EMI energy mitigation

Patent metadata

Field	Value
Publication number	US-10998899-B2
Application number	US-201916700253-A
Country	US
Kind code	B2
Filing date	Dec 2, 2019
Priority date	Oct 30, 2017
Publication date	May 4, 2021
Grant date	May 4, 2021

Abstract

Various examples related to electromagnetic interference (EMI) energy mitigation techniques are provided. In one example, a method includes determining electromagnetic interference (EMI) spectrum information based upon switching angles of a switching circuit and processing harmonic magnitudes (C i ) associated with the switching angles using an artificial neural network to determine adjusted switching angles for the switching circuit; and applying the adjusted switching angles to control the switching circuit thereby reducing generated EMI energy by the switching circuit.

First claim

Therefore, at least the following is claimed:

1. A method for electromagnetic interference energy mitigation, comprising:
determining electromagnetic interference (EMI) spectrum information based upon switching angles of a switching circuit, the EMI spectrum information comprising harmonic magnitudes (C i ) associated with the switching angles;
processing the harmonic magnitudes (C i ) using an artificial neural network to determine adjusted switching angles for the switching circuit based on resolution bandwidths (RBWs) associated with the electromagnetic interference (EMI) spectrum of the switching circuit and weight coefficients of harmonics associated with the RBWs; and
applying the adjusted switching angles to control the switching circuit thereby reducing generated EMI energy by the switching circuit.

2. The method of claim 1 , wherein only one significant harmonic of an EMI frequency spectrum is within each RBW.

3. The method of claim 2 , wherein each RBW is associated with weight coefficients comprising a middle weight coefficient corresponding to one significant harmonic in the RBW and other weight coefficients, and the middle weight coefficient is smaller than the other weight coefficients in the RBW to reduce the one significant harmonic.

4. The method of claim 1 , comprising determining training data for the artificial neural network comprising a first set of operational harmonic magnitudes associated with first adjusted switching angles, wherein the first adjusted switching angles are adjusted in response to a comparison of a maximum weighted error and a preset tolerance threshold, the maximum weighted error determined from errors between reference harmonic magnitudes (C i *) associated with the RBWs and the operational harmonic magnitudes (C i ) determined using the switching angles of the switching circuit, where the errors between C i and C i * are weighted by corresponding weight coefficients of the RBWs.

5. The method of claim 4 , comprising determining revised adjusted switching angles for the switching circuit in response to the maximum weighted error exceeding the preset tolerance threshold, wherein at least a portion of the errors between C i and C i * are reduced by the revised adjusted switching angles.

6. The method of claim 5 , wherein the RBWs are defined based upon an EMI standard limiting the harmonics associated with the RBWs.

7. The method of claim 5 , comprising training the artificial neural network using the training data to generate new adjusted switching angles for the switching circuit.

8. The method of claim 1 , comprising applying a DC offset to a modulation waveform to change an average duty cycle of the switching circuit.

9. The method of claim 1 , wherein the switching angles correspond to rise and fall times of switches in the switching circuit.

10. A system, comprising:
a switching circuit comprising an array of semiconductor switches that control application of a voltage source to a load;
controller circuitry configured to control switching of the array of semiconductor switches by adjusting switching angles of the switching circuit based on resolution bandwidths (RBWs) associated with an electromagnetic interference (EMI) frequency spectrum of the switching circuit and weight coefficients of the RBWs; and
the controller circuitry further configured to process, using an artificial neural network, harmonic magnitudes (C i ) associated with the switching angles to determine the adjusted switching angles for the switching circuit based on the resolution bandwidths (RBWs) associated with the electromagnetic interference (EMI) frequency spectrum of the switching circuit and the weight coefficients of the RBW.

11. The system of claim 10 , wherein the RBWs are associated with pass-bands of intermediate frequency (IF) filters.

12. The system of claim 11 , comprising EMI filters having stop bands corresponding to one or more of the RBWs.

13. The system of claim 11 , wherein the controller circuitry is configured to process training data using the artificial neural network to generate new adjusted switching angles, wherein the training data comprises a first set of operational harmonic magnitudes (C i ) associated with first adjusted switching angles, wherein the first adjusted switching angles are adjusted in response to a comparison of a maximum weighted error and a preset tolerance threshold, the maximum weighted error determined from errors between reference harmonic magnitudes (C i *) associated with the RBWs and the operational harmonic magnitudes (C i ) determined using the switching angles of the switching circuit, where the errors between C i and C i * are weighted by corresponding weight coefficients of the RBWs.

14. The system of claim 10 , wherein the switching angles correspond to rise and fall times of switches in the switching circuit.

15. The system of claim 10 , wherein the controller circuitry is configured to adjust an average duty cycle of the switching circuit to control a total energy of the switching circuit.

16. The system of claim 11 , wherein the RBWs are defined based upon an EMI standard limiting the harmonics associated with the RBWs.

17. A method for electromagnetic interference energy mitigation, comprising:
determining electromagnetic interference (EMI) spectrum information based upon switching angles of a switching circuit, the EMI spectrum information comprising determined harmonic magnitudes (C i ) associated with the switching angles;
determining weighted errors corresponding to differences between reference harmonic magnitudes (C i *) and the determined harmonic magnitudes (C i );
in response to a comparison of a maximum weighted error of the weighted errors to a preset tolerance threshold, determining adjusted switching angles for the switching circuit,
processing, using an artificial neural network, training data to determine new adjusted switching angles for the switching circuit, wherein the training data comprises harmonic magnitudes associated with the determined adjusted switching angles; and
applying the new adjusted switching angles to control the switching circuit thereby reducing generated EMI energy by the switching circuit.

18. The method of claim 17 , wherein the weighted errors are determined by weighting errors between a reference harmonic magnitude and a corresponding determined harmonic magnitude with a corresponding weight coefficient.

19. The method of claim 17 , wherein the switching angles correspond to rise and fall times of switches in the switching circuit.

20. The method of claim 17 , comprising adjusting an average duty cycle of the switching circuit to control a total energy of the switching circuit.

Assignees

• Univ Florida

Inventors

• Wang Shuo
• Zhao Hui
• Yang Le

Classifications

• G06N3/0499

  Feedforward networks · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/126

  Evolutionary algorithms, e.g. genetic algorithms or genetic programming · CPC title

• G06N3/006

  based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO] · CPC title

• H03K5/003

  Changing the DC level (reinsertion of DC component of a television signal H04N5/16) · CPC title