of an electric machine for an electric or hybrid vehicle, the electric machine comprising a rotor, a stator and a transmitter comprising a rectifier and a transformer;

[0041] [Fig. 2] is another schematic representation of part of the electric machine of figure 1, further comprising a converter;

[0042] [Fig. 3] is a schematic representation of a first embodiment of a current estimation method, such method being used to determine a current value for the rotor of figure 1;

[0043] [Fig. 4] is a schematic representation of a second embodiment of the current estimation method used to determine the current value for the rotor of figure 1;

[0044] [Fig. 5] is a schematic representation of a third embodiment of the current estimation method used to determine the current value for the rotor of figure 1 ;

[0045] [Fig. 6] is a schematic representation of a current estimation method according to the invention, such method being used to determine the current value for the rotor of figure 1;

[0046] [Fig. 7] is a schematic representation of a combination of the current estimation method of figures 3 to 5 and the current estimation method according to the invention of figure 6 with a machine model.

[0047] The characteristics, variants and different modes of realization of the invention may be associated with each other in various combinations, in so far as they are not incompatible or exclusive with each other. In particular, variants of the invention comprising only a selection of features subsequently described in from the other features described may be imagined, if this selection of features is enough to confer a technical advantage and / or to differentiate the invention from prior art.

[0048] Like numbers refer to like elements throughout drawings.

[0049] Figures 1 and 2 are schematic representations of an electric machine 1, or part of said electric machine 1. The electric machine 1 is destined to be mounted in a vehicle such as an electric or hybrid vehicle, where it can be used to provide torque to the wheels of the vehicle. Here, the electric machine 1 is an inductive electrically excited synchronous machine, which is also known as its acronym iEESM. As can be seen on Figure 1, the electric machine 1 comprises a rotor 2 and a stator 4. In this example, the stator 4 is positioned around the rotor 2, but in other embodiments the rotor 2 could be positioned around the stator 4. In inductive electrically excited synchronous machines such as this electric machine 1, power is transferred to a winding mounted on the rotor 2 using a transformer 6. The transformer 6 is more precisely a rotating high frequency transformer. The transformer 6 comprises a primary side 8 and a secondary side 10, these two sides 8, 10 being different in that the primary side 8 is on a non-rotating part of the electric machine 1, i.e the stator, whereas the secondary side 10 is on a rotating part of said electric machine 1, i.e the rotor. On figure 1, a separation between the non-rotating part and the rotating part is illustrated as a dashed line.

[0050] A rectifier 12 is associated with the transformer 6. The rectifier 12 is a rotating rectifier, and as such it is positioned on the secondary side 10 of the transformer 6, i.e. on the rotating part of the electric machine 1. The role of the rectifier 12 is to provide only positive voltage to the rotor 2. This rectifier 12 is, along with the transformer 6, implemented in a transmitter 14, here an inductive transmitter.

[0051] As is visible on Figure 2, a converter 16 is connected to the primary side 8 of the transformer 6. The converter 16 is a phase shift full bridge converter or PSFBC. It works at a variable pulse width modulation frequency, or PWM frequency. The PWM frequency of the converter 16 can indeed vary in a wide range, for instance from 5 to 100 kHz, depending on the design of the converter 16, the transformer 6 and the required current in the rotor 2.

[0052] The converter 16 comprise an input capacitor 17. The converter 16 is controlled by an electronic control unit of the electric machine such as a microcontroller, a field- programmable gate array (FPGA) or a digital signal processor (DSP) to convert direct current coming from a battery of the vehicle into alternating current in order to provide an appropriate voltage to the primary side 8 of the transformer 6. As shown here, the converter 16 comprises four transistors 18, with two top transistors 18A and two bottom transistors 18B. Each top transistor 18A is associated with a bottom transistor 18B, and form a respective half-bridge, or phase leg. The top 18A and the bottom transistor 18B of a respective half-bridge are controlled inversely. In order to avoid short circuit of the converter input, an additional interlock time is introduced in the command between the top 18A and the bottom 18B transistors of each phase leg. This additional interlock time leads to a delayed switch-on command of the top transistor 18A or the bottom transistor 18B of one half-bridge compared to the switch-off command for the top transistor 18A or the bottom transistor 18B of one half-bridge.

