Back to results
US 12,613,139 B2Granted
System for monitoring a switchgear
AssigneeABB Schweiz AG
InventorsRalf Gitzel, Marcel Dix, Holger Kaul, Tomas Kozel
Life of this patent
Estimating the term…
Abstract
A system and method for monitoring a switchgear includes an infrared camera, a processing unit, and an output unit. The infrared camera acquires an infrared image of the switchgear, and the processing unit converts it into a binary image. Pixels in the infrared image having a temperature equal to or above a first threshold value are given the same first value. Pixels in the infrared image having a temperature below the first threshold value are given the same second value. The processing unit is configured to implement a Siamese neural network to determine if a hot spot exists in the infrared image.
Claims14
A system for monitoring a switchgear, the system comprising:
- an infrared camera;
- a processing unit;
- and
- an output unit;
- wherein, the infrared camera is configured to acquire an infrared image of the switchgear;
- wherein the processing unit is configured to convert the infrared image into a binary image, wherein pixels in the infrared image having a temperature equal to or above a first threshold value are given the same first value and wherein pixels in the infrared image having a temperature below the first threshold value are given the same second value;
- wherein the processing unit is configured to implement a Siamese neural network to determine if a hot spot exists in the infrared image, wherein the determination comprises analysis by the Siamese neural network of the binary image and a reference binary image, and wherein the reference binary image was generated from a reference infrared image of the switchgear or of a reference switchgear, and wherein pixels in the reference infrared image having a temperature equal to or above a second threshold value were given the same first value and wherein pixels in the reference infrared image having a temperature below the second threshold value were given the same second value to generate the reference binary image;
- wherein the output unit is configured to output an indication of a fault in the switchgear on the basis that a hot spot has been determined to exist in the infrared image;
- and
- wherein the Siamese neural network is a trained Siamese neural network, wherein training of the Siamese neural network comprises utilization of a plurality of synthetic binary images, wherein the plurality of synthetic binary images are derived from a real infrared image through random noise being added to a binary image generated from the real infrared image, and wherein the binary image is a black and white mask.
7 more elements
A method for monitoring a switchgear, the method comprising:
- acquiring with an infrared camera an infrared image of the switchgear;
- converting by a processing unit the infrared image into a binary image, wherein pixels in the infrared image having a temperature equal to or above a first threshold value are given the same first value and wherein pixels in the infrared image having a temperature below the first threshold value are given the same second value;
- implementing by the processing unit a Siamese neural network and determining if a hot spot exists in the infrared image, wherein the determining comprises analyzing by the Siamese neural network the binary image and a reference binary image, and wherein the reference binary image was generated from a reference infrared image of the switchgear or of a reference switchgear, and wherein pixels in the reference infrared image having a temperature equal to or above a second threshold value were given the same first value and wherein pixels in the reference infrared image having a temperature below the second threshold value were given the same second value to generate the reference binary image;
- and
- outputting by an output unit an indication of a fault in the switchgear on the basis that a hot spot has been determined to exist in the infrared image,
- wherein the Siamese neural network is a trained Siamese neural network, wherein training of the Siamese neural network comprises utilization of a plurality of synthetic binary images, wherein the plurality of synthetic binary images are derived from a real infrared image through random noise being added to a binary image generated from the real infrared image, and wherein the binary image is a black and white mask.
3 more elements
Dependencies are inferred from the claim text.