APPARATUS AND METHOD FOR DETERMINING A TEMPERATURE OF A ROTOR

Abstract

The invention relates to an apparatus for determining a temperature of a rotor of an electric machine. The apparatus comprises an interface and a computing device. Input variables which are dependent on the operation of the electric machine are received via the interface. The computing device calculates, by means of a physical model, a first contribution to the heat transfer on the rotor on the basis of at least one of the input variables. In addition, the computing device calculates, by means of an artificial intelligence model, a second contribution to the heat transfer on the rotor on the basis of at least one of the input variables. The computing device calculates the temperature of the rotor on the basis of the first contribution and the second contribution to the heat transfer on the rotor.

Claims

1. An apparatus (1) for determining a temperature of a rotor (3) of an electric machine (2), the apparatus comprising: an interface (11) configured to receive input variables which are dependent on the operation of the electric machine (2); and a computing device (12) which is configured to: use at least one of the input variables to calculate a first contribution to the heat transfer on the rotor (3) by means of a physical model, use at least one of the input variables to calculate a second contribution to the heat transfer on the rotor (3) by means of an artificial intelligence model (101), and use the first contribution and the second contribution to the heat transfer on the rotor (3) to calculate the temperature of the rotor (3).

2. The apparatus (1) according to claim 1, wherein the first contribution to the heat transfer on the rotor (3) includes a heat flow from a stator (4) of the electric machine (2) to the rotor (3) and/or a heat flow from a coolant (5) of the electric machine (2) to the rotor (3).

3. The apparatus (1) according to claim 2, wherein the at least one input variable includes a current temperature of the stator (4) and/or a current temperature of the coolant (5), and wherein the computing device (12) is configured to calculate the first contribution to the heat transfer on the rotor (3) using a temperature difference between the current temperature of the stator (4) and the most recently calculated temperature of the rotor (3) and/or using a temperature difference between the current temperature of the coolant (5) and the most recently calculated temperature of the rotor (3).

4. The apparatus (1) according to claim 1, wherein the input variables include at least one current of the electric machine (2), at least one voltage of the electric machine, a DC link voltage of a battery coupled to the electric machine (2), an effective phase current of the electric machine (2), a pulse width modulation frequency, a rotational speed of the electric machine (2), a torque of the electric machine (2), at least one modulation variable of the electric machine (2), an ambient temperature of the electric machine (2) and/or a transmission temperature of a transmission coupled to the electric machine (2).

5. The apparatus (1) according to claim 1, wherein the computing device (12) is further configured to check the plausibility of the input variables and/or the first contribution to the heat transfer on the rotor (3) and/or the second contribution to the heat transfer on the rotor (3).

6. The apparatus (1) according to claim 5, wherein the computing device (12) is configured to check the plausibility using at least one physical formula and/or an artificial intelligence method.

7. A computer-implemented method for determining a temperature of a rotor (3) of an electric machine (2), comprising: providing (S1) input variables which are dependent on the operation of the electric machine (2); calculating (S2) a first contribution to the heat transfer on the rotor (3) using at least one of the input variables and by means of a physical model; calculating (S3) a second contribution to the heat transfer on the rotor (3) using at least one of the input variables and by means of an artificial intelligence model (101); and calculating (S4) the temperature of the rotor (3) using the first contribution and the second contribution to the heat transfer on the rotor (3).

8. A method according to claim 7, wherein the artificial intelligence model (101) is trained using test bench data of the electric machine (2).

9. (canceled)

10. A non-transitory, computer-readable storage medium comprising executable instructions that when executed by a computer cause the computer to determine a temperature of a rotor (3) of an electric machine (2), by: obtaining (S1) input variables which are dependent on the operation of the electric machine (2); calculating (S2) a first contribution to the heat transfer on the rotor (3) using at least one of the input variables and by means of a physical model; calculating (S3) a second contribution to the heat transfer on the rotor (3) using at least one of the input variables and by means of an artificial intelligence model (101); and calculating (S4) the temperature of the rotor (3) using the first contribution and the second contribution to the heat transfer on the rotor (3).