[0053] The converter 16 is operated by phase shifts, with each phase shift being comprised between 0 and 180 °. Additionally, a duty cycle ratio is associated with each phase shift of the converter 16, such duty cycle ratio being comprised between 0 and 1. As an example, a phase shift of 0 ° corresponds to a duty cycle ratio of 0, a phase shift of 90 ° corresponds to a duty cycle ratio of 0.5 and a phase shift of 180 ° corresponds to a duty cycle ratio of 1.

[0054] The converter 16 generates a voltage waveform 20, such waveform 20 being visible on Figures 3 to 5. The voltage waveform 20 can be divided into a plurality of PWM periods 22, with both a positive pulse 24 and a negative pulse 26 occurring in the waveform 20 for each PWM period 22. For a duty cycle ratio equal to 1, positive or negative voltages are applied successively during all given PWM periods, without zero voltage values, each positive or negative voltage being applied during half of the PWM period, whereas for a duty cycle ratio equal to 0.5 for instance, positive and negative voltages are applied for smaller duration during the given PWM period, with zero voltage values provided between them, each positive or negative voltage being applied for half the time of what is done for the duty cycle ratio equal to 1.

[0055] As has been mentioned hereinbefore, the electric machine 1 can be used to provide torque to the wheels of the hybrid or electric vehicle. To control torque, and more generally to ensure an efficient operation of the electric machine 1, it is necessary to accurately determine the current of the rotor 4.

[0056] To this end, a method for rotor current estimation can be used. A first current estimation method, which covers several embodiments, will now be described in relation with Figures 3 to 5. On Figures 3, 4 and 5, respectively corresponding to a first, a second and a third embodiment of the first current estimation method, different waveforms are illustrated. These waveforms correspond, from top to bottom, to the voltage waveform 20 generated by the converter 16, a corresponding primary transformer current waveform 28 generated in the transformer 6, and a corresponding rotor current waveform 30 generated in the rotor 4. It is here meant by “corresponding” that the voltage waveform 20, the primary transformer current waveform 28 and the rotor current waveform 30 are observed in the same PWM period.

[0057] The first embodiment of the first current estimation method, which will hereinafter also be called Method A, is illustrated on Figure 3. In this first embodiment, i.e. in Method A, the current estimation method comprises an identification step 32 in which, for a given PWM period 22, sampling center locations for the primary transformer current are identified in the voltage waveform 20 such that they correspond to both the middle of a positive pulse and the middle of a negative pulse. In other words, a first sampling center point is identified in location 34, which is in the middle of a positive voltage pulse and a second sampling center point is identified in location 36, which is identified in the middle of the following negative voltage pulse.

[0058] The first embodiment then comprises a sampling step 38 in which sampling of the primary transformer 6 in the primary transformer current waveform 28 is performed; such current values are obtained from the first site SI. The sampling is performed with respect to the two sampling center locations 34 and 36 which were obtained from the identification step 32. In the sampling step 38 the absolute of the current value 40 corresponding to the first sampling center location 34 in the voltage waveform 20 is obtained. This value corresponds to the average value of the primary transformer current within the positive voltage pulse 24. Correspondingly, within the negative voltage pulse 26 the absolute of the current value 42 corresponding to the sampling center location 36 is taken.

[0059] Once the absolute values corresponding to the samples have been determined, the current estimation method comprises, for Method A, an averaging step 37 in which the absolute of the current value 40 corresponding to the positive voltage pulse 24 and the absolute of the current value 42 corresponding to the negative voltage pulse 26 are averaged together. In other words, a first value 40 corresponding to the first sampling center location 34 is taken in the primary transformer current waveform 28, a second value 42 corresponding to the second sampling center location 36 is taken in said primary transformer current waveform 28, and then the absolute of the first value 40 and the second value 42 are added together and divided by two to obtain an averaged absolute current value. The primary transformer current measurement resulting in the values 40, 42 is for instance carried out on the first site S 1. It should be noted that such values 40, 42 may be acquired either through single- shot sampling, meaning by sampling a single value, or through oversampling, meaning by sampling a plurality of values on the same pulse. More precisely, in case of oversampling the samples are taken such they their center location corresponds to the sampling center location 34 in case of a positive voltage pulse 24 or to the sampling center location 36 in case of the negative voltage pulse 26. The samples taken in the fast oversampling case are all located within the positive or negative voltage pulse 24, 26. The given PWM period 22 thus corresponds to an evaluation period.

