Feedback control system and method for glass tube conversion process - Patents.com

Abstract

A method for providing feedback control of a conversion unit for converting a glass tube into a glass article includes a model predictive control framework. The method includes operating the conversion unit, providing target values ​​for attributes of the glass article or the glass tube, measuring attributes of the glass article and the glass tube, adjusting the measurement data to exclude outlying data points, calculating statistics representative of the measured attributes, and determining updated settings for one or more process parameters from previous settings, statistical characteristics, and target values, the updated settings minimizing a target control function for the conversion unit. The method further includes adjusting the process parameters to the updated settings. The model predictive control framework enables feedback control of the conversion unit to compensate for disturbances acting on the process.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of priority under 35 U.S.C. §120 to U.S. patent application Ser. No. 17 / 746,396, filed May 17, 2022, and also claims the benefit of priority under 35 U.S.C. §119 to U.S. Provisional Patent Application No. 63 / 344,851, filed May 23, 2022, the contents of each of which are relied upon and incorporated by reference in their entireties herein. [Technical field]

[0002] FIELD OF THE DISCLOSURE The present disclosure relates generally to systems and methods for producing glass articles from glass tubes, and more particularly to systems and methods for feedback controlling the glass tube conversion process. [Background technology]

[0003] Glass has long been a preferred material for containing drugs because of its hermeticity, optical transparency, and excellent chemical durability compared to other materials. In particular, glasses used for containing drugs must have adequate chemical durability to prevent the stability of the drug formulation contained therein from being compromised. Glasses having suitable chemical durability include glass compositions within the ASTM standard "Type IA" and "Type IB" glass compositions that have previously demonstrated chemical durability.

[0004] Glass tubes may be converted into other glass articles, such as various glass containers for pharmaceutical applications, including, but not limited to, vials, syringes, ampoules, cartridges, and other glass articles. Glass tubes may be converted, for example, with a "converter." Converters have been around for over 75 years and are currently manufactured by a variety of commercial and internal equipment suppliers. These converters typically transform long lengths of glass tube into multiple glass articles using processes that include flame working, rotating and stationary forming, heat separation, or scoring and impact cutting processes. Various burners and formers are often used to form one or more articles from the glass tube and separate the articles from the glass tube. Summary of the Invention [Problem to be solved by the invention]

[0005] During the conversion of a glass tube into a glass article using a converter (i.e., converter section), a heating section, such as a burner, heats the glass in the glass tube to a temperature where the viscosity of the glass allows the glass to be formed into one or more features of the glass article. The forming station includes formers, such as pin and wheel assemblies, that contact the heated glass tube to form the inner and outer dimensions of the features of the finished glass article. Conventional converters have used simple needle valves to adjust the burner power at the heating station. The position of the formers in conventional converters has been adjusted by mechanical linkages, often connected to a cam drive shaft. The burner power, burner placement, former position, and former contact timing can affect the dimensional yield and defect rate of the converter section. Additionally, hundreds of other process settings and inputs in a typical converter can affect the dimensional yield and defect rate that the converter can achieve.

[0006] Historically, the control strategy for managing hundreds of process settings and inputs for conventional converters has relied on human operators with varying experience levels to specifically vary burner parameters, former parameters, and overall machine timing for vial geometries of different shapes. Human operators manage the normal day-to-day variations that occur during this process by adjusting these same parameters. A manufacturer's ability to consistently produce high quality glass articles with high yields from a converter is highly dependent on the skill level and experience of the machine operators operating the converter. As the experience levels of operators vary from line to line and shift to shift, this can result in large variations in the yield and quality of vials produced in the process. Despite the development of "add-on" controls and automation, such as mass flow control valves, servo motors, and / or PLC controls, current technology still relies on human operators making decisions that ultimately drive all input parameters for the converters.

[0007] Therefore, there is a need for a system and method that provides feedback control of a conversion section that converts glass tubing into glass articles, such as pharmaceutical containers, to reduce variability in the operation, increase yields, and reduce defect rates. [Means for solving the problem]

[0008] In a first aspect of the present disclosure, a method of controlling a conversion unit for producing glass articles from a glass tube includes operating a conversion unit including a plurality of processing stations to produce a plurality of glass articles from a plurality of glass tubes, and operating the conversion unit may include translating the glass tube through each of the plurality of processing stations in succession. The method may further include providing a target value for at least one attribute of the plurality of glass articles or the plurality of glass tubes during or after conversion, measuring at least one attribute of the plurality of glass articles or the plurality of glass tubes during or after conversion, recording settings of at least one processing parameter of the conversion unit for the at least one attribute to generate a data set including the measured values ​​of the at least one attribute and the settings of the at least one processing parameter of the conversion unit, processing the data set to generate a statistical characteristic of the distribution of the measured values ​​of the at least one attribute, and determining an updated setting of each of the at least one processing parameter from the statistical characteristic of the distribution of the measured at least one attribute, the target value of the at least one attribute, and the setting of the at least one processing parameter. The updated setting of the at least one processing parameter may be a value of the setting that minimizes a target control function for the at least one attribute. The method may further include adjusting each of the at least one processing parameter of the transform unit to the updated setting.

[0009] A second aspect of the present disclosure may include the first aspect and further include repeating the steps of measuring the at least one attribute, recording settings for each of the at least one processing parameter, processing the dataset, determining updated settings for each of the at least one processing parameter, and adjusting each of the at least one processing parameter until the updated settings for each of the at least one processing parameter converge.

[0010] A third aspect of the present disclosure includes the first or second aspects, and the step of processing the dataset may include removing outlying data points from the dataset of measurements of the at least one attribute, and calculating statistical properties of the distribution of the measurements of the at least one attribute from the dataset after removing the outlying data points.

[0011] A fourth aspect of the present disclosure includes the third aspect, wherein the statistical characteristic of the distribution of the dataset can be a mean, a median, a range, a standard deviation, a variance, or a combination thereof.

[0012] A fifth aspect of the present disclosure includes any one of the first to fourth aspects, and may further include providing a specification range for each at least one attribute. The specification range for the attribute may include a minimum value for the attribute below which the glass article is considered out of specification, and a maximum value for the attribute above which the glass article is considered out of specification. The method may further include applying, in the target control function, to each at least one attribute, an attribute weighting factor based on a spread of the specification range for each at least one attribute.

[0013] A sixth aspect of the present disclosure includes the fifth aspect and may further include determining the attribute weighting coefficients from a specification range for each of the at least one attribute.

[0014] A seventh aspect of the present disclosure includes the sixth aspect, further comprising: pk from the specification range, and determining the attribute weighting coefficient for each of at least one attribute as a process capability index C pk and determining based on the

[0015] An eighth aspect of the present disclosure may include any one of the first to seventh aspects, and may further include developing, for each of the at least one processing parameter, a penalty factor; and applying the penalty factor to each of the at least one processing parameter in the target control function, the penalty factor acting to reduce the magnitude of changes to processing parameters that have a greater impact on one or more of the at least one attribute.

[0016] A ninth aspect of the present disclosure includes the eighth aspect and may further include repeating the method of the eighth aspect for a number of iterations, identifying divergence or oscillation of the updated settings for one or more of the at least one processing parameter indicative of reduced performance to control one or more aspects of the transform unit, and adjusting a penalty factor for one or more of the at least one processing parameter. The adjusting penalty factor may reduce the magnitude of the change to the one or more processing parameters in each iteration of the method, thereby reducing the divergence or oscillation of the updated settings.

[0017] A tenth aspect of the present disclosure includes the ninth aspect, wherein the divergence can be indicated by an oscillation or a consistent increase in the target control function from one iteration of the method to the next.

[0018] An eleventh aspect of the present disclosure includes any one of the first to tenth aspects, and may further include providing a maximum setting and a minimum setting for each of the at least one processing parameter, and maintaining the updated setting for each of the at least one processing parameter within a range between the minimum and maximum settings for the processing parameter.

[0019] A twelfth aspect of the present disclosure includes any one of the first to eleventh aspects, wherein the target control function includes a mean squared error cost function according to:

[0020]

number

[0021] where J(k) is the mean squared error cost function as a function of k, where k is an integer indicating the current iteration for minimizing the mean squared error cost function, G is a sensitivity coefficient matrix representing the degree to which a change in each of the at least one process parameter causes a change in each of the at least one attribute, and Attrib measured (k) is a vector of statistical properties of the distribution of measurements of at least one attribute during iteration k, Act(k) is the setting of the processing parameters at iteration k, Act(k-1) is the setting of the processing parameters at iteration k-1, and Attrib targ is a vector of target values ​​for each at least one attribute, and Q T Q is a symmetric weighting matrix of attribute weighting coefficients for the error of attribute measurements from attribute target values, and R T R may be a symmetric weighting matrix of penalty coefficients for variation of at least one process parameter.

[0022] A thirteenth aspect of the present disclosure includes the twelfth aspect and may further include developing a target control function.

[0023] A fourteenth aspect of the present disclosure includes the thirteenth aspect, wherein developing the target control function may include developing at least one model for a predicted value of the at least one attribute for each setting of the at least one process parameter. The at least one model may include an equation in which the predicted value of the at least one attribute may be a sum of at least one term depending on the setting of the at least one process parameter and an offset constant. Developing the target control function may further include providing an initial mean squared error cost function that is a function of the predicted value for each at least one attribute, the target value for each at least one attribute, and the setting for each at least one process parameter, substituting the at least one model into the initial mean squared error cost function for the predicted value of the at least one attribute, and solving the at least one model for the offset constant to generate an offset constant function. Solving the at least one model for the offset constant may include substituting statistical characteristics of measurements of each at least one attribute into the predicted value of each at least one attribute. Developing the target control function may further include substituting an offset constant function for an offset constant of the initial mean squared error cost function to generate the target control function.

[0024] A fifteenth aspect of the present disclosure includes the fourteenth aspect, and wherein the step of developing the at least one model can include a step of performing an experimental design process, a step of deriving the at least one model from first principles, or a combination thereof to generate the at least one model.

[0025] A sixteenth aspect of the present disclosure may include any of the fourteenth or fifteenth aspects and further include periodically redeveloping at least one model to account for processing changes over time or in response to known changes in the operation of the converter.

[0026] A seventeenth aspect of the present disclosure includes any one of the fourteenth to sixteenth aspects, and may further include applying an attribute weighting coefficient for each of the at least one attribute to one or more terms of the target control function.

[0027] An eighteenth aspect of the present disclosure includes any one of the fourteenth to seventeenth aspects and may further include applying a penalty coefficient for each of the at least one processing parameter to one or more terms of the target control function.

