Search

EP-4738652-A1 - WIRELESS POWER SYSTEM AND METHOD FOR TRAINING A MACHINE LEARNING MODEL

EP4738652A1EP 4738652 A1EP4738652 A1EP 4738652A1EP-4738652-A1

Abstract

The invention concerns a wireless power system (1), comprising: a first unit (10, 20) including a pad (11, 21) and a wireless communication unit (12, 22), wherein the pad (11, 21) is configured to wirelessly transmit or receive power to or from a second unit (10, 20), and the wireless communication unit (12, 22) is configured to receive a communication signal from the second unit (10, 20); and a control unit (2), wherein the control unit (2) comprises: an input unit configured to receive the communication signal from the wireless communication unit; a detection unit configured to detect at least one signal strength value and/or at least one phase value of the communication signal; at least one trained machine learning model (200, 204 - 206) programmed to determine a relative position between the first unit (10, 20) and the second unit (10, 20) based at least on the detected signal strength value and/or phase value as input parameters (201); and an output unit (203) configured to output the relative position. The invention also concerns a power transmission system (100) and a method for training a machine learning model (200, 204 - 206).

Inventors

  • Enderlin, Jonas
  • LOESER, LEON ANDREA

Assignees

  • Delta Electronics (Thailand) Public Co., Ltd.

Dates

Publication Date
20260506
Application Date
20241029

Claims (15)

  1. Wireless power system (1), comprising: • a first unit (10, 20) including a pad (11, 21) and a wireless communication unit (12, 22), wherein the pad (11, 21) is configured to wirelessly transmit or receive power to or from a second unit (10, 20), and the wireless communication unit (12, 22) is configured to receive a communication signal from the second unit (10, 20); and • a control unit (2), wherein the control unit (2) comprises: • an input unit configured to receive the communication signal from the wireless communication unit; • a detection unit configured to detect at least one signal strength value and/or at least one phase value of the communication signal; • at least one trained machine learning model (200, 204 - 206) programmed to determine a relative position between the first unit (10, 20) and the second unit (10, 20) based at least on the detected signal strength value and/or phase value as input parameters (201); and • an output unit configured to output the determined relative position.
  2. Wireless power system (1) according to claim 1, wherein the detection unit is configured to detect, as the phase value input parameter (201) for the trained machine learning model (200, 204- 206), a phase-difference between multiple receptions of the communication signal.
  3. Wireless power system (1) according to any one of the foregoing claims, wherein the detection unit is configured to detect, as the signal strength value input parameter (201) for the trained machine learning model (200, 204- 206), an amplitude of the received communication signal.
  4. Wireless power system (1) according to claims 2 or 3, wherein the wireless communication unit (12, 22) comprises at least one main communication antenna (13, 23) for transmitting and receiving the communication signal, and at least one, especially two to four, auxiliary sensing antennas (24) for detecting the relative position, and wherein the detection unit is especially configured to detect a signal strength difference and/or phase-difference between multiple receptions of the communication signal via the at least one main communication antenna (13, 23) and the at least one auxiliary sensing antenna (24) as input parameters (201).
  5. Wireless power system (1) according to any one of the foregoing claims, wherein the control unit (2) comprises at least two trained machine learning models (200, 204 - 206), wherein a first trained machine learning model (204) is programmed to determine the relative position based on the signal strength value as an input parameter (201), and a second trained machine learning model (205) is programmed to determine the relative position based on the phase value as an input parameter (201).
  6. Wireless power system (1) according to claim 5, wherein the control unit (2) comprises a further third trained machine learning model (206) programmed to classify, based on the signal strength value and the phase value as input parameters (201), the output relative position as output (203) from the first or the second trained machine learning model (204, 205).
  7. Wireless power system (1) according to claim 6, wherein the third trained machine learning model (206) is programmed to estimate the relative position and output from the first or the second trained machine learning model (204, 205) based on the estimated relative position.
  8. Wireless power system (1) according to claim 7, wherein the third trained machine learning model (206) is programmed to output from the first trained machine learning model (204) when the estimated relative position is above a predetermined distance threshold and to output from the second trained machine learning model (205) when the estimated relative position is below the predetermined distance threshold.
  9. Wireless power system (1) according to any one of the foregoing claims, wherein the at least one trained machine learning model (200, 204 - 206) is pre-trained with predetermined relative positions, detected signal strength values, and detected phase values.
  10. Wireless power system (1) according to claim 9, wherein the predetermined relative positions are based on two- or three-dimensional Cartesian coordinates representing a distance between the first unit (20) and the second unit (10).
  11. Wireless power system (1) according to any one of the foregoing claims, wherein the at least one trained machine learning model (200, 204 - 206) is respectively a trained neural network, especially a trained feedforward neural network.
  12. Wireless power system (1) according to claim 11, wherein each of the at least one trained neural networks (200, 204 - 206) comprises a set of neuron-parameters (202).
  13. Power transmission system (100), comprising the wireless power system (1) according to any one of the foregoing claims and the second unit (10).
  14. Method for training a machine learning model, especially for training the machine learning model (200, 204 - 206) of the foregoing claims, for determining a relative position between a first unit (20) of a wireless power system (1) and a second unit (10), the first unit (20) and the second unit (10) comprising a first wireless communication unit (22) and a second wireless communication unit (12), respectively, the method comprising: • placing (S1) the first unit (20) and/or the second unit (10) on a positioning jig; • a first iteration of: • recording (S2) a first relative position of the first unit (20) and/or the second unit (10); • transmitting (S3) a communication signal between the first unit (20) and the second unit (10); • detecting (S4) at least one first phase value and/or at least one first signal strength value of the communication signal received by the first unit (20) or the second unit (10) at the first relative position; and at least a second iteration of: • changing (S1'), via the positioning jig, the relative position to a second relative position and repeating said recording (S2), transmission (S3), and detecting (S4) steps, and • inputting (S5) of the recorded relative position(s) and the detected phase value(s) and/or the detected signal strength value(s) as a set of training data points for input parameters (201) of the machine learning model (200, 204 - 206).
  15. Wireless power system according to any one of claims 1 to 12, wherein the trained machine learning model (200, 204 - 206) is trained by the method of claim 14.