[0060] After the averaging step 37, Method A comprises an estimation step 44. In this estimation step 44, the averaged current value obtained in the averaging step 38 is used to determine a corresponding value of the rotor current in the rotor current waveform 30. For this estimation step 44, a transformer transformation ratio, corresponding to a relation between the inductance of the transformer 6 and its mutual inductance, is considered. The estimation of the rotor current value can then be obtained thanks to the following formula: with:

[0061] “ip” is the rotor current value to be estimated;

[0062] “Ln” is the inductance of the transformer 6;

[0063] “M12” is the mutual inductance of the transformer 6; and

[0064] “ii.avg” is the averaged current value obtained during the averaging step 38.

[0065] Alternatively, the estimation of the rotor current value can be obtained thanks to the following formula: ip« xil avg, with x being an effective transformation ratio which can be used instead of the transformer transformation ratio when inductance values are current dependent rather than constant.

[0066] The rotor current estimate value is marked on the rotor current waveform 30 as a star. As can be seen on Figure 3, this rotor current value obtained through the aforementioned formula is valid at a point corresponding to the middle between the first value 40 and the second value 42 in the primary transformer current waveform 28, which also means that it is valid at a point corresponding to the middle between the first sample 36 and the second sample 38 in the voltage waveform 20. There is thus one valid rotor current value for each combination of a first value 40 and a second value 42, meaning one valid rotor current value for each combination of a positive pulse and a negative pulse. There is one valid current value per PWM period 22.

[0067] The second embodiment of the first current estimation method, which will hereinafter also be called Method B, is illustrated on Figure 4. It contains an identification step 32, a sampling step 38, a correction step 39 and an estimation step 44.

[0068] The identification step 32 is similar to the identification step of Method A, in which, for given PWM half-periods 23 and 25, sampling locations for the primary transformer current are identified in the voltage waveform 20 such that they correspond to both the middle of a positive pulse 24 and the middle of a negative pulse 26. In other words, a first sampling center location for a first PWM half-period 23 is identified in location 34, which is in the middle of a positive voltage pulse 24, and a second sampling center location for a second PWM half-period 25 is identified in location 36, which is identified in the middle of the following negative voltage pulse 26. Method B contains an additional identification step 46 in which two surrounding additional sampling center locations in zero voltage phase before and after the corresponding active pulses 24, 26 are identified. In case of the positive voltage pulse 24, the additional sampling center locations 48 and 50 are identified. In case of the negative voltage pulse 26 the additional sampling center locations 50 and 51 are identified.

[0069] Method B also comprises a sampling step 38 and an additional sampling step 47 in which sampling is performed within the primary transformer current waveform 28; such current values are obtained from the first site S 1. Similarly to method A the sampling in the sampling step 38 is more specifically performed with respect to the sampling center locations which were identified in the identification step 32. In case of the positive voltage pulse 24 the sample 40 corresponding to the sampling center location 34 from the identification step is obtained. In case of the negative voltage pulse 26 the sample 42 corresponding to the sampling center location 36 from the identification step is obtained. The additional sampling step 47 takes samples which correspond to the additional sampling center locations identified in the additional identification step 46. In case of the positive voltage pulse 24 the samples 52 and 54 are taken in the primary transformer current which correspond to the additional sampling center locations 48 and 50 from the additional identification step 46. In case of the negative voltage pulse 26 the samples 54 and 55 are taken in the primary transformer current which correspond to the additional sampling center locations 50 and 51 from the additional identification step 46.

[0070] Unlike the sample 34 in the middle of the positive voltage pulse 24 and the sample 36 in the middle of the negative voltage pulse 26, the additional samples 48, 50 in case of a positive voltage pulse 24 and the additional samples 50, 51 in case of a negative voltage pulse 26 are not in an active pulse but at zero-voltage, i.e. between two pulses. For instance, the first additional sample 52 is taken before a given positive pulse, the second additional sample 54 is taken after said positive pulse, or conversely the first additional sample 54 is taken before a given negative pulse, the second additional sample 55 is taken after said negative pulse. The two consecutive additional primary transformer current samples 52, 54 in case of the first PWM halfperiod interval 23, or the two additional primary transformer current samples 54, 55 in case of second PWM half-period interval 25, are taken for instance but not necessarily at the beginning and at the end of a given PWM half-period.

[0071] During the correction step 39, the current values obtained during the sampling step 38 and the additional sampling step 47 are processed. In case of the first evaluation period 23 for the positive pulse a correction value is calculated from the average of the primary transformer current samples from locations 52 and 54. The correction value for the positive pulse obtained from samples 52 and 54 is then subtracted from the primary transformer current sample 40 resulting in a corrected primary transformer current value for the positive pulse corresponding to sampling center location 34. In case of the second evaluation period 25 for the negative pulse a correction value is calculated from the average of the primary transformer current samples from location 54 and 55. The correction value is then subtracted from the primary transformer current sample 42 resulting in a corrected primary transformer current value for the negative pulse corresponding to the sampling center location 36. Thus a corrected primary transformer current value for each PWM half-period is obtained. In this case, the acquisition period is thus the PWM half-period.