[0028] A 19th aspect of the present disclosure includes any one of the 1st to 18th aspects, wherein the step of determining an updated setting for each at least one processing parameter may further include a step of providing constraints on the at least one processing parameter including maximum and minimum values ​​for the at least one processing parameter.

[0029] A twentieth aspect of the present disclosure includes any one of the first to nineteenth aspects, and may further include providing target values ​​for a plurality of attributes of the plurality of glass articles, the plurality of glass tubes, or both, measuring the plurality of attributes of the plurality of glass articles, the plurality of glass tubes, or both for the plurality of glass articles, and recording settings of a plurality of processing parameters of the converter for the plurality of attributes to generate a plurality of data sets. Each of the plurality of data sets may include measurements for each of the plurality of attributes over a period of time and settings of each of the plurality of processing parameters of the converter. The method may further include processing each of the plurality of data sets to generate a statistical characteristic of a distribution for each of the plurality of attributes, determining updated settings for each of the plurality of processing parameters from the statistical characteristic of the distribution for each of the plurality of attributes, the target values ​​for each of the plurality of attributes, and settings of each of the plurality of processing parameters, and adjusting each of the plurality of processing parameters of the converter to the updated settings for each of the plurality of processing parameters. The updated settings for the plurality of processing parameters may be values ​​of the updated settings that minimize the target control function.

[0030] A twenty-first aspect of the present disclosure includes any one of the first to twentieth aspects and may further include a step of displaying a user interface on a display unit.

[0031] A 22nd aspect of the present disclosure may include the 21st aspect and further include receiving one or more user inputs from a user interface and varying the updated setting of at least one processing parameter based on the one or more user inputs.

[0032] A twenty-third aspect of the present disclosure includes any one of the first to twenty-second aspects, wherein the at least one attribute may include one or more attributes of the converted glass article, one or more attributes of one or more features of the partially formed glass article at the working end of the glass tube, one or more attributes of the glass tube, one or more attributes of the preform at the working end of the glass tube, or combinations thereof.

[0033] A twenty-fourth aspect of the present disclosure relates to a system for producing a plurality of glass articles from a plurality of glass tubes. The system includes a conversion unit and a control system, the conversion unit may include a plurality of holding units, each of which is operable to hold a glass tube and rotate the glass tube about a central axis of the glass tube, a plurality of processing stations including at least one heating station, at least one forming station, and at least one separation station, and at least one measurement device operable to measure one or more attributes of each glass article produced from the glass tube, each glass tube, or both. The conversion unit may be operable to translate the glass tube through each of the plurality of processing stations in succession to produce the plurality of glass articles. The control system may be communicatively coupled to the conversion unit and include one or more processors, one or more memory modules communicatively coupled to the one or more processors, and machine-readable, executable instructions stored in the one or more memory modules. The machine-readable, executable instructions, when executed by the one or more processors, may cause the control system to automatically measure at least one attribute of the glass article, the glass tube, or both with the at least one measurement device, record settings of at least one processing parameter of the converter for the at least one attribute to generate a dataset including the measurements of the at least one attribute and settings of each of the at least one processing parameter of the converter, process the dataset to generate a statistical characteristic of a distribution of the measurements of the at least one attribute, determine updated settings for the at least one processing parameter from the statistical characteristic of the distribution of the measurements of the at least one attribute, a target value for the at least one attribute, and a setting for each of the at least one processing parameter, and further adjust the at least one processing parameter of the converter to the updated setting. The updated setting for each of the at least one processing parameter may be a setting value that minimizes a target control function for the at least one attribute.

[0034] It should be understood that both the foregoing general description and the following detailed description are intended to provide an overview or framework for understanding the nature and features of the claimed subject matter described in various embodiments. The accompanying drawings are included to provide a further understanding of the various embodiments and are incorporated in and constitute a part of this specification. The drawings illustrate various embodiments described herein and, together with the description, serve to explain the principles and operation of the claimed subject matter. [Brief description of the drawings]

[0035] [Figure 1] FIG. 1 is a front view schematic diagram of an embodiment of a system including a converter for producing glass articles from glass tubes according to one or more embodiments shown and described herein. [Diagram 2] 2 is a schematic top view of a primary turret, a secondary turret, and a feed turret of the converter of FIG. 1 according to one or more embodiments shown and described herein. FIG. [Diagram 3] 2 is a schematic diagram illustrating a heating station of the converter of FIG. 1 according to one or more embodiments shown and described herein. [Figure 4] 2 is a schematic diagram illustrating one embodiment of a forming station of the transformer of FIG. 1 according to one or more embodiments shown and described herein. [Diagram 5] 2 is a schematic diagram illustrating another embodiment of a forming station of the conversion portion of FIG. 1 according to one or more embodiments shown and described herein. [Figure 6] 2 is a schematic diagram illustrating a cooling station of the converter of FIG. 1 according to one or more embodiments shown and described herein. [Figure 7] 2 is a schematic diagram illustrating a separation station of the conversion unit of FIG. 1 according to one or more embodiments shown and described herein. [Figure 8] 2 is a schematic diagram illustrating a drilling station of the converter of FIG. 1 according to one or more embodiments shown and described herein. [Figure 9] 2 is a schematic diagram illustrating a measurement station of the converter of FIG. 1 according to one or more embodiments shown and described herein. [Figure 10] FIG. 2 is a perspective view that illustrates a cross section of a glass tube prior to conversion at the conversion portion of FIG. 1 according to one or more embodiments shown and described herein. [Figure 11] FIG. 1 is a schematic diagram illustrating another embodiment of a system including a conversion unit for converting glass tubes into glass articles according to one or more embodiments shown and described herein. [Figure 12] 12 is a schematic diagram illustrating a control sequence for providing feedback control of the converter of FIG. 11 according to one or more embodiments shown and described herein. [Figure 13] 1 is a graph illustrating dimensions (y-axis) of a glass article according to one or more embodiments shown and described herein as a function of the glass article's unique identifier (x-axis) for multiple condition sets. [Figure 14] 14 is a graph illustrating several operating models developed from the dimensional data and condition sets from FIG. 13 in accordance with one or more embodiments shown and described herein. [Figure 15] 1 is a graph illustrating dimensions (y-axis) of a glass article according to one or more embodiments shown and described herein as a function of the glass article's unique identifier (x-axis) for multiple condition sets. [Figure 16] 16 is a graph illustrating several operating models developed from the dimensional data and condition sets from FIG. 15 in accordance with one or more embodiments shown and described herein. [Figure 17] 1 is a graph plotting normalized measured values ​​of various attributes of glass vials as a function of normalized predicted values ​​of those attributes using the operating model developed in Example 2 in accordance with one or more embodiments shown and described herein. [Figure 18] 1 is a graph plotting the sum of squares error (eg, the value of a target control function) as a function of the number of iterations for the control methods of Examples 2 and 3 in accordance with one or more embodiments shown and described herein. [Figure 19]1 is a graph showing normalized measurements of 13 attributes of glass vials produced in Example 2 for different iterations of the control method of Example 2 according to one or more embodiments shown and described herein. [Figure 20] 1 is a graph illustrating normalized measurements of maximum and minimum flange outer diameter, flange inner diameter, flange height, and top height of glass vials according to one or more embodiments shown and described herein as a function of time over four iterations of the control method of Example 2. [Figure 21] 1 is a graph illustrating normalized measurements of bottom flange angle, top flange angle, shoulder angle, and shoulder radius of glass vials according to one or more embodiments shown and described herein as a function of time over four replicates of the control method of Example 2. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0036]

[0023] Now, embodiments of the disclosed system and method for controlling the operation of a conversion process of glass tubes to produce glass articles will be described in detail, examples of which are illustrated in the accompanying drawings. Wherever possible, the same reference numbers are used to refer to the same or similar parts throughout the figures. A method disclosed herein for controlling a conversion unit for converting glass tubes to glass articles includes operating a conversion unit including a plurality of processing stations to produce a plurality of glass articles from a plurality of glass tubes, and operating the conversion unit may include translating the glass tube through each of the plurality of processing stations in succession. The method may further include providing a target value during or after conversion for at least one attribute of the plurality of glass articles or the plurality of glass tubes, measuring at least one attribute of the plurality of glass articles or the plurality of glass tubes during or after conversion, and recording the setting of at least one processing parameter of the conversion unit for the at least one attribute to generate a data set including the measured value of the at least one attribute and the setting of the at least one processing parameter of the conversion unit. The method may further include processing the data set to generate a statistical characteristic of the distribution of measurements of the at least one attribute, and determining an updated setting for each of the at least one processing parameter from the statistical characteristic of the distribution of the measured at least one attribute, a target value for the at least one attribute, and a setting for the at least one processing parameter. The updated setting for the at least one processing parameter may be a setting value that minimizes a target control function for the at least one attribute. In an embodiment, the target control function for the transform may be derived from a cost function, such as, but not limited to, a mean squared error cost function. The method may further include adjusting each of the at least one processing parameter of the transform to the updated setting.

[0037] As used herein, directional terms, such as upper, lower, right, left, front, back, top, and bottom, are used only with reference to the illustrated drawings and their coordinate axes and are not intended to imply absolute orientations.

[0038] Unless otherwise stated, it is not intended that any method set forth herein be construed as requiring its steps to be performed in a particular order or that any apparatus require a particular orientation. Thus, unless a method claim actually recites the order in which steps are performed, or any apparatus claim actually recites the order or orientation of individual components, or unless there is a specific statement in the claims or specification that the steps are limited to a particular order, or a specific order or orientation of the apparatus components, no order or direction is intended to be inferred in any way. This applies to any interpretation based on the absence of a description, including logical issues regarding the sequencing of steps, operational flow, order or orientation of components, grammatical structure or punctuation, and the simple meaning derived from the number or type of embodiments described herein.

[0039] In this specification, the singular indefinite and definite articles in the original English language include the plural unless otherwise clear from the context. Thus, for example, when an element is described using an indefinite article in the original English language, it includes embodiments having two or more of such elements unless otherwise clear from the context.

[0040] As used in this specification, the "working end" of the glass tube is the end of the glass tube that faces toward the processing station of the main turret of the conversion section, relative to the holding section, and the "non-working end" of the glass tube is the end of the glass tube that faces away from the processing station of the main turret.

[0041] As used herein, the "dwell time" of a converter refers to the amount of time that a glass tube spends at a particular processing station before proceeding to the next processing station.

[0042] As used herein, the term "active time" refers to the amount of time that a glass tube is maintained in interaction with at least one heating element or at least one former while the glass tube is in a particular processing station.

[0043] As used herein, the term "index time," when used in connection with an indexing converter, refers to the amount of time it takes to index a glass tube from one processing station to the next. "Dwell time," "active time," and "index time" are all measured in units of time.