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The figures show:

[0031] FIG. 1 a schematic block diagram of an electric machine and an apparatus for determining a temperature of a rotor of the electric machine according to one embodiment of the invention;

[0032] FIG. 2 a schematic block diagram explaining the calculation of the temperature of the rotor of the electric machine;

[0033] FIG. 3 a schematic block diagram explaining a training method of the artificial intelligence model; and

[0034] FIG. 4 a flow chart of a method for determining a temperature of a rotor of an electric machine according to one embodiment of the invention.

[0035] In all figures, identical or functionally identical elements and devices are provided with the same reference sign. The numbering of method steps is for the sake of clarity and is generally not intended to imply a specific chronological order. It is in particular also possible to carry out multiple method steps at the same time.

DETAILED DESCRIPTION

[0036] FIG. 1 shows a schematic block diagram of an electric machine 2 and an apparatus 1 for determining a temperature of a rotor 3 of the electric machine 2. The electric machine 2 further comprises a stator 4 and a coolant reservoir 5.

[0037] The apparatus comprises an interface 11 and a computing device 12. The interface 11 is preferably wired, but can also be a wireless interface. The computing device 12 comprises memories for storing the received data and computing components, such as microprocessors, application-specified circuits (ASICs) or the like.

[0038] Input variables which are dependent on the operation of the electric machine are received via the interface 11. The input variables include physical or mechanical variables (e.g. temperatures, a torque, a rotational speed and the like) and/or electrical variables (e.g. currents and the like).

[0039] The computing device 12 uses at least one of the input variables (such as the temperature of the stator 4 and the temperature of the coolant 5) to calculate a first contribution to the heat transfer on the rotor 3 by means of a physical model. The computing device 12 further uses at least one of the input variables (such as currents and a torque) to at the same time or subsequently calculate a second contribution to the heat transfer on the rotor 3 by means of an artificial intelligence model. The same or at least partially different input variables can be taken into account for this purpose. The artificial intelligence model can include a neural network or a Gaussian process regressor. The computing device uses the first contribution and the second contribution to the heat transfer on the rotor to calculate the temperature of the rotor 3.

[0040] The calculated temperature can be output via the interface 11. A control unit can receive the calculated temperature of the rotor 3, for example, and use it to control the electric machine.

[0041] FIG. 2 shows a schematic block diagram explaining the calculation of the temperature of the rotor 3 of the electric machine 2 by means of a hybrid thermal model. This is a 1-node network model.

[0042] The temperature T.sub.cool(t) of the coolant 5 and the temperature T.sub.start(t) on the stator 4 are received as an input variable. Input variables I.sub.1, . . . I.sub.n of the artificial intelligence model 101 are furthermore provided, which include currents of the electric machine 2, voltages of the electric machine 2, a DC link voltage of a battery coupled to the electric machine 2, an effective phase current of the electric machine 2, a pulse width modulation frequency, a rotational speed of the electric machine 2, a torque of the electric machine 2, modulation variables of the electric machine 2, an ambient temperature of the electric machine 2 and/or a transmission temperature of a transmission coupled to the electric machine 2.

[0043] A difference between the current temperature of the coolant 5 and the most recently calculated temperature of the rotor 3 is calculated by the computing device 12:

T.sub.cool(t)T.sub.rot(t1).

[0044] The computing device 12 also calculates a difference between the current temperature of the stator 4 and the most recently calculated temperature of the rotor 3:

T.sub.start(t)T.sub.rot(t1).

[0045] A heat flow from the coolant 5 to the rotor 3 is calculated by multiplying (104) by a specific thermal conductivity of the coolant 5 stored in a first lookup table 102. A heat flow from the stator 4 to the rotor 3 is calculated as well, by multiplying (105) by a specific thermal conductivity of the stator 4 stored in a second lookup table 103. The two calculated heat flows constitute a first contribution to the, in particular convective, heat transfer on the rotor 3.

[0046] The artificial Intelligence Model 101 calculates a second contribution to the heat transfer on the rotor 3 using the input variables I.sub.1, . . . I.sub.n. The first contribution to the heat transfer on the rotor 3 is added (106) to the second contribution to the heat transfer on the rotor 3. The heat losses within the rotor 3 in particular constitute a second contribution to the heat transfer on the rotor 3.