[0072] Interestingly, similarly to Method A the current values 40, 42 and the additional current values 52, 54, 55 of Method B may be collected by the single-shot sampling technique or by the oversampling technique.

[0073] In the estimation step 44, the rotor current is estimated using the result from the subtraction of the correction value from the corresponding current value. Such estimation is determined using the following formula:

[0074] • 11 • * 1

[0075] Ij? ~ — l- corr-> With.

[0076] M12

[0077] “ip” is the rotor current value to be estimated;

[0078] “Ln” is the inductance of the transformer 6;

[0079] “M12” is the mutual inductance of the transformer 6; and

[0080] “ii.corr” is the corrected current value obtained during the averaging step 38.

[0081] Alternatively, the estimation of the rotor current value can be obtained thanks to the following formula: ip« xil corr, with x being an effective transformation ratio which can be used instead of the transformer transformation ratio when inductance values are current dependent rather than constant.

[0082] As is illustrated on Figure 4, the rotor current value which is marked as a star is valid at a point of the rotor current waveform 30 corresponding to each first value 40 and each second value 42 of the primary transformer current waveform 28. As a result, there is one valid rotor current value for each value 40, 42 taken in an active pulse. There are two valid current values per PWM period 22.

[0083] Compared to the first embodiment, Method B offers a more accurate rotor current estimation with less delay in dynamic conditions.

[0084] The third embodiment of the first current estimation method, which combines elements from Methods A and B, will hereinafter also be called Method A+B. Method A+B is illustrated on Figure 5. Method A+B implements the identification step 32 as it has been previously described, thus identifying the first sampling center location 34 and the second sampling center location 36 from the voltage waveform 20, more particularly from one of its positive pulses and one of its negative pulses in the same PWM period. The third embodiment comprises, as Method B, the additional identification step 46. In this additional identification step 46, the sampling center locations 48, 50, 51 for sampling in the zero-voltage phases are identified around each of the first voltage pulse 24 and the second voltage pulse 26. In other words, a first additional sampling center location 48 is identified at zero-voltage before the positive pulse 24, a second additional sampling center location 50 is identified at zero-voltage after said positive pulse, and a third additional sampling center location 51 is identified at zero-voltage after the negative pulse.

[0085] Once both the identification step 32 and the additional identification step 46 have been completed, the third embodiment comprises a sampling step 38 and an additional sampling step 47, alike to that of Method B. Multiple current values are taken in the primary transformer current waveform 28, namely the first value 40 corresponding to the sampling center location 34, the second current value 42 corresponding to the second sampling center location 36, as well as a first additional current value 52 for the first additional sampling center location 48, a second additional current value 54 for the second sampling center location 50 and a third additional current value 55 for the third sampling center location 51. Similarly to the correction step 39 of Method B the first additional current value 52 and the second additional current value 54 corresponding to each additional sampling center location 48, 50 are averaged in order to obtain the first correction value. Also, the second additional current value 54 and the second additional current value 55 corresponding to each additional sampling center location 50, 51 are averaged in order to obtain the second correction value. Then, the first and second correction values are subtracted from the corresponding current value 40, 42. As a result, a first corrected current value is obtained for the first current value 40 and a second corrected current value is obtained for the second current value 42.

[0086] Finally, an averaging step 37 similar to Method A is applied using the first and second corrected current value. The difference to Method A is that it is not the current values 40, 42 which are used for the averaging step, but rather corrected values corresponding to the same sampling center locations 34 and 36, namely the first and second corrected current values, obtained from the previous correction step. The first corrected value and the second corrected value are then averaged together. Such averaging is similar to the averaging step 37 carried out in Method A.

[0087] The averaged corrected value of Method A+B is then used in the estimation step 44, which to determine the rotor current uses the following formula: with.

[0088] “ip” is the rotor current value to be estimated;

[0089] “Ln” is the inductance of the transformer 6;

[0090] “M12” is the mutual inductance of the transformer 6; and

[0091] “ii.avgcorr”” is the averaged corrected current value obtained during the averaging step 38.