[0044] When used in relation to a heating station, a burner "in engagement with" the glass tube refers to positioning the burner such that a flame from the burner extends toward or contacts the glass tube to heat it. Conversely, when the burner is not in engagement with the glass tube, the burner is positioned such that a flame from the burner is directed away from the glass tube or moved far enough away from the glass tube that the flame does not contact or directly heat the glass tube.

[0045] The term "actuate" when used in relation to a former at a forming station refers to the former contacting the glass tube. If the former does not act on the glass tube, the former does not contact the glass tube.

[0046] As used herein, the term "part speed" refers to the production or processing rate of a conversion part in units of number of glass articles per unit time.

[0047] As used herein, the term "circumference" of a glass tube refers to the 360 ​​degree collection of points on the glass tube at a certain radius r from the central axis D of the glass tube at a particular Z location (i.e., a location on the + / -Z axis of the drawing). The circumference of the glass tube may, for example, coincide with the outer surface of the glass tube at a particular Z location, or with the inner surface of the glass tube at a particular Z location.

[0048] As used herein, the term "operating" refers to the normal steady-state operation of the converter. Thus, as used herein, "operating settings" refers to the settings of the converter for normal steady-state operation of the converter.

[0049] As used herein, the terms "upstream" and "downstream" refer to the relative location of processing stations in a converter. If the glass tube encounters a second processing station before encountering a first processing station, the first processing station is considered "downstream" of the second processing station. Similarly, if the glass tube encounters a first processing station before encountering a second processing station, the first processing station is considered "upstream" of the second processing station.

[0050] Glass tubes are converted into glass articles, particularly glass articles for pharmaceutical applications, which may include, but are not limited to, vials, syringes, ampoules, cartridges, and other glass articles. Glass tubes may be converted into these glass articles using a conversion section (i.e., a converter) that includes multiple processing stations. The processing stations may include heating stations, forming stations, thermal separation stations, punching stations, and the like. Converters typically transform long lengths of glass tube into multiple glass articles using processes that include, but are not limited to, flame processing, rotary and stationary forming, separation (e.g., thermal separation or notching and impact cutting processes), punching, cooling, measuring, or other processing steps. Thus, glass articles produced through the conversion process performed by a converter are those that have been exposed to a series of flame burners or other heating sections and forming tools to form the glass tube into a particular shape and size, and then the formed glass articles are separated from the glass tube.

[0051] 1, there is shown a schematic diagram of an embodiment of a conversion unit 100 for producing glass articles from a glass tube 102. The conversion unit 100 may be used to convert the glass tube 102 into a plurality of glass articles. The conversion unit 100 may include a base 104 having a plurality of processing stations 106 and a main turret 108 disposed above the base 104 and rotatable about a central axis A relative to the base 104. The conversion unit 100 may further include a glass tube loading turret 110 disposed above the main turret 108 and feeding the glass tube 102 to the main turret 108. The conversion unit 100 may also include a plurality of secondary processing stations 112 on the base 104, and a secondary turret 114, which may be rotatable relative to the base 104.

[0052] As shown generally in FIG. 1, the base 104 of the conversion unit 100 may be stationary and the processing stations 106 may be coupled to an upper portion 105 of the base 104. The processing stations 106 may be arranged in a spaced apart arrangement on the main circuit 116. In an embodiment, the main circuit 116 may be circular and the main turret 108 may index or continuously move the glass tubes 102 through the processing stations 106 by rotation of the main turret 108 about a central axis A. Alternatively, in other embodiments, the main circuit 116 may be a linear array of processing stations 106. Although a circular arrangement of processing stations 106 is described herein, it should be understood that the subject matter disclosed herein may be equally applicable to conversion units having other arrangements of processing stations 106, such as linear, curvilinear, or irregularly arranged processing stations 106.

[0053] The type and / or shape of the glass article to be produced from the glass tube 102 may affect the total number of processing stations 106 in the converting unit 100. The number of processing stations 106 in the main turret 108 may be between 14 and 32. Although the converting unit 100 and converting process are described herein with respect to the converting unit 100 having 16 processing stations 106 in the main circuit 116, it should be understood that the converting unit 100 may have more or less than 16 processing stations 106 in the main circuit 116. The processing stations 106 may include, for example and without limitation, one or more heating, forming, polishing, cooling, separating, punching, measuring, feeding, ejection stations, other processing stations, or combinations thereof for producing glass articles from the glass tube 102. The type and / or shape of the article to be produced from the glass tube 102 may also affect the type of processing stations 106 and / or the order of the processing stations 106 in the converting unit 100.

[0054] The main turret 108 may be disposed above the base 104 and rotatably coupled to the base 104 such that the main turret 108 is rotatable relative to the base 104 about a central axis A. A drive motor (not shown) may be used to rotate the main turret 108 relative to the base 104. The main turret 108 may include a number of holders 130 configured to removably secure each glass tube 102 to the main turret 108. The holders 130 may be clamps, chucks, or other holding devices, or a combination of holding devices. The holders 130 may orient each glass tube 102 such that the glass tube 102 is substantially parallel to the central axis A of the main turret 108 and substantially perpendicular to the upper portion 105 of the base 104. Although the conversion unit 100 is described herein with respect to a vertically oriented conversion unit 100, it should be understood that the conversion unit 100 may be oriented horizontally or at an angle at which the glass tube 102 is not vertical during processing. Each holder 130 may extend in a direction from the bottom 109 of the main turret 108 toward the base 104 (i.e., in a -Z direction relative to the coordinate axes of FIG. 1). Each holder 130 may be oriented such that the glass tube 102 is located at or adjacent to each successive processing station 106 of the main circuit 116 of the base 104 as the main turret 108 is indexed about the central axis A. The vertical orientation of the glass tubes 102 allows the working end 150 of each glass tube 102 to move or index forward through the processing stations 106.

[0055] In an embodiment, the conversion unit 100 may be operable to index each of the plurality of holders 130 through the plurality of processing stations 106. Indexing may refer to a stepwise process of moving the glass tube 102 into a processing station 106, maintaining the glass tube 102 at a stationary position XYZ of the processing station 106 for a dwell time, and then indexing the glass tube 102 to the next processing station 106. Alternatively, in an embodiment, the conversion unit 100 may be operable to translate the plurality of holders 130 sequentially through the conversion process. In an embodiment, the processing stations 106 may translate with the glass tube 102 during the active time of the glass tube 102 at the processing station.

[0056] Each holder 130 may be individually rotatable relative to the main turret 108 to rotate the glass tube 102 about a central axis D of the glass tube 102, which may be parallel to a central axis A of the main turret 108. Each holder 130 may be operatively coupled to a motor (not shown), continuous drive belt, or other drive mechanism for rotating each holder 130 about the central axis D of the glass tube 102 relative to the main turret 108. Rotation of the holders 130 allows the glass tube 102 to rotate about the central axis D of the glass tube 102 relative to a stationary burner, former, cooling nozzle, or other feature of the processing station 106. A heating element or former of the processing station 106 may be maintained in a fixed position relative to the glass tube 102, allowing the glass tube 102 to rotate about the central axis D to expose the entire circumference of the glass tube 102 to the heating element or former.

[0057] 1 and 2, the conversion unit 100 may include a plurality of secondary processing stations 112, which may also be spaced apart and arranged in a secondary circuit 118 (FIG. 2). The conversion unit 100 may include a secondary turret 114 (FIG. 1) that indexes or continuously moves articles 103 (FIG. 1) separated from glass tubes 102 through the plurality of secondary processing stations 112. The secondary turret 114 may be rotatable relative to the base 104 about a second axis B. The second axis B may be generally parallel to the central axis A of the main turret 108. The secondary turret 114 may also include a plurality of holders 130 that hold the glass articles 103 and position them for each of the secondary processing stations 112 to act on in turn. The secondary turret 114 receives the glass articles 103 from a separation station 206 (FIG. 2) of the main turret 108 and may index or move the articles 103 through a plurality of secondary processing stations 112 by rotation of the secondary turret 114 to discharge the finished articles from the conversion section 100. Although shown in a circular pattern, it should be understood that the secondary processing stations 112 may be arranged in a linear, curvilinear, or irregular manner. The secondary processing stations 112 may be referred to as bottom formers. For vials, the secondary processing stations 112 may be operable to form the bottom of the vial.

[0058] The conversion section 100 may include an apparatus for loading a new length of glass tube 102 into the holder 130. In an embodiment, the conversion section 100 may include a glass tube loading turret 110 adjacent to the main turret 108 in a position where the glass tube loading turret 110 can load a new length of glass tube 102 into the holder 130 of the main turret 108 at at least one processing station 106. In an embodiment, the glass tube loading turret 110 may be positioned off of a central axis A of the main turret 108. The glass tube loading turret 110 may be rotatable about an axis C, which may be substantially parallel to the central axis A of the main turret 108. The glass tube loading turret 110 may be independently supported in a stationary position relative to the main turret 108, and the rotation of the glass tube loading turret 110 may be independent of the rotation of the main turret 108. In an embodiment, the processing station 106 aligned with the glass tube loading turret 110 may be a tube loading station 214 (FIG. 2). When the conversion unit 100 has converted all or substantially all of the glass tubes 102 into one or more articles at a particular holder position 136, the glass tube loading turret 110 may deliver the new length of glass tube 102 through the top of the main turret 108 to the holder 130 at the holder position 136 when the holder position 136 is indexed into alignment with the tube loading station 214 (FIG. 2). Alternatively, or in addition, in an embodiment, the conversion unit 100 may include an arm (not shown) movable between the main turret 108 and the glass tube loading turret 110. When the conversion unit 100 has converted all or a portion of the glass tube 102 at a particular holder location 136, the arm unit may grab a new length of glass tube 102 from the glass tube loading turret 110 or other glass tube staging device and deliver the new length of glass tube 102 to the main turret 108 at the particular holder location 136. Other methods and apparatus for delivering the new length of glass tube 102 to the main turret 108 are also contemplated.

[0059] 2, as previously described, the multiple processing stations 106 of the conversion unit 100 may include one or more of a heating station 202, a forming station 204, a separating station 206, a cooling station 210, a piercing station 212, a tube loading station 214, a discharge station 216, a measuring station 218, a tube length dropping station 220, or other stations and / or combinations of these stations. FIG. 2 illustrates a schematic arrangement of the processing stations 106 for a conversion unit 100 having a main circuit 116 having 16 processing stations 106 and a secondary circuit 118 having 8 secondary processing stations 112. As previously described, the processing stations 106 of the main circuit 116 may be equally spaced and uniformly distributed in a circular circuit, and the secondary processing stations 112 of the secondary circuit 118 may also be equally spaced and uniformly distributed in a circular circuit. FIG. 2 also shows diagrammatically a glass tube loading turret 110 having a number of loading channels 132 .