[0047] The computing device 12 divides (107) the summed contributions to the heat transfer on the rotor 3 by the specific heat capacity C.sub.th,rot of the rotor 3, wherein the specific heat capacity C.sub.th,rot of the rotor 3 is the ratio of the added or removed heat and the resulting change in the temperature:

[00001]$C_{\mathrm{th},\mathrm{rot}}=\frac{dQ}{\mathrm{dT}}$

[0048] This variable is integrated (108) over time in order to calculate (109) the instantaneous temperature of the rotor 3.

[0049] The calculation of the first contribution and the calculation of the second contribution can preferably be carried out in parallel.

[0050] FIG. 3 shows a schematic block diagram explaining a training method of the artificial intelligence model 101. Separation into a static component (steady-state loss map) and a dynamic component facilitates the training of the artificial intelligence model 101 and reduces its size. A separate artificial intelligence model is preferably trained for each type of electric machine 2. In the application phase, the artificial intelligence model 101 is trained with test bench measurements in order to calculate features (labels). The test bench measurements can be telemetry measurements of the temperature of the rotor 3 under different load and ambient conditions.

[0051] To train the artificial intelligence model 101, the required label is determined backward from the telemetry measurements of the temperature of the rotor 3 (109). For this purpose, a time integration is carried out to calculate a difference between the current temperature of the rotor 3 and the previous temperature of the rotor 3 (A):

T=T.sub.rot(t)T.sub.rot(t1).

[0052] This is followed by a multiplication with the specific heat capacity C.sub.th,rot of the rotor 3 (B):

C.sub.th,rot.Math.T.

[0053] Lastly, the heat loss P.sub.loss on the rotor is calculated as the label by subtracting the first contribution. This label is used to train the artificial intelligence model 101. The trained artificial intelligence model 101 for the heat loss modeling is then used forward as a building block throughout the hybrid model for calculating the rotor temperature.

[0054] The robustness of the artificial intelligence model 101 can optionally be ensured by training with artificially noisy data. The data sets are enriched by systematically adding noise to the training signals. This method ensures the robustness and a certain degree of noise tolerance of the artificial intelligence model 101.

[0055] It is also possible to use plausibility check functions to check the input variables (e.g. the temperature of the stator 4 or the coolant 5) for (sensor) errors. This plausibility check can be carried out with at least one physical formula, which is used to check whether the input values are within a physically valid range. The plausibility check can also be carried out using machine learning methods that can detect anomalies.

[0056] In the event of a sensor failure or error detection, the artificial intelligence model 101 will reference the last plausible value of the affected input signal and provide a worst case prediction of the temperature of the rotor 3 to ensure component protection.

[0057] FIG. 4 shows a flow chart of a method for determining a temperature of a rotor 3 of an electric machine 2.

[0058] In a first step S1, input variables which are dependent on the operation of the electric machine 2 are received. The input variables are provided to a computing device 12.

[0059] In a second step S2, the computing device 12 uses at least one of the input variables to calculate a first contribution to the heat transfer on the rotor 3 by means of a physical model. The computing device 12 can in particular take into account a heat flow from a stator 4 of the electric machine 2 to the rotor 3 and a heat flow from a coolant 5 of the electric machine 2 to the rotor 3 to calculate the first contribution to the heat transfer on the rotor 3.

[0060] In a third step S3, the computing device 12 uses at least one of the input variables to calculate a second contribution to the heat transfer on the rotor 3 by means of an artificial intelligence model 101. For this purpose, the artificial intelligence model 101 can have been trained in a preceding training procedure using telemetry data, as explained in more detail in connection with FIG. 3.

[0061] In a fourth step S4, the computing device 11 lastly uses the first contribution and the second contribution to the heat transfer on the rotor 3 to calculate the temperature of the rotor 3. For this purpose, the specific heat capacity C.sub.th,rot of the rotor 3 can be taken into account as described above.

[0062] The calculated temperature of the rotor 3 is output and taken into account, for example, by a control unit when controlling the electric machine 3. The determined temperature of the rotor 3 can in particular be taken into account in a derating method.