[0092] Alternatively, the estimation of the rotor current value can be obtained thanks to the following formula: ip« i ,avgcorr with x being an effective transformation ratio which can be used instead of the transformer transformation ratio when inductance values are current dependent rather than constant.

[0093] As can be seen on Figure 5, the rotor current value marked as a star in the rotor current waveform 30 is valid at a point corresponding to the middle between the first current value 40 and the second current value 42 in the primary transformer current waveform 28; in other words, it is valid at a point corresponding to the middle between the first sampling center location 36 and the second sampling center location 38 in the voltage waveform 20. In Method A+B, there is thus one valid rotor current value for each combination of a first current value 40 and a second current value 42, meaning one valid rotor current value for each combination of a positive pulse and a negative pulse. There is one valid current value per PWM period 22.

[0094] Compared to the first embodiment and the second embodiment, Method A+B offers an accurate rotor current estimation even in dynamic conditions. The delay characteristics are the same as in Method A. It should be noted that, while each of the first, second and third embodiments has been described in relation to an acquisition period corresponding to either a given PWM period or a given PWM half-period, the sampling step 32 could alternatively be carried out in an acquisition period covering multiple PWM periods or multiple PWM half-periods. In this case, it is not necessary for the sampling step 32 to be performed in each PWM period or in each PWM half-period. The samples from these longer acquisition periods would however need to be averaged, for instance via a weighted averaging method.

[0095] While methods A, B and A+B are embodiments of the current estimation method according to the invention in which the rotor current is estimated based on the primary transformer current waveform 28, other current estimation methods can be used to determine the rotor current.

[0096] A second current estimation method, this time a current estimation method according to the invention, will now be described in relation to Figure 6. The current estimation method according to the invention will hereinafter also be referred to as Method C.

[0097] The current estimation method according to the invention starts with a determination step 131. In this determination step 131, a primary transformer current estimate is calculated in a given PWM half-period.

[0098] As an example, a first position Pl can be estimated in the primary transformer current waveform 28 in case of a positive voltage pulse half-period 23, or a second position P2 can be estimated in said waveform 28 in case of a negative voltage pulse half-period 25. It should be noted that direct measurement of the primary transformer current from the first site SI, as is the case with the previous methods A, B and A+B, requires very precise timing and very small sensor delay variation, whereas an estimate of a corresponding value such as that of Method C with first position Pl for a positive voltage pulse or second position P2 for a negative voltage pulse is more robust in terms of timing errors.

[0099] In order to calculate the primary transformer current estimate, the following formula is used: ilpwrbalance i Pin, halfperiod, avg / duty half period / DC , with.

[0100] “i ipwrbaiance” is the primary transformer current estimate; +- corresponds to the sign information of the expected primary transformer current, with + for positive voltage pulse and, - for negative voltage pulse;

[0101] “r|” is an efficiency parameter, which is used to account for the power losses happening in the power converter 16; in general this parameter is expected to be operating point dependent and therefore can be represented by a look-up table;

[0102] “Pin, halfperiod, avg” is an average value of the input power of the converter 16 in the halfperiod under evaluation;

[0103] “dutyhalfperiod” corresponds to the duty cycle ratio which corresponds to the halfperiod under evaluation; and

[0104] “UDC” is the input voltage of the converter 16.

[0105] It should be noted that “Pin, halfperiod, avg” may also be expressed as: with.

[0106] “UDC, avg, halfperiod” corresponding to the input voltage in the PWM half-period under evaluation, which can be measured on the input capacitor 17, and

[0107] “IDC, avg, halfperiod” corresponding to the input current in said PWM half-period under evaluation, which can be measured from a second site S2 illustrated on Figure 2, such second side S2 being on the converter 16 input side.

[0108] Said input current is defined as:

[0109] In order to implement such integral, either fast oversampling using a delta- sigma analog-to-digital converter (DSADC) or fast acquisition and accumulation using a successive-approximation-register analog-to-digital converter (SAR-ADC) may be used. It should be noted that in order to obtain a representative value of the DC input current over a PWM half-period, which corresponds to the sampling interval, the acquisition rate must be as fast as possible. In addition, it is necessary to synchronize the acquisition with the PWM generation in order to obtain the average value of the DC current which corresponds to the duty cycle realized in a given PWM halfperiod. Once the primary transformer current estimate has been determined, it needs to be corrected. To this end, an identification step 132 is performed. In this identification step 132, sampling center locations during zero-voltage phase are identified in the voltage waveform 20. More particularly, a first sampling center location sample 134 is identified and a second sample sampling center location 136 is identified, both at zero-voltage. The first sampling center location 134 and the second sampling center location 136 are identified within a PWM / 2 interval. As an example, the first sampling center location 134 is identified at the start of a given PWM half-period 23 and the second sampling center location 136 is identified at the end of said PWM half-period 23.