[0060] The converter main circuit 116, as shown generally in FIG. 2, may include one or more heating stations 202, a separation station 206, a piercing station 212, one or more forming stations 204, one or more cooling stations 210, a measuring station 218, a tube length drop station 220, and a tube loading station 214. Although FIG. 2 illustrates the main circuit 116 as having a circular arrangement of processing stations 106, as previously described, the main circuit 116 may have processing stations 106 arranged in other non-circular geometric arrangements, such as linear, curved, irregular, or other arrangements. A heating station 202 may be positioned before each forming station 204 and separation station 206 in the index feed direction 222 of the main turret 108 to preheat a target area of ​​the glass tube 102 to a viscosity at which the glass can be deformed, effectively forming or stretching, and separating. In the separation station 206, the formed glass article 103 (FIG. 1) may be separated from the glass tube 102 (FIG. 1) so as to simultaneously form its bottom. The separation station 206 is also a processing station 106, where the partially formed glass article 103, once separated, may be transferred to the secondary turret 114 (FIG. 1) and indexed through the secondary circuit 118 with the secondary processing station 112. The perforation station 212 may be located in the main circuit 116 downstream of the separation station 206 with respect to the index feed direction 222 of the main turret 108. In the perforation station 212, the meniscus portion 350 (FIG. 8) previously formed in the separation station 206 is perforated, thereby reopening the working end 150 of the glass tube 102.

[0061] Referring again to FIG. 2, the forming station 204 of the main turret 108 may be located downstream of the punching station 212 and the one or more heating stations 202 in the index feed direction 222. The one or more forming stations 204 may repeatedly form the glass tube 102 to form one or more features of the finished glass article. As described above, one or more heating stations 202 may be located before each forming station 204 to preheat a target area of ​​the glass tube 102 to a temperature at which the glass tube 102 may be shaped to form a desired feature. The forming station 204 of the main turret 108 may shape the working end 150 (FIG. 3) of the glass tube 102 to form a feature at one end of the glass article 103, and the forming station 204 of the secondary turret 114 may shape the other end of the glass article 103 after the glass article 103 is separated from the glass tube 102. In an embodiment, a conversion unit 100 is used to produce vials from glass tubes 102, and the forming stations 204 of the conversion unit 100 may include one or more shoulder forming stations, flange forming stations, flange finishing stations, or combinations thereof, and may further include one or more heating stations 202 disposed before and between each forming station 204.

[0062] The main circuit 116 further includes a measurement station 218, where at least one measurement device may be used to measure one or more attributes of the glass tube 102, such as, for example, diameter and thickness, or one or more dimensions of a feature of the glass article 103 formed by the forming station 204. The feature dimensions may include, but are not limited to, flange thickness, flange length, neck length, neck thickness, overall article length, flange inner diameter, flange outer diameter, flange height, top height, bottom flange angle, top flange angle, eccentricity, article inner or outer diameter, shoulder thickness, shoulder angle, shoulder radius, other feature dimensions, or combinations thereof. One or more cosmetic attributes of the glass tube 102 or glass article 103 may also be evaluated at the measurement station 218. The cosmetic attributes may include, but are not limited to, defects in one or more features of the glass article 103 (e.g., defects in the flanges, neck, etc.), overall desirability, or combinations thereof. The overall desirability may be a composite characteristic based on multiple other dimensional or cosmetic attributes measured for the glass article 103. In an embodiment, the measurement station 218 may be located immediately after the last forming station 204 to measure dimensions while the glass tube 102 is still hot. Alternatively, the measurement station 218 may be located after one or more cooling stations 210 to measure dimensions of the glass tube 102 and / or glass article 103 at a lower temperature. In an embodiment, the secondary circuit 118 of the conversion unit 100 may include the measurement station 218.

[0063] With further reference to FIG. 2, one or more cooling stations 210 may be positioned after the forming station 204 in the index feed direction 222 of the main turret 108. A tube length drop station 220 may be positioned after the forming station 204 between the forming station 204 and the separation station 206 to drop the partially formed glass tube 102 so that the glass tube 102 separates the glass article 103 from the glass tube 102 at the separation station 206. The main circuit 116 may also include a tube loading station 214 that loads a supply of new glass tube lengths 102 from the glass tube loading turret 110 to the main turret 108 (FIG. 1). In an embodiment, the tube loading station 214 may be incorporated into the cooling station 210. The tube loading station 214 may be positioned between the last forming station 204 and the separation station 206.

[0064] The forming station 204 of the main turret 108 may form features at a first end of the glass article 103. For example, as shown in FIG. 4, the forming station 204 may form a shoulder 142 and a flange 144 at a top (first end) of the glass article 103, which may be a vial or cartridge. Referring again to FIG. 2, once the glass article 103 is separated from the glass tube 102 at the separation station 206, the glass article 103 may be transferred to a secondary processing station 112 of the secondary turret 114. The secondary processing station 112 may include one or more forming stations 204 that form a second end of the glass article 103 opposite the first end of the glass article 103. For example, the forming station 204 of the secondary processing station 112 may form one or more features at a bottom (second end) of the glass article 103. The secondary turret 114 may rotate about axis B in a direction 224 opposite the main turret 108. In an embodiment, the secondary turret 114 may rotate in the same direction as the primary turret 108 .

[0065] The secondary processing stations of the secondary circuit may include one or more of the heating station 202, forming station 204, polishing station 208, cooling station 210, drain station 216, or other stations or combinations of the secondary processing stations 112. Although FIG. 2 illustrates the secondary circuit as having a circular arrangement of the secondary processing stations 112, as previously described, the secondary circuit may have the secondary processing stations 112 arranged in other non-circular arrangements, such as linear, curved, irregularly shaped, or other arrangements. In embodiments, the secondary processing stations 112 of the secondary circuit 118 may be used to form one or more features of the glass article 103, such as, for example, a vial, an ampoule, a cartridge, or a syringe, at an end of the glass article 103 opposite the end formed by the main turret 108. For example, in some embodiments, the glass article 103 is a vial and the forming station 204 of the secondary circuit 118 may form the bottom of the vial. Other features are contemplated, such as distinctive features of ampoules, cartridges, syringes, etc. The secondary circuit 118 may include one or more polishing stations 208 to finish the surface of the glass articles. The secondary circuit 118 may further include multiple cooling stations 210, and a discharge station 216, at which the finished glass articles 103 may be discharged from the conversion section 100.

[0066] The above description of the processing stations 106 in the primary circuit 116 and the secondary processing stations 112 in the secondary circuit 118 may represent an exemplary converter 100 that produces vials from glass tubes 102. However, it should be understood that a greater or lesser number of processing stations 106 and secondary processing stations 112 may be used to produce other glass articles, such as vials, cartridges, syringes, ampoules, or other pharmaceutical glass articles having different shapes or features. Furthermore, it should be understood that the processing stations 106 and secondary processing stations 112 may be arranged in any of a number of different sequences and / or configurations to produce glass articles of different shapes.

[0067] Referring now to FIG. 3, the heating stations 202 of the conversion section 100 are shown in a schematic manner. Each heating station 202 may include one or more heating sections 301. As shown in FIG. 3, in an embodiment, the heating section 301 may include one or more burners 302 that may be used to heat a region of interest on the glass tube 102 prior to a forming operation performed at the forming station 204 (FIG. 2) or a separating operation performed at the separating station 206 (FIG. 2). Although FIG. 3 shows a single burner 302, it should be understood that multiple burners 302 may be used in a single heating station 202. Each burner 302 may be fluidly connected to a fuel gas supply 304, an oxygen supply 306, and optionally an air supply 308. Examples of fuel gases for the burners 302 may include, but are not limited to, hydrogen, hydrocarbon fuel gases such as methane, propane, and butane, other fuel gases, or combinations thereof.

[0068] Each burner 302 may include a fuel control valve 310 to control the flow of fuel gas to the burner 302. Each burner 302 may also include an oxygen control valve 312 to control the mass flow rate of oxygen to the burner 302. Each burner 302 may further include an air control valve 314 to optionally control the flow rate of air to the burner 302. The burners 302 combust the fuel gas in the presence of oxygen and / or air to generate a flame to heat at least a target area of ​​the glass tube 102. Although the heating station 202 of the conversion unit 100 is described herein as heating the glass tube 102 with a burner, it should be understood that other heating units or methods other than a burner may be used to heat the glass tube 102. Other heating units may include, but are not limited to, CO 2 These may include lasers, such as lasers, for example, induction heaters, other heating devices, and combinations thereof.

[0069] The heating station 202 may further include a burner arrangement 318 coupled to the burner 302. The burner arrangement 318 may be operable to position the burner 302 vertically (e.g., in the + / -Z direction of the coordinate axes of FIG. 3), horizontally (e.g., in the XY plane identified by the coordinate axes of FIG. 3), or in a combination of these directions relative to the glass tube 102 at the heating station 202. In an embodiment, each burner arrangement 318 may include one or more servo motors operable to automatically and / or incrementally adjust the position of the burner 302 in one or more directions. Any other type of arrangement that is commercially available or will be commercially available may also be used for the burner arrangement 318. The burner arrangement 318, the fuel control valve 310, the oxygen control valve 312, the air control valve 314, or a combination thereof, may be communicatively coupled to the control system 402 to enable the control system 402 to control the vertical position, horizontal position, heat output, or a combination thereof, of the burner 302.

[0070] 4 and 5, an example of a forming station 204 of the conversion section 100 is shown in schematic form. Each forming station 204 may include one or more formers 324 rotatable about a tooling axis E relative to the base 104 (FIG. 1). To proceed to a forming station 204, the glass tube 102, which has been heated in the previous heating station 202, is rotated by the holding section 130. The formers 324 may act on the glass tube 102 as it rotates. When acting, the formers 324 may contact the heated glass tube 102 to form the glass tube 102 into a desired shape. The formers 324 may contact the glass tube 102 during the active time of the formers 324. When the active time is over, the former actuator 326 retracts the formers 324 from acting on the glass tube 102. Figure 4 illustrates a schematic of an embodiment of a forming station 204 for forming the shoulder 142 of a glass vial. Figure 5 illustrates a schematic of an exemplary embodiment of a forming station 204' for forming the flange 144 of a glass vial. The forming station 204' for forming the flange 144 includes three formers 324a, 324b, 324c. Other types of formers 324 may also be used at the forming station 204 depending on the desired characteristics of the glass article 103.

