Method and apparatus for automatic localization of a fault

Patent metadata

Field	Value
Publication number	US-10955456-B2
Application number	US-201816139669-A
Country	US
Kind code	B2
Filing date	Sep 24, 2018
Priority date	Sep 26, 2017
Publication date	Mar 23, 2021
Grant date	Mar 23, 2021

Abstract

Provided is an in-field apparatus and method for automatic localization of a fault having occurred at power transmission lines of a power supply system, the in-field apparatus includes a preprocessing unit configured to process measured voltage and/or current raw time series data of the power transmission lines to provide a normalized raw data and/or feature representation of the measured raw time series data, and an artificial intelligence module configured to predict an optimal evaluation time used for evaluation of the measured voltage and/or current raw time series data to localize the fault based on the normalized raw data and/or feature representation.

First claim

The invention claimed is:

1. An in-field apparatus for automatic localization of a fault having occurred at power transmission lines of a power supply system, the in-field apparatus comprising:
(a) a preprocessing unit configured to process measured voltage and/or current raw time series data of the power transmission lines to provide a normalized raw data and/or feature representation of the measured raw time series data; and
(b) an artificial intelligence module configured to predict an optimal evaluation time used for evaluation of the measured voltage and/or current raw time series data to localize the fault based on the normalized raw data and/or feature representations;
wherein the artificial intelligence module comprises:
a window detection unit configured to determine an optimal window size of a window used for feature extraction from the measured voltage and/or current raw time series data; and
a time point detection unit configured to determine an optimal evaluation time point for the evaluation of features extracted from the measured voltage and/or current raw time series data.

2. The in-field apparatus according to claim 1 , wherein the preprocessing unit is configured to perform a z-score normalization of measured voltage and/or current raw data of the power transmission lines.

3. The in-field apparatus according to claim 1 , wherein the window detection unit is configured to generate learned regression curves for different predefined window sizes.

4. The in-field apparatus according to claim 1 , wherein the artificial intelligence module comprises:
a false positive detection unit configured to determine a probability whether the predicted optimal evaluation time does lead to a valid fault localization.

5. The in-field apparatus according to claim 1 , wherein the artificial intelligence module is configured to perform a feature-based fault classification based on the feature representation of the preprocessed measured raw time series data provided by said preprocessing unit.

6. The in-field apparatus according to claim 5 , wherein the artificial intelligence module is configured to perform the feature-based fault classification using a random forest, a decision tree, a support vector machine, and/or a fully connected neural network.

7. The in-field apparatus according to claim 1 , wherein the artificial intelligence module is configured to perform a not feature-based direct fault classification based on the normalized measured raw time series data provided by the preprocessing unit.

8. The in-field apparatus according to claim 7 , wherein the artificial intelligence module is configured to perform the not feature-based direct fault classification using a feed-forward neural network or a recurrent neural network, RNN.

9. The in-field detection apparatus according to claim 1 , wherein the preprocessing unit is configured to normalize raw data of a measured fault record comprising:
voltage raw data of different voltage phases transported through corresponding power transmission lines,
current raw data of associated electrical currents transported through the respective power transmission lines, and
current raw data of an electrical current flowing through earth.

10. The in-field detection apparatus according to claim 9 , wherein the voltage and/or current raw data of the measured fault record processed by the preprocessing unit comprises data samples measured by corresponding sensors within a time window.

11. The in-field detection apparatus according to claim 1 , wherein a database or an online stream of fault records with known fault types is provided and used to train the artificial intelligence module of the in-field detection apparatus.

12. The in-field detection apparatus according to claim 11 , wherein the database of fault records comprises simulated fault records and/or fault records of historical faults having occurred at power transmission lines of said power supply system.

13. The in-field detection apparatus according to claim 1 , wherein countermeasures to remove a fault are initiated automatically by a controller depending on a detected fault type of the localized fault.

14. A method for automatic localization of a fault having occurred at power transmission lines of a power supply system, the method comprising the steps of:
(a) preprocessing measured voltage and/or current raw time series data of the power transmission lines to provide a normalized raw data and/or feature representation of the raw time series data;
(b) calculating an optimal evaluation time point and/or evaluation time window based on the normalized raw data and/or feature representation using a trained artificial intelligence module;
(c) evaluating by the trained artificial intelligence module the measured voltage and/or current raw time series data using the calculated optimal evaluation time point and/or evaluation time window to localize the fault; and
(d) performing by the trained artificial intelligence module a feature-based fault classification based on the feature representation of the preprocessed raw time series data.

15. The method according to claim 14 , wherein the feature-based fault classification is performed using a random forest, a decision tree, a support vector machine, and/or a fully connected neural network.

16. An apparatus for automatic localization of a fault in a power supply system, the apparatus comprising:
(a) a preprocessing unit configured to process measured voltage and/or current raw time series data of the power transmission lines to provide a normalized raw data and/or feature representation of the measured raw time series data; and
(b) an artificial intelligence module configured to predict an optimal evaluation time used for evaluation of the measured voltage and/or current raw time series data to localize the fault based on the normalized raw data and/or feature representation, wherein the artificial intelligence module is configured to perform a feature-based fault classification based on the feature representation of the preprocessed measured raw time series data provided by said preprocessing unit.

17. The apparatus according to claim 16 , wherein the artificial intelligence module is configured to perform the feature-based fault classification using a random forest, a decision tree, a support vector machine, and/or a fully connected neural network.

Assignees

• Siemens Ag

Inventors

• Becher Silvio
• Buggenthin Felix
• Böhme Klaus
• Büttner Florian
• Kereit Matthias
• Kogan Igor

Classifications

• G06N5/01

  Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound · CPC title

• G06N3/045

  Combinations of networks · CPC title

• G06N3/044

  Recurrent networks, e.g. Hopfield networks · CPC title

• G06N3/08

  Learning methods · CPC title

• G06N3/09

  Supervised learning · CPC title