[0110] As is visible on Figure 6, the first sampling center location 134 and the second sampling center location 136 are identified so that they surround voltage pulses, for instance before a given positive pulse 124 and after said positive pulse 124, or alternatively before a given negative pulse 126 and after said negative pulse 126 in case of PWM half-period 25.

[0111] After the identification step 132, the current estimation method according to the invention comprises a sampling step 133. During this sampling step 133, current values corresponding to the sample center locations 134, 136 are acquired in the primary current waveform 28, such that a first current value 140 corresponding to the first sampling center location 134 and a second current value 142 corresponding to the second sampling center location 136 are taken. The primary transformer current samples from the sampling step are than averaged in the correction step 139. More precisely, the first current value 140 and the second current value 142 from the sampling step 133 are averaged together to obtain a correction value. The correction value is then subtracted from the primary transformer current estimate calculated during the previously performed determination step 131. As a result, a corrected current estimate “ii.pwrbaiancecorr” is obtained. This corrected current estimate corresponds to either a positive voltage pulse or a negative voltage pulse.

[0112] According to the embodiment, the first current value 140 and the second current value 142 may be acquired either thanks to a single-shot sampling method or thanks to an oversampling method; in other words, each of the first value 140 and the second value 142 corresponds to either a single value or a plurality of values respectively.

[0113] Method C concludes with an estimation step 144, in which the rotor current is estimated based on the corrected current estimate from the calculation step 133, the duty cycle ratio and a transformer transformation ratio, here a relation between the inductance of the transformer 6 and its mutual inductance. Consequently, the rotor current can be estimated thanks to the following formula: with.

[0114] “ip” is the rotor current value to be estimated;

[0115] “Ln” is the inductance of the transformer 6;

[0116] “M12” is the mutual inductance of the transformer 6; and

[0117] “ii.pwrbaiancecorr” is the averaged corrected current value obtained during the calculation step 33.

[0118] Alternatively, the estimation of the rotor current value can be obtained thanks to the following formula: ip« xil pwrbaiancecorr, with x being an effective transformation ratio which can be used instead of the transformer transformation ratio when inductance values are current dependent rather than constant.

[0119] The rotor current value is marked on the rotor current waveform 30 as a star on Figure 6. This rotor current value is valid at a point of the rotor current waveform 30 corresponding to the first position Pl and at a point corresponding to the second position P2 in the primary transformer current waveform 28. This means that there is a valid rotor current value for each active pulse of the primary transformer current waveform 28. There are two valid current values per PWM period 22.

[0120] Correspondingly, there is one valid rotor current value per PWM half period 23,25.

[0121] An alternative of Method C can be implemented in the context of the invention. Above, method C has been described with mainly a determination step 131 wherein a current estimate is calculated in a given PWM half-period from a power balance, an identification step 132 where sampling center locations 134, 136 for the additional primary transformer current samples are identified, a sampling step 133 where the sampling of the primary transformer current in zero voltage phases happens and a correction step 139 with current values 140, 142 resulting from the sampling step 133 with samples in the zero voltage phase. In the same way as the averaging step 37 in Method A and Method A+B was described above, an averaging step 137 wherein the current values sampled at positive and negative pulses are taken in account may be implemented. More precisely, during the determination step 131 it is possible to use two consecutive samples of rotor current estimate corresponding respectively to a positive voltage pulse 24 and to a negative voltage pulse 26. The correction step 139 is then applied, taking into account the current values 140, 142 resulting from the sampling step 133 to correct the two consecutive samples of rotor current estimate. This leads to the corrected current values Pl and P2. Similarly to the averaging step 37 in Method A and Method A+B, in the averaging step 137 in Method C the average of the absolute value of the corrected current value Pl, corresponding to the first PWM half-period 23, and the absolute value of the corrected current value P2, corresponding to the second PWM half-period 25, is calculated. In addition, it is also possible to use more than two consecutive PWM half-periods, and to use weighted averages of these consecutive PWM half-periods in order to get an improved rotor current estimate.