[0071] Referring again to FIG. 4, the former actuator 326 may be operable to move the former 324 into and out of action on the glass tube 102. By moving the former 324 into and out of action on the glass tube 102, the timing of contact of the former 324 with the glass tube 102 may be controlled. The timing of contact of the former 324 with the glass tube 102 refers to the timing at which each former 324 in the forming station 204 acts or does not act on the glass tube 102. By adjusting the timing of contact of the former 324, the total contact time that each former 324 makes contact with the glass tube 102 may be adjusted. The contact time refers to the length of time that the former 324 acts or makes contact with the glass tube 102.

[0072] The former actuators 326 may further be operable to change the position of the former 324 at the forming station 204 vertically (e.g., in the + / -Z direction of the coordinate axes of FIG. 3), horizontally (e.g., in the XY plane identified by the coordinate axes of FIG. 4), or a combination of these directions relative to the glass tube 102. The forming position of the former 324 refers to the former position when the former 324 acts on the glass tube 102. In embodiments, each former actuator 326 may include one or more servo motors operable to automatically and / or incrementally adjust the position of the former 324 in one or more directions of the coordinate axes of FIG. 4. Any other type of arrangement that is commercially available or will be commercially available may also be used as at least a portion of the former actuator 326. The former actuator 326 may be communicatively coupled to the control system 402 to enable the control system 402 to vary the vertical position, horizontal position, or both, of the former 324 when in the forming position. The vertical and / or horizontal position of the former 324 refers to the position of the former 324 when it acts on the glass tube 102.

[0073] 6, the cooling station 210 is shown in schematic form. The cooling station 210 may include one or more cooling nozzles 340 arranged to direct a cooling fluid 342, such as, for example, cooling air or an inert gas, toward the glass tube 102. One or more of the cooling nozzles 340 may be arranged to direct the cooling fluid 342 toward a particular region of the glass tube 102. One or more cooling fluid control valves 344 may be fluidly coupled to the cooling nozzles 340 to control the mass flow rate of the cooling fluid 342 to the cooling nozzles 340, thereby enabling control of the cooling rate of the glass tube 102, as well as the temperature of the glass tube 102 and the temperature gradient in the glass tube 102. The cooling station 210 may also include a cooling nozzle positioner (not shown) coupled to the cooling nozzles 340 and operable to position the cooling nozzles 340 relative to the glass tube 102. The cooling nozzle arrangement, the cooling fluid control valve 344, or both, may be communicatively coupled to a control system 402 (FIG. 1) enabling the control system 402 to control the vertical and / or horizontal position of the cooling nozzles 340, the flow rate of the cooling fluid 342, or a combination thereof.

[0074] 7, a separation station 206 of the conversion unit 100 is shown in schematic form. The separation station 206 shown in FIG. 7 is a thermal separation station and may be located after one or more heating stations 202 in the index feed direction 222 of the main turret 108. The heating station 202 located before the separation station 206 may heat the glass tube 102 to make the glass viscous. The separation station 206 may include a separation burner 348. The separation burner 348 may include any of the features previously described for the burner 302, including, but not limited to, a fuel gas control valve 310, an oxygen control valve 312, and / or an air control valve 314. While the glass tube 102, viscously deformed by the previous heating station 202, is rotated by the holding unit 130 about the central axis D of the glass tube 102, the separation burner 348 may act on the outer surface 140 of the glass tube 102 to heat the glass tube 102 to a temperature where the viscosity of the glass causes the partially formed glass article to separate from the glass tube 102. Once separated from the glass tube 102, the partially formed article may be transferred to the secondary turret 114 (FIG. 1) or ejected from the conversion unit 100. Although shown in FIG. 7 as a thermal separation station, the separation station 206 may also be a non-thermal separation station, such as a notch breakage technique separation station that may be used for syringes and cartridges.

[0075] Similar to the heating station 202, the separation station 206 may also include a burner arrangement 318 coupled to a separate burner 348. The burner arrangement 318 may be operable to position the separate burner 348 in the separation station 206 relative to the glass tube 102 vertically (e.g., in the + / -Z direction of the coordinate axes of FIG. 3), horizontally (e.g., in the XY plane identified by the coordinate axes of FIG. 7), or a combination of these orientations. The burner arrangement 318, the fuel control valve 310, the oxygen control valve 312, the air control valve 314, or a combination thereof, may be communicatively coupled to a control system 402 (FIG. 3) to enable the control system 402 to control the vertical position, horizontal position, heat output, or a combination thereof, of the separate burner 348. If the separation station 206 is a cut-break separation station, the separation station 206 may include one or more cut tools and / or break tools and may further include a tool actuator operable to vary the positioning of the cut tools and / or break tools.

[0076] 8, one embodiment of the piercing station 212 of the conversion portion 100 is shown diagrammatically. The piercing station 212 may be located after the separation station 206 in the index feed direction 222 of the main turret 108. As previously described, the article 103 may be thermally separated from the glass tube 102 at the separation station 206 to form a meniscus 350 of glass over the working end 150 of the glass tube 102. At the piercing station, the meniscus 350 is pierced in preparation for forming the next article at the working end 150 of the glass tube 102.

[0077] In an embodiment, the perforation station 212 may include a perforation burner 352. The perforation burner 352 may be positioned below and directed toward the working end 150 of the glass tube 102. The perforation burner 352 may be fluidly connected to one or more of a fuel gas supply 304, an oxygen supply 306, an air supply 308, or a combination thereof. The fuel gas supply 304, the oxygen supply 306, and the air supply 308 are as previously described for the burner 302 of FIG. 3. The perforation station 212 may also include a fuel gas control valve 310, an oxygen control valve 312, and / or an air control valve 314 that control the heat output from the perforation burner 352. As the main turret 108 indexes the glass tube 102 to the piercing station 212, a flame from the piercing burner 352 heats the meniscus portion 350 of the glass, melting the meniscus portion 350, piercing the meniscus portion 350, and reopening the working end 150 of the glass tube 102. In an embodiment, the meniscus portion 350 may be pierced by directing a flow of gas, such as compressed air, nitrogen, argon, or other gas, at or across the meniscus portion 350. In an embodiment, instead of using the piercing burner 352, the meniscus portion 350 may be pierced using mechanical means or other methods.Various methods for perforating the meniscus portion 350 are described in U.S. Pat. No. 10,968,133, issued April 6, 2021, entitled "METHODS FOR MINIMIZING SHR IN GLASS ARTICLES BY PRODUCING A GAS FLOW DURING PHARMACEUTICAL PART CONVERTING," and in co-pending U.S. patent application Ser. No. 16 / 197,187, filed November 20, 2018, entitled "SYSTEMS AND METHODS FOR MINIMIZING SHR FROM PIERCING DURING PHARMACEUTICAL PART CONVERTING USING A GAS FLOW," and in co-pending U.S. patent application Ser. No. 16 / 197,187, filed November 20, 2018, entitled "SYSTEMS AND METHODS FOR MINIMIZING SHR FROM PIERCING DURING PHARMACEUTICAL PART CONVERTING USING NEGATIVE PRESSURE." No. 16 / 197,971, filed Nov. 21, 2018, entitled "MINIMIZING SHR FROM PIERCING FROM PHARMACEUTICAL PART CONVERTING USING PULSED EJECTION," and co-pending U.S. patent application Ser. No. 16 / 198,041, filed Nov. 21, 2018, entitled "SYSTEMS AND METHODS FOR MINIMIZING SHR FROM PIERCING FROM PHARMACEUTICAL PART CONVERTING USING PULSED EJECTION," the entire contents of which are incorporated herein by reference. Positioning units, control valves, and other controls may be incorporated in the drilling station 212 and communicatively coupled to a control system (FIG. 1) to control various operating parameters of the drilling station 212.

[0078] 9, the measurement station 218 is shown diagrammatically. The measurement station 218 may include one or more measurement devices 360 arranged to measure one or more attributes of the glass tube 102 and / or the glass article 103. The attributes may include one or more physical dimensions, one or more cosmetic characteristics, or both, of the glass tube 102, the glass article 103, or / and. The measurement device 360 ​​of the measurement station 218 may be any of the measurement devices described herein. The measurement device 360 ​​may be communicatively coupled to the control system 402 to transmit information regarding one or more attributes of the glass tube 102, the glass article 103, or / and to the control system 402. In an embodiment, the measurement device 360 ​​may be a thermal imaging device. An example of a thermal imaging device for measuring attributes and characteristics of the glass tube 102 or features formed at the working end 150 of the glass tube 102 is described in U.S. Pat. No. 10,773,989, issued September 15, 2020, the entire contents of which are incorporated herein by reference.

[0079] 3-9 include schematic diagrams of several different example processing stations 106 that may be used in the conversion portion 100. However, it should be understood that other processing stations 106 having different structures, combinations of structures, or functionality may be used to achieve the desired conversion of the glass tube 102 into one or more glass articles.

[0080] 10, the glass tube 102 may be an elongated hollow cylindrical tube made of glass. The glass tube 102 may have a circular cross-sectional shape and may have an outer surface 140, an inner surface 146, and a thickness t. The thickness t of the glass tube 102 may be the radial distance between the inner surface 146 and the outer surface 140 of the glass tube 102. The glass tube 102 may have a length L in the + / -Z direction of the coordinate axis of FIG. 10. As shown in FIG. 10, the glass tube 102 may have an outer diameter OD. As previously described, the glass tube 102 may be rotated about the central axis D of the glass tube 102 throughout the conversion process. The working end 150 of the glass tube 102 is the end of the glass tube 102 that faces the -Z direction of the coordinate axis of FIG. 1 when the glass tube 102 is fixed to the holding portion 130 of the conversion portion 100. The non-working end of the glass tube 102 is the end opposite the working end 150 (ie, the end of the glass tube 102 in the +Z direction of the coordinate axis of FIG. 10).

[0081] 1 and 2, in operation, the main turret 108 may index or move the glass tube 102 secured in the holder 130 to the processing stations 106. A particular operation may be performed on the glass tube 102 at each processing station 106, such as heating, forming, piercing, separating, cooling, dropping, feeding, measuring, etc. The converter 100 may be adjusted so that all of the processing stations 106 complete their operations within the dwell time. At the end of the dwell time, during the index time, the main turret 108 may index the glass tube 102 to the next processing station 106. As used in this disclosure, for an indexing converter, the total time per section per station is the sum of the dwell time and the index time.

[0082] In embodiments, the converter 100 may be a continuous converter operable to move the glass tube 102 and holder 130 continuously through a plurality of processing stations 106. In embodiments, the heaters, burners, formers, measuring devices, and other components of the converting process may move with the glass tube 102 as it advances through the processing stations 106. For both indexed feed converters and continuous converters, the "active time" of a processing station is the amount of time that the glass tube 102 remains under the influence of at least one heater, at least one former, at least one cooling nozzle, or other device while in the processing station 106.

