Methods and systems for non-destructive evaluation of stator insulation condition

Patent metadata

Field	Value
Publication number	US-12437415-B2
Application number	US-202217937644-A
Country	US
Kind code	B2
Filing date	Oct 3, 2022
Priority date	Oct 3, 2022
Publication date	Oct 7, 2025
Grant date	Oct 7, 2025

Abstract

Methods and systems are provided for an insulation system of a stator. In one example, a method may include receiving images of the stator at an automated tool implemented at a processor of a computing system, the images depicting slots in in an inner surface of the stator, and processing the images using image processing and deep learning algorithms to provide processed images. The processed images may be input to an artificial intelligence (AI) model of the automated tool, the AI model trained to identify varnish in the processed images. obtained two-dimensional presentation of varnish distribution may be output from the AI model.

First claim

The invention claimed is:

1. A method of evaluating a varnish condition of a stator, comprising:
illuminating an inner surface of the stator using a UV light source, the inner surface including slots;
obtaining a fluorescence image of the inner surface via digital imaging equipment and transmitting the fluorescence image to a processor;
segmenting the fluorescence image into slot profiles, using image processing algorithms implemented at the processor, by aligning the fluorescence image with a photograph of the inner surface of the stator and dividing the fluorescence image into segments, each segment corresponding to a slot of the slots;
constructing a fluorescence signature for the slot, via a machine learning model implemented at the processor, by deconstructing a segment corresponding to the slot at a pixel-scale resolution and matching pixels of the segment to pixels of ground truth images;
outputting a two-dimensional (2D) representation of the fluorescence signature of the slot from the machine learning model; and
displaying the 2D representation at a display device.

2. The method of claim 1 , wherein the fluorescence image is obtained without cutting the stator, and wherein the stator remains intact after the 2D representation is displayed.

3. The method of claim 1 , wherein further comprising resizing the fluorescence image prior to segmenting the fluorescence image, and wherein resizing the fluorescence image includes adjusting a scale of the fluorescence image to a first value representing a width of a corresponding slot of the slots, and to a second value representing a length of the corresponding slot.

4. The method of claim 3 , wherein segmenting the fluorescence image includes slicing the fluorescence image into a number of the slot profiles equal to the first value.

5. The method of claim 1 , wherein deconstructing the segment includes comparing the pixels of the segment to the pixels of the ground truth images and minimizing loss between the pixels of the segment and the pixels of the ground truth images.

6. The method of claim 5 , wherein minimizing loss includes generating an algorithm based on pixel pairs generated by the matching of the pixels of the segment to the pixels of the ground truth images to determine loss between the pixel pairs and updating the machine learning model.

7. The method of claim 6 , wherein outputting the 2D representation includes outputting predicted pixels corresponding to the pixel pairs and assembling the predicted pixels into a reconstruction of the segment.

8. A system for evaluating a varnish condition of a stator, comprising:
a housing enclosing a UV light source and digital imaging equipment; and
a processor configured with executable instructions stored in non-transitory memory that, when executed, cause the processor to:
receive images of an inner surface of the stator from the digital imaging equipment;
process the images using deep learning algorithms by segmenting and aligning the images to generate segmented images;
input the segmented images to a machine learning model trained to identify varnish in the segmented images based on color distribution analysis;
deconstruct the images at a pixel-scale resolution to construct predicted images from the segmented images; and
display the predicted images, each of the predicted images corresponding to a slot of the stator, at a display device.

9. The system of claim 8 , wherein, prior to segmenting and aligning the images, the images are modified by applying image parameter adjustments intrinsic to the digital imaging equipment, and wherein the image parameter adjustments includes removing skew and a fish-eye lens effect from the images.

10. The system of claim 8 , wherein each of the predicted images show a fluorescence signature of the slot indicating regions corresponding to varnish based on a color distribution of the predicted images.

Assignees

• Ford Global Tech Llc

Inventors

• Schroeter Robert
• Lenz Jackson
• Tan Boratha
• Siffre Ife

Classifications

• G06T7/38

  Registration of image sequences · CPC title

• G06V10/26

  Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion · CPC title

• G06T2207/10064

  Fluorescence image · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title

• G06T2207/30164

  Workpiece; Machine component · CPC title