[0122] The method of rotor current estimation, be it Method A, Method B, Method A+B or Method C, can be combined with a machine model 56 in order to improve the accuracy of the rotor current estimation, as it is shown on Figure 7. The machine model 56 associated with the electric machine 1 corresponds to the following formula: with:

[0123] “uf” is the voltage supplied to the rotor;

[0124] “Rf” is a rotor resistance parameter which is temperature-dependent;

[0125] “if” is the rotor current to be estimated;

[0126] “Lff” is a rotor self-inductance parameter which is dependent on the rotor current, a rotor position, a stator d-current and a stator q-current;

[0127] “id”, “if” are stator d-current and stator q-current; “Mdf” is a stator d-axis to rotor winding coupling inductance parameter, which is dependent on the rotor current, the rotor position, the stator d-current, and the stator q-current; and

[0128] “Mqf” is a stator q-axis to rotor winding coupling inductance parameter which is dependent on the rotor current, the rotor position, the stator d-current and the stator q-current.

[0129] Figure 7 represents an observer structure 58, using the machine model 56 and possibly one or more of the aforementioned current estimation methods, i.e. Method A, Method B, Method A+B or Method C. This observer structure 58 is for instance chosen among a Luenberger observer as that of Figure 7, an extended Luenberger observer, or a Kalman filter.

[0130] As can be seen on Figure 7, the derivative of stator d-current the voltage supplied cZ i to the rotor Uf and the derivative of stator q-current are input into the machine model 56. The derivative of stator d-current — is associated with the stator d-axis to dt rotor winding coupling inductance and the rotor self-inductance parameters, the voltage supplied to the rotor Uf is associated with the rotor self-inductance as well, cZ i and the change of stator q-current is associated with the stator q-axis to rotor winding coupling inductance and the rotor self-inductance parameters in order to correspond to a differential equation 60 as follows:

[0131] Such a differential equation is implemented into an integrator 62 of the machine model 56 where it is integrated to provide the rotor current.

[0132] The Machine model can be used for open loop estimation by feeding in the necessary inputs Uf, The current derivatives can be obtained from sensor values or by prediction. Besides, for parameter adaptation the actual rotor winding temperature, stator d-current, stator q-current and actual rotor current estimate have to be used. In order to improve the efficiency of the machine model 56, it is when possible combined with at least one of the current estimation methods as previously described, or any other current estimation method. More precisely, one or more of Method A, Method B, Method A+B, Method C or another current estimation method is used, when reliable, to correct the rotor current provided by the integrator 62.

[0133] On the embodiment of Figure 7, both Method A+B and Method C may be used to provide a rotor current estimate to a correction signal consolidation system 64. It should be noted that while only Method A+B and Method C are used here, other embodiments may use, additionally or alternatively, among Method A, Method B or any other current estimation method. Here, the correction signal consolidation system 64 receives the rotor current estimate of Method A+B and that of Method C, and it can either use both, use only one of them or use neither according to their reliability.

[0134] When both the rotor current estimate from Method A+B and the rotor current estimate from Method C are unreliable, for instance because of a near-zero duty cycle or in case of a pulse-off state of the converter 16, said current estimates are canceled out by sending to the correction signal consolidation system 64 a corresponding negative value of the rotor current estimate. To this end, the observer structure 58 comprises a canceling mechanism 66 to send the corresponding negative value back to the correction signal consolidation system 64. The cancelling mechanism 66 is part of an observer correction mechanism 68. When both rotor current estimates are unreliable, a corrective value 70 obtained by the observer correction mechanism 68 is thus set to zero.

[0135] When only one of the rotor current estimate from Method A+B and the rotor current estimate from Method C is reliable, the correction signal consolidation system 64 isolates it and sends it to the observer correction mechanism 68.

[0136] When both the rotor current estimate from Method A+B and the rotor current estimate from Method C are reliable, the correction signal consolidation system 64 can choose only one of them to be sent to the observer correction mechanism 68, or alternatively use both. When both estimates are used, the correction signal consolidation system 64 can perform a weighted average operation. Additionally, smooth blending may be implemented in order to ensure there is a smooth transition between different current estimation methods.

[0137] In the observer correction mechanism 68, the reliable current estimate or weighted reliable current estimates, corresponding to an observer error signal, are multiplied by an observer gain 72. Such observer gain 72 may depend on the stator d-current, the stator q-current, the estimated rotor current, the rotor position and the rotor winding temperature. As a result of the multiplication with the observer gain 72, the observer correction mechanism 68 generates the corrective value 70, which is then sent to the equation 60 of the machine model 56 in order to correct it before it is integrated in the integrator 62.