[0083] Examples of converters 100 that convert glass tubes 102 into glass vials include Vial Forming Machine Models RP16 or RP18 with Automatic Tube Feeder manufactured by AMBEG Dr. J. Dichter GmbH, which include 16 processing stations 106 in a main circuit 116 and 8 secondary processing stations 112. Another example includes Vial Forming Machine Model RP32 manufactured by AMBEG Dr. J. Dichter GmbH, which includes 32 processing stations 106 in a main circuit 116 and 2 secondary circuits 118, each secondary circuit 118 including 8 secondary processing stations 112, and yet another example includes the Zeta 098 Vial Forming Machine manufactured by Euromatic SRL, which includes 36 processing stations. Another example can include the Zeta 103 Cartridge Forming Machine manufactured by Euromatic SRL, which is a converter that converts glass tubes into glass cartridges. The cartridge conversion unit has similar features as the vial conversion unit 100 previously described, but is used to produce glass articles having a cartridge form factor rather than a vial.

[0084] Although described in the context of the conversion unit 100 producing glass vials from glass tubing 102, it should be understood that by varying the formers 324 and / or the sequence or configuration of the processing stations 106 in the primary circuit 116 or the secondary processing stations 112 in one or more secondary circuits 118, the conversion unit 100 may be configured to produce one or more other articles, such as other types of pharmaceutical containers or articles. The pharmaceutical articles may include, but are not limited to, vials, cartridges, syringes, ampoules, jars, or other glass pharmaceutical articles.

[0085] During operation of the converting section 100, the burner power of the burner 302 and the positioning of the burner 302 can greatly affect the dimensional yield and defect rate of the converting section 100. The position of the former 324 relative to the glass tube 102 can also greatly affect the dimensional yield and defect rate of the converting section 100. Furthermore, the contact timing of the former 324 contacting the glass tube 102 can greatly affect whether the glass article 103 fabricated from the glass tube 102 meets the dimensional tolerances and defects. The total contact time in a conventional converting machine is inherently tied to the index feed rate of the machine, further complicating and limiting the processing window of a conventional converting machine to produce a high quality product with a high yield. Furthermore, the converting section 100 can have hundreds of other inputs and processing settings across all processing stations 106 that need to be properly and consistently managed to produce a high quality product with a high manufacturing yield.

[0086] Traditionally, in conventional converters, the burner power required to heat the glass tube 102 to a viscosity sufficient for forming has been controlled by simple needle valves fluidly connected to the fuel gas, oxygen, and air manifolds. In conventional converter forming stations, the position of the formers, which often include forming pins and forming wheels, has been set by simple mechanical linkages connected to cam drive shafts. Several advanced automation techniques have been developed and have become "add-on" features for some recent glass tube converters. In particular, mass flow control valves have been incorporated to replace manual needle valves and set the gas and oxygen flows to the various burners. Servo motors have replaced adjustable linkages used to position formers at the forming station. More advanced PLCs have allowed for much more advanced control of the forming station, with timing and speed profiles applied to each servo drive compared to conventional cam-driven machines. However, current technology still requires a human operator to make decisions regarding the selection and adjustment of process parameters, albeit through interaction with more sophisticated software and technology.

[0087] The efficiency of operation of the converter 100 for producing glass articles 103 from glass tubes 102 depends heavily on the ability to timely compensate for disturbances in the process that cause one or more attributes of the glass article 103 to deviate from the target value and / or increase the variance of these attributes. It is difficult to consistently adjust the large number of attributes of a glass article or glass tube by managing and adjusting hundreds of input and processing parameter settings of the glass tube converter 100.

[0088] Additionally, normal variations in the conversion process or variations in external inputs to the conversion process may also require adjustments to the converter. The variations may be due to input material or fuel gas, thermal expansion of the turret over time, gradual changes in equipment dimensions as the equipment wears, deterioration of the burner over time, or other factors. This variation may manifest itself as a shift or trend in the measured glass attribute over time. Even after determining the nominal set point, changes in input material or input fuel gas, wear of the equipment, changes in the environment surrounding the converter, and numerous other disturbances may require the actuator set point itself to be adjusted in real time to compensate the process. The source of the disturbance is often unknown, and even if known, some causes are less feasible to economically eliminate than the development and use of feedback controls. From the large number of possible actuator set point values, 10 to about 30 attributes of the glass article or glass tube are considered initially, and perhaps 10 to 20 adjustments are considered, but most may be fixed. Typically there are fewer actuator settings to control than attributes to control, and because actuators have maximum and minimum limiting values, some form of optimization or tradeoff may be necessary. Furthermore, the actuators controlling the various burners and formers have complex interactions, and the sensitivity of attributes to actuator changes is often variable and not well known.

[0089] For conventional glass tube converters, the control strategy to manage all of the process inputs and settings has been to use human operators to vary the process parameters of the converter 100 based on human intuition and experience. These human operators of various levels of experience have been relied upon to vary the burner parameters, former parameters, and overall machine timing for glass articles of different geometries, such as vials, syringes, ampoules, cartridges, etc. Furthermore, human operators have been relied upon to manage and compensate for the normal day-to-day variations that occur during the converting process by adjusting these same parameters. Some experienced operators have learned through experience and understanding of basic first principles to adjust the set points of numerous actuation sections to maintain the glass article attribute values ​​within specification targets. However, the operator's response to these changing attributes can have a significant impact on steady-state glass article yield and quality. Thus, a manufacturer's ability to consistently produce high quality glass articles from a converter at high yields has been heavily dependent on the skill level and experience of the human operators used to operate the converter. Different experience levels from operator to operator, machine to machine, production line to production line, and shift to shift can result in wide variations in yield and quality of glass articles produced by the converting process.

[0090] The task of controlling the glass tube converter 100 is multidimensional. It is often difficult and impossible for human performance to simultaneously consider more than about three actuators and about three attributes. Furthermore, some flange dimensions have been shown to have some degree of variance heterogeneity. Thus, since maximization of the process capability index Cpk is ultimately desired, one must consider balancing the attribute variances, not just that the attribute means are at or near the target. Finally, the operator in the loop approach is highly manual and therefore requires frequent attention to notice if and when the process drift becomes excessive or there is a sudden change in the performance of the converter 100.

[0091] Thus, there remains a need for a system and method for controlling the operation of a conversion unit 100 and conversion process to produce glass articles 103 from glass tubes 102. The systems and methods disclosed herein address problems in the current art by providing automatic or semi-automatic feedback control based on a model predictive control framework to efficiently control a conversion unit 100 that produces glass articles 103 from glass tubes 102. The feedback control method disclosed herein efficiently controls multiple process parameters of the conversion unit 100 to produce glass articles having attributes within specifications during steady-state operation of the conversion unit 100.

[0092] The model predictive control framework includes developing one or more behavioral models for the converter 100, each of which relates at least one processing parameter of the converter 100 to at least one attribute of the glass article 103, the glass tube 102, or both. The behavioral models included in the controller may be determined by experimental design techniques developed through first principles, or a combination of both. The behavioral models for the converter 100 may be substituted into a target control function. In an embodiment, the target control function may be a mean squared error cost function for the converter 100. Although described herein as using a mean squared error cost function, it is understood that the target control function may be based on other types of cost functions or any other suitable control function. The target control function may include attribute weighting factors and penalty factors for adjustment of the processing parameters. Updated settings for the processing parameters of the converter may be determined by solving the target control function for values ​​of the processing parameters that minimize the target control function. The solution to minimize the target control function may further consider maximum and minimum values ​​for each process parameter.

[0093] A model predictive control framework may be used to provide feedback control of the conversion unit 100. The feedback control method disclosed herein for controlling the operation of the conversion unit 100 may include performing one or many iterations of a control sequence. The control sequence may include providing target values ​​for attributes of the glass article 103 or glass tube 102 during or after conversion. The feedback control method may further include operating the conversion unit 100 to produce a glass article 103 from the glass tube 102 by translating the glass tube 102 through a number of processing stations 106 in succession, measuring attributes of the glass article or glass tube for a statistically relevant number of glass tubes, and recording the settings of the conversion unit process parameters for those attributes to generate one or more data sets including the measured values ​​of the attributes and the settings of the process parameters. The data sets may be processed to remove outlying data points and generate statistical characteristics (e.g., mean, median, spread, standard deviation, variance, etc.) of the distribution of the measured values ​​of the attributes. The method may further include determining updated settings from the target control function by substituting statistical characteristics of the distribution of each measured attribute, the target values ​​of the attributes, and the settings of the process parameters into the target control function, and then finding the updated settings that minimize the value of the target control function. The method may further include adjusting the process settings to the updated settings determined from the minimization of the target control function. The control sequence may be repeated until the updated settings of the process parameters converge.

[0094] Over time, the operating conditions of the converter 100 may change, leading to a large difference between the operating model and the actual performance of the converter 100. The model predictive control framework disclosed herein uses feedback control to compensate for the mismatch between such operating models and the actual converter performance, such that the converter operation produces glass articles that remain within acceptable specifications despite the initial operating model potentially being substantially inaccurate. Thus, the initial operating model may generally be in error for nonlinearity, offset, and sensitivity errors. The model predictive control framework disclosed herein compensates for process changes over time, such as changes in glass composition, wear of formers and burners, changes in the converter's ambient environment (e.g., temperature, humidity, air pressure, etc.), or other process changes. The feedback control process disclosed herein may also reduce the reliance on human operators to control the converter 100 for more consistent and reliable performance. The feedback control method may also allow the operation of the converter 100 to be adjusted to compensate for normal process variations in a consistent, data-driven manner, rather than relying on guesswork or models based on human intuition and experience.

[0095] It should also be noted that some anomalous behavior of the conversion unit 100 may be difficult to capture algorithmically, but may be observed and noticed by a human operator. The feedback control methods disclosed herein may be adapted to take into account varying degrees of human interaction, providing the opportunity to observe these anomalous behaviors and compensate when controlling the conversion process.