Description

Field of the Invention The invention concerns a wireless power system and a method for training a machine learning model for determining a relative position between a first unit and a second unit of a wireless power system. Background of the Invention The use of computer-implemented methods in a wireless power system for determining a relative position between a first unit and a second unit of a wireless power system are known from for example US 2020/0353832 A1, US 2019/0381891 A1, and US 2023/0131879 A1. Furthermore, from US 2020/0339000 A1, a vehicle positioning system is known in which a ground assembly for the vehicle positioning system includes a first ground assembly at antenna, wherein the ground assembly is configured to function together with a vehicle assembly of the vehicle positioning system to determine a position of the vehicle assembly relative to the ground assembly, and wherein a determination of the position is performed by utilizing a neural network. In some cases, for example in US 2020/0353832 A1, positioning is based on detecting an environment of the wireless power system, for example with cameras. In US 2020/0339000 A1, a neural network is trained using signals from magnetic antennas. The magnetic antennas are provided in addition to RF antennas which are used for communication between the units (assemblies) thereof. However, the known solutions have multiple drawbacks. For one, systems based on environment detection for example with the use of cameras require costly components, and commonly achieve only a low accuracy. Furthermore, the use of magnetic antennas additional to a communication antenna increase the component costs, the size and weight of the devices, and increase the complexity of computing the corresponding signals. Summary It is an object of the present invention to overcome these deficiencies. In particular, it is an object of the present invention to provide a wireless power system and a method for training a machine learning model for determining a relative position between a first unit and a second unit of a wireless power system, wherein positioning can be carried out using low-cost components, in a wide range of applications and environments as well as weather conditions, and with high accuracy. The solution of these objects is achieved by the subject matter of the independent claims. The dependent claims contain advantageous embodiments of the present invention. In particular, the solution of these objects is achieved by the wireless power system according to claim 1. The wireless power system comprises a first unit including a pad and a wireless communication unit, wherein the pad is configured to wirelessly transmit or receive power to or from a second unit. Therein, the wireless communication unit is configured to receive, or transmit and receive, a communication signal from the second unit. Furthermore, the wireless power system comprises a control unit. The control unit comprises an input unit configured to receive the communication signal from the wireless communication unit. The control unit comprises a detection unit configured to detect at least one signal strength value and/or at least one phase value of the communication signal. The control unit furthermore comprises at least one trained machine-learning model programmed to determine a relative position between the first unit and the second unit based at least on the detected signal strength value(s) and/or the detected phase value(s) as input parameters. The control unit also comprises an output unit configured to output the relative position. Thereby, the wireless power system of the present invention utilizes the communication signal transmitted from a second unit to the first unit for determining the relative position between the two units. Since determining the relative position is not (solely) based on detecting the environment of the wireless power system and furthermore does not necessitate additional power transfer coils or magnetic antennas, the component costs are greatly reduced. Furthermore, determining the relative position based on the communication signal achieves greatly enhanced accuracy, especially over the use of power transfer coils or magnetic antennas for position detection. In some examples, the wireless power system additionally uses power transfer coils or magnetic antennas for relative position determination, i.e. in addition to using the communication signal. In other words, the control unit is configured to determine the relative position or distance of the wireless communication unit, which is included in the first unit of the wireless power system, to an external second unit. For example, the first unit may be a reception unit (secondary side unit), and the control unit is configured to determine its alignment with the transmission unit (primary side unit). In some examples, the first unit is comprised by a vehicle, whereas the second unit is comprised by an external gr