[0138] As mentioned before, if there is no reliable rotor current estimate sent to the correction signal consolidation system 64 then the corrective value 70 is set to zero. In this case, the machine model 56 can be used in an open loop mode. Such open loop mode means that the observer structure 58 mainly relies on the change of stator d-current, on the voltage supplied to the rotor 2 and on the change of stator q-current inputs to determine the rotor current.

[0139] The present invention thus covers a current estimation method capable of estimating, in an accurate manner, the output current on the rotating part of an electric machine both in steady state conditions and in dynamic conditions, by sampling data from a non-rotating part of said electric machine.

[0140] Many modifications and other embodiments of the invention set forth herein will come to mind to one skilled in the art to which the invention pertains having the benefit of the teachings presented in the foregoing descriptions and the associated drawings. Therefore, it is to be understood that the invention is not to be limited to the specific embodiments disclosed and that modifications and other embodiments are intended to be included within the scope of the appended claims. Although specific terms are employed herein, they are used in a generic and descriptive sense only and not for purposes of limitation.

Claims

CLAIMS1. A method of rotor current estimation for a rotating part of an electric or hybrid vehicle electric machine (1), in which the electric machine (1) comprises a converter (16) which is operated with phase shifts, with a duty cycle ratio associated with a phase shift, the converter (16) providing voltage to a transformer (6) of the electric machine (1), such voltage being provided to a primary side (8) of the transformer (6), with the converter (16) generating a voltage waveform (20) between the converter (16) and the transformer (6), the current estimation method comprising:- a determination step (131) in which a current estimate in one PWM half-period is calculated,- an identification step (132) in which in a corresponding PWM half-period of a voltage waveform (20) generated by the converter (16), a first sampling center location (134) and a second sampling center location (136) are identified at zero-voltage at a PWM / 2 interval;- a sampling step (133) in which a first current value (140) and a second current value (142) corresponding respectively to the first sampling center location (134) and the second sampling center location (136) are taken in a primary transformer current waveform (28);- a correction step (139) in which the current estimate calculated in the determination step (131) is corrected based on the current values (140, 142) taken during the sampling step (133); and- an estimation step (144) in which a current value for the rotating part is estimated according to the corrected calculated current estimate from the correction step (139), such current value estimation taking into account a transformer transformation ratio.

2. The method of rotor current estimation according to the preceding claim, wherein during the determination step (131) two consecutive primary transformer current values corresponding respectively to a positive voltage pulse (24) and a negative voltage pulse (26) in the voltage waveform (20) are taken, and wherein the correction step (139) comprises a substep wherein corrected consecutive current values (Pl, P2) are calculated based on the current values (140, 142) taken during the sampling step (133) and an additional averaging step (137) in which the absolute values of the corrected current values (Pl, P2) are averaged.

3. The method of rotor current estimation according to the preceding claims, wherein the current value (140, 142) corresponds to a single value.

4. The method of rotor current estimation according to claim 1 and claim 2, wherein the current value (140, 142) corresponds to a plurality of values.

5. The method of rotor current estimation according to any of the preceding claims, wherein the first sampling center location (134) is identified before a given positive or negative pulse (124, 126) and the second sampling center location (136) is identified after the given pulse (124, 126).

6. The method of rotor current estimation according to any of the preceding claims, wherein the first sampling center location (134) and the second sampling center location (136) are identified at the start and at the end of the PWM half-period.

7. The method of rotor current estimation according to any of the preceding claims, wherein during the determination step (131), the current estimate corresponding to the PWM half-period is calculated from an input power of the converter (16), the duty cycle ratio and an input voltage of the converter (16).

8. The method of rotor current estimation according to the preceding claim, wherein the input power of the converter (16) is obtained from at least an input current of the converter (16), an average value of the input current of one PWM half-period corresponding to the duty cycle of said PWM half-period.

9. The method of rotor current estimation according to any of the preceding claims, wherein the transformer transformation ratio is a relation between inductance and mutual inductance of the transformer (6).

10. The method of rotor current estimation according to any of the preceding claims, being combined with a machine model (56) by embedding the machine model (56) in an observer structure (58).

11. The method of rotor current estimation according to the preceding claim, wherein a corrective value (70) for the observer structure (58) is calculated from an observer error signal based on the rotor current estimate from the estimation step (144).

12. The method of rotor current estimation according to the preceding claim, wherein the corrective value (70) is set to zero when the observer error signal is unreliable.

Images