[0096] The feedback control method disclosed herein may be performed at least partially or completely by a control system communicatively connected to the conversion unit. Referring again to FIG. 1, a system 400 for producing a plurality of glass articles from a glass tube 102 may include a conversion unit 100 having a plurality of processing stations 106, such as, but not limited to, at least one heating station 202, at least one forming station 204, and a separation station 206. As previously described, the conversion unit 100 may be operable to move or advance the glass tube 102 through each of the plurality of processing stations 106. The conversion unit 100 may include a plurality of holding units 130. Each of the plurality of holding units 130 may be operable to secure the glass tube 102 and rotate the glass tube 102 about a central axis D of the glass tube 102. The conversion unit 100 may further include at least one measurement device 360 ​​operable to measure one or more attributes of each glass article 103 produced from the glass tube 102. The conversion portion 100 may further include a control system 402 communicatively coupled to the conversion portion 100 and the at least one measurement device 360. It is to be understood and intended that the conversion portion 100 may include any of the features, processing stations, or operating parameters previously described herein for the conversion portion 100.

[0097] Referring again to FIG. 9, as previously described, the system 400 may include one or more measuring devices 360. The measuring devices 360 may enable 100% online inspection of various dimensional and cosmetic attributes of the glass tube 102, the glass article 103, or both, and provide real-time feedback to the system 400. The measuring devices 360 may be disposed at one or more measuring stations 218, which may be in the primary circuit, the secondary circuit, or both. Additionally or alternatively, in an embodiment, the conversion unit 100 may include one or more measuring devices 360 disposed at one or more processing stations 106 that are not measuring stations 218, such as, but not limited to, the heating station 202, the forming station 204, the separating station 206, the cooling station, the punching station, or other types of processing stations. Additionally or alternatively, the one or more measuring devices 360 may be coupled to one of the multiple holding units 130 and translated with the holding unit 130 through the multiple processing stations 106. In an embodiment, the conversion unit 100 includes multiple measurement devices 360, each of which may be coupled to a processing station 106 and / or to one of the multiple holders 130 for translation with the holder 130 and glass tube 102 through the multiple processing stations 106. Additionally or alternatively, the measurement devices 360 may be located downstream of the conversion unit 100, such as in a quality control station downstream of the conversion unit 100.

[0098] The measurement device 360 ​​may be operable to measure one or more attributes of the glass tube 102, one or more attributes of a feature of the glass article partially formed at the working end 150 of the glass tube 102, one or more attributes of each glass article 103 produced from the glass tube 102, or a combination thereof. The attributes of the plurality of glass tubes 102, glass articles 103, partially formed glass articles 103, or combinations thereof may include one or more temperatures of the glass tube 102, one or more dimensions of the glass tube 102, one or more dimensions of the plurality of glass articles 103 or a feature of the glass article 103 partially formed at the working end 150 of the glass tube 102, one or more cosmetic attributes of the plurality of glass articles 103, or a combination thereof. In an embodiment, the measurement device 360 ​​may be positioned and operable to measure one or more attributes of the glass preform at the working end 150 of the glass tube 102. The glass preform refers to the heated portion of the glass tube 102 at the working end 150 of the glass tube 102 after the heating station 202 and before the forming station 204. In other words, the glass preform refers to the working end 150 of the glass tube 102 in a condition suitable for forming (e.g., viscosity of the glass). Because the glass preform is the portion of the glass tube 102 undergoing transformation, any reference herein to attributes of the glass tube 102 is intended to include attributes of the glass preform as well. Attributes of the glass preform may include the temperature of the glass, dimensions of the preform, other attributes of the preform, or combinations thereof. Other attributes of the glass tube 102, the glass article 103, or both may be measured and are contemplated by the present disclosure.

[0099] The one or more measuring devices 360 may include any measuring device capable of measuring one or more dimensions, temperature, or cosmetic attributes of the glass tube 102 and / or the glass article 103 produced therefrom. The measuring device 360 ​​may include, but is not limited to, optical measuring systems, laser measuring devices, measuring devices using sound waves, other electromagnetic waves, or other measurement techniques. In an embodiment, the measuring device 360 ​​may include a thermal imaging system such as the thermal imaging system disclosed in U.S. Patent No. 10,773,989, filed March 22, 2018, entitled "SYSTEMS AND METHODS FOR MEASURING THE TEMPERATURE OF GLASS DURING TUBE CONVERSION," the contents of which are incorporated herein by reference in their entirety. The thermal imaging system may be operable to measure one or more temperatures or dimensions of the glass tube 102, the glass article 103, or both during heating or after heating to form the glass tube 102 into the glass article 103. Additionally or alternatively, the measurement device 360 ​​may include one or more dimensional measurement systems, such as one or more of a visual imaging system, a laser reflectometer, a laser gauge, an optical micrometer, or other measurement device operable to measure one or more dimensions of the glass tube 102, features of the partially formed glass article 103, the finished glass article 103, or combinations thereof. Other available measurement devices 360 that identify one or more temperatures, dimensions, cosmetic attributes, or combinations thereof of the glass tube 102, the glass article 103, or both are also contemplated.

[0100] Referring again to FIG. 1 , the system 400 for producing a plurality of glass articles from a glass tube 102 may further include a control system 402 communicatively coupled to the conversion unit 100. In particular, the control system 402 may be communicatively coupled to the measurement device 360 ​​and various controllers of the processing stations 106 of the conversion unit 100. The controllers of the conversion unit 100 that may be communicatively coupled to the control system 402 may include, but are not limited to, the burner arrangement 318, the fuel gas control valve 310, the oxygen control valve 312, the air control valve 314, the former actuator 326, the cooling fluid control valve 344, the main turret drive motor, a drive motor operatively coupled to the holder for rotation of the glass tube 102, a timer, a ventilation system, other controllers, or combinations thereof. The number and type of controllers may depend on the particular conversion unit 100 used and the number and type of processing stations 106 employed by the conversion unit 100.

[0101] The control system 402 may include one or more processors 404, one or more memory modules 406 communicatively coupled to the processors 404, and machine-readable, executable instructions 408 stored in the one or more memory modules 406. The machine-readable, executable instructions 408, when executed by the processors 404, may cause the system to automatically perform any of the operations and / or method steps described in more detail herein without the explicit mention of the machine-readable, executable instructions 408.

[0102] 11, in an embodiment, the system 400 may further include a display 430 operable to display a graphical user interface 432. The display 430 may be directly communicatively coupled to the control system 402 or may be in electronic communication with the control system 402 through the network 410. The display 430 may be a touch screen or may include one or more input devices (not shown) that allow a user to input information into the graphical user interface 432. The graphical user interface 432 may be operable to display information about the control system 402 and the control sequences. The graphical user interface 432 may also be operable to accept input from a user and forward the user input to the control system 402.

[0103] The control system 402 includes machine-readable, executable instructions 408 that, when executed by the processor 404, may cause the system 400 to automatically perform one or more steps of a control sequence for controlling the converter 100 in accordance with the model predictive control framework described herein. The control system 402 may provide feedback control of the converter 100.

[0104] As previously described, the model predictive control framework disclosed herein incorporates one or more models relating the processing parameters of the conversion unit 100 to attributes of the glass article 103, the glass tube 102, or both, to predict values ​​of the attributes from various processing parameter settings. A model for the conversion unit 100 is developed and then incorporated into a target control function to provide feedback control of the conversion unit. During operation of the conversion unit 100, the control system 402, through execution of one or more computer readable and executable instructions, may measure attributes of the glass article 103, the glass tube 102, or both, adjust the measured data, and then utilize a target control function to adjust the processing parameters of the conversion unit 100 to maintain the attributes of the glass article 103 within acceptable targets. In particular, the control system 402 may minimize the target control function that takes into account target values ​​of the attributes and differences between the measured values ​​of the attributes and the predicted values ​​of the attributes obtained from the models.

[0105] As already indicated, the first action when providing feedback control of the transformation unit 100 through a model predictive control framework is the development of one or more models that relate values ​​of one or more process parameters of the transformation unit 100 to one or more attributes of the glass article 103, the glass tube 102, or both. The models for the transformation unit may be developed through first principles considerations, performing a design of experiments procedure, a combination of these two approaches, or other methods. Typically, first principles will suggest a model structure, and a design of experiments may be used to confirm the model structure or to suggest changes to the model structure and determine the model coefficients. In an embodiment, the model structure of the operational model is linear, which may improve the computational efficiency of the model predictive control framework. However, it is understood that the model structure of the operational model may be nonlinear.

[0106] In embodiments, a model for the conversion unit 100 may be developed, entirely or at least partially, through an experimental design process. The experimental design process may utilize 100% real-time online measurements of various attributes of the glass tube 102 and / or glass article 103, and statistical data analysis tools to automatically develop an operating model for the conversion unit 100. In particular, the experimental design process of the present disclosure for determining an operating model may include developing a number of condition sets for the conversion unit 100, each condition set including a number of process parameter settings. The conversion unit 100 may then be operated under each condition set to produce a glass article 103. During the production of the glass article 103, various attributes of the glass article 103, the glass tube 102, or both may be measured in real time by the measurement device 360 ​​of the conversion unit 100. Each glass article 103 may be assigned a unique identifier that may be associated with the condition set used to produce the glass article 103 and the attributes measured for the glass article. Once all condition sets have been run, the data for the condition sets and the measured attributes may then be used to develop one or more operating models that correlate one or more of the processing settings for the conversion unit 100 to the measured attributes. The operating models may be used to automatically determine initial operating settings for each of the plurality of processing parameters of the conversion unit 100. Additionally or alternatively, the operating models may be incorporated into a target control function to provide feedback control of the conversion unit 100.

[0107] Each of the plurality of condition sets of the conversion unit 100 may include a plurality of settings for a plurality of processing parameters of the conversion unit 100. Each of the plurality of condition sets of the conversion unit 100 may be different from the other condition sets. Thus, each condition set may represent a unique group of settings for the processing parameters of the conversion unit 100. In some cases, such as after a former replacement, many of the processing parameters remain the same and only a subset of the condition sets is needed to redevelop the operating model for the forming station 204 of the conversion unit 100, the subset of the condition sets being one that includes a number of condition sets that is less than the total number of possible condition sets for the conversion unit 100.

[0108] The process parameters of each condition set may include, but are not limited to, overall part speed, rotational speed of the holding member 130, burner position at one or more heating stations 202 or burner heat output at the heating stations 202, position of the former 324 at one or more forming stations 204, timing of contact between the former 324 and the glass tube 102 at one or more forming stations 204, other process parameters, or combinations thereof. With reference to FIG. 3, the burner position of the burner 302 at the heating station 202 may include a vertical position (e.g., in the + / -Z direction of the coordinate axes of FIG. 3), a horizontal position (e.g., a position in the XY plane of the coordinate axes of FIG. 3) of the burner 302 with respect to the glass tube 102, or combinations thereof. The burner heat output of the burner 302 at the heating station 202 may include the position of one or more of the fuel gas control valve 310, the oxygen control valve 312, the air control valve 314, or combinations thereof, which may control the burner heat output of the burner 302.

