Granted Patent B2

US 12,379,717 · App. 18/600,258 · Granted Aug 5, 2025

Inventors: Slawek Frackowiak (Fairfax, VA); Leszek Frackowiak (Vienna, VA)

Assignee: Industrial Video Solutions Inc.

G05B19/41875G05B2219/32368

Abstract

Claims (90)

1. A computer-implemented data processing method for identifying and classifying defects in a paper web produced by a manufacturing process, the method comprising:

receiving, by computing hardware, current thermal imaging data for a paper web;

analyzing, by the computing hardware, the thermal imaging data to identify a first portion of the paper web having a temperature that deviates from a surrounding portion of the paper web, the first portion comprising: at least one hot spot, at least one hot streak, at least one cold spot, or at least one cold streak on the paper web;

determining, by the computing hardware for the first portion of the paper web, a set of properties for the first portion of the paper web, the set of properties comprising

a cross-direction location of the first portion having the temperature deviation with respect to an edge of the paper web;

processing, by the computing hardware, the set of properties for the first portion of the paper web using at least one of a rules-based model, a machine-learning model, or a classification model to produce a defect classification for the first portion of the paper web;

generating, by the computing hardware based on the defect classification, a graphical user interface comprising defect map for the paper web, the defect map comprising at least the first portion of the paper web and a visual indication of the defect classification;

providing, by the computing hardware, the graphical user interface for display on a computing device; and

providing, by the computing hardware, the defect classification to at least one of a second machine-learning model or a second rules-based model to cause the second machine-learning model or the second rules-based model to generate a prediction as to an occurrence of a breakage event on the first portion of the paper web based on the defect classification.

2. The method of claim 1 , further comprising:

receiving, by the computing hardware via the graphical user interface, feedback data related to the defect classification; and

responsive to receiving the feedback data, providing the feedback data as training data for one or more of the rules-based model, the machine-learning model, or the classification model.

3. The method of claim 1 , further comprising:

determining, by the computing hardware based on the defect classification, a manufacturing risk for the manufacturing process;

identifying, by the computing hardware based on the manufacturing risk, a corrective action to take in response to the manufacturing risk; and

performing, by the computing hardware, at least a portion of the corrective action.

4. The method of claim 1 , further comprising:

analyzing, by the computing hardware, the thermal imaging data to identify a second portion of the paper web having a second temperature that deviates from the surrounding portion of the paper web;

determining, by the computing hardware for the second portion of the paper web, a second set of properties for the second portion of the paper web, the second set of properties comprising one or more of:

second location data for the second portion of the paper web;

a second temperature of the second portion;

a second shape of the second portion of the paper web;

a second edge uniformity of the second portion of the paper web; or

a second temperature deviation of the second portion with respect to the surrounding portion of the paper web;

processing, by the computing hardware, the second set of properties for the second portion of the paper web using at least one of the rules-based model, the machine-learning model, or the classification model to produce a second defect classification for the second portion of the paper web; and

providing, by the computing hardware, an indication of the second defect classification on a computing device.

5. A system comprising:

a non-transitory computer-readable medium storing instructions; and

a processing device communicatively coupled to the non-transitory computer-readable medium, wherein the processing device is configured to execute the instructions and thereby perform operations comprising:

determining temperature data for at least part of an article of manufacture in a manufacturing process;

analyzing the temperature data to identify a first portion of the article of manufacture having a temperature that deviates from a surrounding portion of the first portion, the first portion comprising: at least one hot spot, at least one hot streak, at least one cold spot, or at least one cold streak on the article of manufacture;

determining, for the first portion of the article of manufacture, a set of properties comprising one or more of:

a temperature of the first portion of the article of manufacture; or

a material type of the article of manufacture;

processing the set of properties for the first portion of the article of manufacture using at least one of a rules-based model, a machine-learning model, or a classification model to produce a defect classification for the first portion of the article of manufacture, the defect classification comprising an identification of a contaminant in the article of manufacture having a thermal property that differs from a thermal property of the article of manufacture; and

providing an indication of the defect classification for display on a computing device; and

providing the defect classification to at least one of a second machine-learning model or a second rules-based model to cause the second machine-learning model or the second rules-based model to generate a prediction as to an occurrence of a breakage event on the first portion of the article of manufacture based on the defect classification.

6. The system of claim 5 , wherein the article of manufacture comprises at least one of:

a paper web;

a non-paper web; or

an extruded material.

