Feedback control systems and methods for glass tube converting processes

Patent metadata

Field	Value
Publication number	US-12459852-B2
Application number	US-202217885210-A
Country	US
Kind code	B2
Filing date	Aug 10, 2022
Priority date	May 24, 2021
Publication date	Nov 4, 2025
Grant date	Nov 4, 2025

Abstract

Methods for providing feedback control of converters for converting glass tubes to glass articles include a model predictive control framework. The methods include operating the converter, providing target values for attributes of the glass articles or glass tubes, measuring the attributes for the glass articles and glass tubes, conditioning the measurement data to remove outlier data points and calculating statistics representative of the measured attributes, and determine updated settings for one or more process parameters from the previous settings, the statistical properties, and the target values, where the updated settings are those that minimize an objective control function for the converter. The methods further include adjusting the process parameters to the updated settings. The model predictive control framework enables feedback control of the converter that compensates for disturbances that act on the process.

First claim

What is claimed is:

1 . A method for controlling a converter for producing glass articles from glass tubes, the method comprising:
operating a converter to produce a plurality of glass articles from a plurality of glass tubes, where the converter comprises a plurality of processing stations and operating the converter comprises translating the glass tubes through each of the plurality of processing stations in succession; providing target values for at least one attribute of the plurality of glass articles or the plurality of glass tubes during or after converting; measuring the at least one attribute of the plurality of glass articles or the plurality of glass tubes during or after converting; recording settings of at least one process parameter of the converter related to the at least one attribute to produce a data set comprising measured values of the at least one attribute and the settings of the at least one process parameter of the converter; processing the data set to produce a statistical property of a distribution of the measured values of the at least one attribute; determining an updated setting for each of the at least one process parameter from the statistical property of the distribution of the at least one attribute measured, the target values of the at least one attribute, and the settings of the at least one process parameter, wherein the updated settings of the at least one process parameter is a value of the settings that minimizes an objective control function for the at least one attribute; adjusting each of the at least one process parameter of the converter to the updated setting; and wherein the objective control function comprises a mean square error cost function according to the following equation:
J ( k )=( G *Act( k )+Attrib measured ( k )− G *Act( k− 1)−Attrib targ ) T Q T Q ( G *Act( k )+Attrib measured ( k )−( G *Act( k− 1))−Attrib targ )+(Act( k )−Act( k− 1)) T R T R (Act( k )−Act( k− 1));

where:
J(k) is the mean square error cost function as a function of k:
k is an integer indicative of a present iteration of minimizing the mean square error cost function;
G is a matrix of sensitivity factors representative of a degree to which a change in each of the at least one process parameter produces a change each of the at least one attribute;
Attrib measured (k) is a vector of the statistical property of the distribution of the measured values of the at least one attribute during iteration k;
Act(k) is a setting of the process parameter at iteration k;
Act(k−1) is a setting of the process parameter iteration k−1;
Attrib targ is a vector of the target values for each of the at least one attribute;
Q T Q is a symmetric weighting matrix of attribute weighting factors for errors in measured values of the attributes from target values of the attributes; and
R T R is a symmetric weighting matrix of penalty factors on the change in the at least one process parameter.

2 . The method of claim 1 , further comprising repeating the measuring the at least one attribute, the recording the setting for each of the at least one process parameter, the processing the data set, the determining the updated setting for each of the at least one process parameter, and the adjusting each of the at least one process parameter until the updated setting for each of the at least one process parameter converges.

3 . The method of claim 1 , wherein processing the data set comprises:
removing outlier data points from the data set of the measured values of the at least one attribute; after removing the outlier data points, calculating the statistical property for the distribution of the measured values of the at least one attributes from the data set.

4 . The method of claim 3 , wherein the statistical property of the distribution of the data set is a mean, a median, a range, a standard deviation, a variance, or combinations of these.

5 . The method of claim 1 , further comprising:
providing a specification range for each of the at least one attribute, wherein the specification range of an attribute comprises:
a minimum value of the attribute below which the glass article is considered out of specification; and
a maximum value of the attribute above which the glass article is considered out of specification; and
applying an attribute weighting factor to each of the at least one attribute in the objective control function based on a spread in the specification range of each of the at least one attribute.

6 . The method of claim 5 , further comprising determining the attribute weighting factor from the specification range for each of the at least one attribute.

7 . The method of claim 6 , comprising determining a process capability index C pk of each of the at least one attribute from the specification range, and determining the attribute weighting factor for each of the at least one attribute based on the process capability index C pk of the at least one attribute.

8 . The method of claim 1 , further comprising:
developing a penalty factor for each of the at least one process parameter; and applying the penalty factor to each of the at least one process parameter in the objective control function, where the penalty factor operates to reduce the magnitude of changes to process parameters that have greater impact on one or more of the at least one attributes.

9 . The method of claim 8 , further comprising:
repeating the method of claim 8 for a plurality of iterations; identifying divergence or oscillation of the updated setting for one or more of the at least one process parameter indicating a reduced ability to control one or more aspects of the converter; and adjusting the penalty factor for one or more of the at least one process parameter, where adjusting the penalty factor reduces a magnitude of changes made to the one or more of the at least one process parameters in each of the plurality of iterations, thereby reducing divergence or oscillation of the updated setting.

10 . The method of 9 , wherein the divergence is indicated by oscillations in or consistent increases in the objective control function per iteration of the method.

11 . The method of claim 1 , further comprising:
providing a maximum setting and a minimum setting of each of the at least one process parameter; and maintaining the updated setting for each of the at least one process parameter in a range between the minimum setting and the maximum setting for the process parameter.

12 . The method of claim 1 , further comprising developing the objective control function.

13 . The method of claim 12 , wherein developing the objective control function comprises:
developing at least one model relating a predicted value of the at least one attribute for each setting of the at least one process parameter, wherein the at least one model comprises an expression in which the predicted value of the at least one attribute is a sum of at least one term dependent on the setting of the at least one process parameter and an offset constant; providing an initial mean square error cost function that is a function of the predicted values for each of the at least one attribute, the target values for each of the at least one attribute, and the settings for each of the at least one process parameter; substituting the at least one model into the initial mean square error cost function for the predicted value of the at least one attribute; solving the at least one model for the offset constant to produce an offset constant function, where solving the at least one model for the offset constant comprises substituting th

Assignees

• Corning Inc

Inventors

• Barnard Christelle Ruby
• Gross Eric Michael
• Matusick Joseph Michael
• Mcnelis Kevin Patrick
• O'Malley Connor Thomas

Classifications

• C03B23/09

  Reshaping the ends, e.g. as grooves, threads or mouths · CPC title

• G05B19/4155

  characterised by program execution, i.e. part program or machine function execution, e.g. selection of a program · CPC title

• G05B2219/45009

  Glassforming · CPC title

• C03B23/045

  Tools or apparatus specially adapted for re-forming tubes or rods in general, e.g. glass lathes, chucks (C03B23/043 takes precedence) · CPC title

• C03B23/095

  by rolling · CPC title