7. The system of claim 5 , wherein analyzing the temperature data to identify the first portion of the article of manufacture having a temperature that deviates from a surrounding portion of the first portion comprises:

determining an expected temperature profile for the article of manufacture based on the material type of the article of manufacture; and

identifying the first portion of the article of manufacture having the temperature that deviates from the surrounding portion of the first portion comprises identifying the first portion as an impurity that has a temperature other than the expected temperature.

8. The system of claim 5 , wherein determining the temperature data for the at least part of the article of manufacture in the manufacturing process comprises determining the temperature at at least one of:

during the manufacturing process; or

substantially immediately following the manufacturing process.

9. A computer-implemented data processing method for identifying and classifying potential defects in an article of manufacture, the method comprising:

determining temperature data from thermal imaging data for at least part of an article of manufacture in a manufacturing process;

analyzing, by computing hardware, the temperature data to identify a first portion of the article of manufacture having a temperature that deviates from an expected temperature;

determining, by the computing hardware for the first portion of the article of manufacture, a set of properties for the first portion of the article of manufacture, the set of properties comprising one or more of:

location data for the first portion of the article of manufacture with respect to the article of manufacture;

an absolute temperature of the article of manufacture;

a shape of the first portion of the article of manufacture; or

a temperature deviation of the first portion with respect to the expected temperature;

processing, by the computing hardware, the set of properties for the first portion of the article of manufacture using at least one of a rules-based model, a machine-learning model, or a classification model to produce a defect classification for the first portion of the article of manufacture;

providing, by the computing hardware, an indication of the defect classification for display on a computing device;

providing, by the computing hardware, the defect classification to at least one of a second machine-learning model or a second rules-based model to cause the second machine-learning model or the second rules-based model to generate a prediction as to an occurrence of a breakage event on the first portion of the article of manufacture based on the defect classification.

10. The method of claim 9 , wherein:

the expected temperature is determined based on one or more material properties of the article of manufacture; and

processing the set of properties for the first portion of the article of manufacture using at least one of the rules-based model, the machine-learning model, or the classification model to produce the defect classification for the first portion of the article of manufacture further comprises processing the one or more material properties of the article of manufacture with the set of properties to produce the defect classification for the first portion of the article of manufacture.

11. The method of claim 9 , wherein the article of manufacture comprises at least one of:

a paper web;

a non-paper web;

an extruded material;

a polymer; or

a metal.

12. The method of claim 9 , wherein:

the defect classification includes an indication of a defect concern level;

the method further comprises:

determining, by the computing hardware, an action based on the defect concern level; and

causing, by the computing hardware, performance of the action.

13. The method of claim 12 , wherein the action comprises

reclassifying at least a portion of the article of manufacture from a first type of classification to a second type of classification.

14. The method of claim 12 , further comprising:

identifying additional portions of the article of manufacture having respective temperatures that deviate from the expected temperature over time; and

generating a graphical user interface comprising a defect map comprising a representation of the article of manufacture and an indication of a respective location of each of the additional portions of the article of manufacture.

15. The method of claim 14 , wherein the indication of the defect classification comprises a trend in defects over time determined based on the identified additional portions of the article of manufacture having the respective temperatures that deviate from the expected temperature over time.

16. The method of claim 9 , wherein the set of properties comprise the location data for the first portion of the article of manufacture with respect to the article of manufacture, the shape of the first portion of the article of manufacture, and the temperature deviation of the first portion with respect to the expected temperature.

17. The method of claim 9 , further comprising:

determining, by the computing hardware, second temperature data from second thermal imaging data for the first portion of the article of manufacture in the manufacturing process; and

analyzing, by the computing hardware, the second temperature data and the defect classification to determine a defect origin for the first portion of the article of manufacture in the manufacturing process.

18. The method of claim 1 , further comprising:

responsive to the prediction of the occurrence of the breakage event, facilitating performance of a preventative action, wherein the preventative action comprises at least one of:

modifying an operating parameter of at least one machine component used in the manufacturing process;

activating a cleaning component for manufacturing process;

activating a siren;

generating an alert and transmitting the alert to a computing device; or

at least temporarily ceasing the manufacturing process.

19. The method of claim 18 , further comprising processing the prediction and historical preventative action success data using at least one of a third rules-based model or a third machine-learning model to select the preventative action.

Assignments (1)

Continuity (6)

References Cited (134)