[0109] 4, the former tool position of the former tool 324 at the forming station 204 may include the vertical position of the former tool 324 relative to the glass tube 102 (e.g., in the + / -Z direction of the coordinate axis of FIG. 3), the horizontal position (e.g., the position in the XY plane of the coordinate axis of FIG. 3), or a combination thereof. The horizontal position of the former tool 324 may refer to the horizontal position of the former tool 324 relative to the glass tube 102 when the former tool 324 is in the operating position. When in the operating position, the horizontal position of the former tool 324 may determine the pressure of the former tool 324 against the glass tube 102. The timing of contact between the former tool 324 and the glass tube 102 may be controlled by controlling the operation of the former tool actuator 326 to adjust the timing of moving the former tool 324 into and out of operation on the glass tube 102 at the forming station 204. As previously described, the timing of contact may be adjusted to control the total contact time at the forming station 204.

[0110] The process parameters associated with the heating station 202 and the forming station 204 are described herein to illustrate the subject matter of the present disclosure and are not intended to limit the process parameters that may be used. It should be understood that other process parameters associated with the separation station 206, the perforation station, the cooling station, or other process stations 106 of the conversion unit 100 may also be included in the condition set. The number and type of process parameters in each condition set of the conversion unit 100 may depend on the type of conversion unit 100, such as the number and type of process stations 106 of the conversion unit 100, the configuration of each process station 106, the process control devices (e.g., control valves, motorized adjustments, speed adjustments, etc.) present at each process station 106, or other factors.

[0111] Each processing parameter of the converter 100 may have a wide range of settings. The range of settings for each processing parameter may represent the full control range of the processing parameter, or may be a narrower range of settings that the converter 100 may be expected to produce glass articles of acceptable quality and yield. Each processing parameter may have a maximum and minimum value. The maximum and minimum values ​​for each processing parameter may be based on the total capacity or range of operation of the actuator or control device that includes the processing parameter, or may be based on values ​​of the processing parameter that are known to produce non-standard glass articles or other problematic conditions for the converter 100.

[0112] A range of settings for each processing parameter may be used to develop a plurality of condition sets for the conversion unit 100. In an embodiment, the range of settings for each processing parameter of the conversion unit 100 may be input into a design of experiments statistical software program, which may be operable to process the range of settings for each processing parameter of the conversion unit 100 to develop a plurality of condition sets. An example of a design of experiments statistical software program may be JMP® developed by SAS. Other commercially available statistical software packages or custom software and / or algorithms may also be used to develop a plurality of condition sets. The plurality of condition sets provide a set of "recipes" for the operation of the conversion unit 100. An example subset of condition sets is shown in Tables 2 and 3 of the Examples.

[0113] In an embodiment, an initial recipe is provided to or stored in one or more memory modules of the control system 402 of the converter 100. The initial recipe is associated with a particular type or size of glass article 103 and may include preset values ​​for each processing parameter of the converter 100. A condition set may be developed from the initial recipe by selecting settings for each processing parameter distributed over a range of values ​​centered around or including the preset value from the initial recipe. In other words, the condition set includes the preset value from the initial recipe for a particular processing parameter and may also include values ​​for the processing parameters on either side of the preset value.

[0114] Once condition sets have been developed for the conversion unit 100, the conversion unit 100 may be operated with each of the multiple condition sets. A certain number of glass articles 103 may be produced with the processing parameters of each condition set. As few as 60 glass articles or as few as 50 glass articles may indicate statistical significance between the condition sets. The control system 402 may be operable to produce more than 50 or more than 60 glass articles 103 with the processing parameters of each condition set for the conversion unit 100. In embodiments, the conversion unit 100 may be operated to produce 50-200, 50-100, 60-200, or 60-100 glass articles 103 for each condition set.

[0115] While the converter 100 is operated to produce glass articles 103 at each set of processing parameters, one or more attributes of the plurality of glass tubes 102, glass articles 103, or both may be measured using a measuring device 360. As previously described, the measuring device 360 ​​may be operable to measure one or more attributes of the plurality of glass tubes 102, glass articles 103, or both. The attributes may include, but are not limited to, one or more temperatures or dimensions of the glass tubes 102 or glass preforms, one or more dimensions of the glass articles 103, one or more cosmetic features of the glass articles 103, or combinations thereof. The attributes may also include intermediate dimensions of the partially formed glass articles during or after one or more forming stations. The measuring device 360 ​​may be communicatively coupled to the control system 402. The measurement device 360 ​​may be operable to transmit one or more signals to the control system 402 indicative of attributes of the glass tube 102, the glass article 103, or both measured by the measurement device 360.

[0116] Once all condition sets have been run through the conversion unit 100, one or more operating models for the conversion unit 100 may be developed based on the measured attribute or attributes and the multiple condition sets. The operating models may relate various processing settings to one or more attributes of the glass article 103.

[0117] The operating model may be generated through statistical analysis of the collected data on the measured attributes of the glass article 103 and the process parameters used to manufacture the glass article 103. In an embodiment, the operating model may be developed external to the converter 100 or the control system 402. Referring again to FIG. 11, the collected data on the measured attributes of the glass article 103 and the process parameters used to manufacture the glass article 103 may be exported from the control system 402 to an external computing device 420. The data may be exported from the control system 402 to the external computing device 420 via a network 410, which may be a wired or wireless network. The external computing device 420 may include statistical analysis software capable of processing the data received from the control system 402 and developing an operating model therefrom. An example of statistical analysis software may include "JMP" developed by SAS, although other statistical analysis software packages may be used. Once developed, the operating model may be transmitted from the external computing device 420 to the control system 402 for use in the operation of the transformer 100 and / or in the development of a target control function.

[0118] Following development of the operating model, an initial operating setting for each process parameter may be determined using the operating model, and the conversion unit 100 may then be operated at the initial operating setting. In an embodiment, the operating model may be input to the control system 402, such that the control system 402 is operable to determine an initial operating setting for each process parameter from the operating model. The term "initial operating setting" refers to the value of the process parameter to which the conversion unit 100 is initially set prior to feedback control using the model predictive control framework disclosed herein. In the model described below, the initial operating setting corresponds to the setting of the process parameter at k=0, where k is an integer indicating the number of iterations of the control sequence, as explained below. Thus, for the first iteration (k=1) of the control sequence, the value of the vector at iteration k-1, denoted by the term Act(k-1), is equal to the initial operating setting of the conversion unit.

[0119] The control system 402 is operable to receive specifications for the glass article 103 to be manufactured, which may include target values ​​for one or more attributes of the glass article 103, the glass tube 102, or both. The target values ​​may be one or more target dimensions, one or more target cosmetic attributes, or a combination thereof, for the glass article 103 to be manufactured or the glass tube 102 during various conversion stages. For example, the target values ​​for the attributes of the glass article 103 may include one or more specification values ​​from an international standard for glass articles 103, such as, but not limited to, ISO 8362-1 "Injection Containers and Accessories-Part 1: Injection Vials Made of Glass Tubing", 3rd Edition, 2009, pp. 12-15, the entire contents of which are incorporated herein by reference. Table 1 below illustrates an example specification from ISO 8362-1, providing dimensions for an injection vial (i.e., Model A) manufactured from glass tubing including a blowback-free neck finish.

[0120] [Table 1]

[0121] The machine-readable, executable instructions 408, when executed by the processor 404, may cause the control system 402 to automatically receive specifications for the glass article 103 to be manufactured and determine initial operating settings for each processing parameter of the conversion unit 100 from the specifications and an operating model developed for the conversion unit 100. The control system 402 may receive the specifications for the glass article 103 from a user input device or from an external computing device 420 communicatively coupled to the control system 402 via a network 410. The initial operating settings for the processing parameters may be determined by inputting values ​​for attributes of the glass article 103 from the specifications into the operating model and then calculating the initial operating settings. Operating the conversion unit 100 at the initial operating settings may include adjusting each processing parameter to the initial operating settings determined from the operating model.

[0122] The control system 402 may be operable to develop an operating model of the conversion unit 100. In particular, the control system 402 may include machine-readable, executable instructions 408 that, when executed by the processor 404, may cause the system to automatically operate the conversion unit 100 to convert the glass tube 102 into a plurality of glass articles 103, measure one or more attributes of the plurality of glass tubes 102, the glass articles 103, or both, using at least one measurement device 360, adjust process parameters of the conversion unit 100 to operate the conversion unit 100 under each of a plurality of condition sets, associate each of the plurality of glass articles 103 with one of a plurality of condition sets used to manufacture the glass articles 103 and one or more attributes of the glass tube 102, the glass articles 103, or both, measured by the measurement device 360, and develop one or more operating models based on the measured attributes and the condition sets. Each operating model may relate one or more of a plurality of process parameters to one or more measured attributes of the glass tube 102, the glass article 103, or both. The machine-readable, executable instructions 408, when executed by the one or more processors 404, may further cause the control system 402 to automatically determine initial operating settings for each process parameter based on the operating model and to operate the conversion unit 100 with each process parameter set to the initial operating settings determined from the operating model.

[0123] Following development of an operating model for the transformer 100, the operating model may then be incorporated into a model predictive control framework to provide feedback control of the transformer 100. In particular, the operating model may be incorporated into a target control function that provides an algorithm for feedback control of the transformer 100. The target control function may be based on a cost function for the transformer 100. In an embodiment, the target control function may be based on a weighted mean squared error cost function for the transformer 100. Other control algorithms for developing the target control function are also contemplated. If the target control function is based on a weighted mean squared error cost function, it penalizes deviations of attribute measurements from attribute target values, attribute variance levels (considered directly or addressed indirectly), and actuator movement magnitudes. The actuator movement magnitude refers to the magnitude of change in position of an actuator or control valve used to control a process parameter of the transformer 100. During a feedback control sequence to control the transformer, the target control function for the transformer 100 is minimized. In the case where the model is linear, this is accomplished by constrained standard quadratic programming. The solution to the optimization problem is applied, and the solution is developed to compensate for any differences between what is expected to occur and what actually occurs, and the entire optimization process is repeated again. Thus, the control sequence and minimization of the target control function are repeated in an iterative process to provide robust performance in the presence of significant modeling errors. Each iteration of the control sequence includes measuring attributes of the glass article and / or glass tube produced at the converter, recording the process parameter settings used to produce the glass article, processing a data set including the measurements and process parameter settings to generate a statistical characteristic matrix representing the measured values ​​of the attributes, substituting the process parameter settings and the statistical characteristic matrix representing the measured values ​​of the attributes into the target control function, determining updated settings for the process parameters by minimizing the target control function, and further adjusting the process parameters to the updated settings. The control sequence may then be repeated a number of times until the updated settings converge.

[0124] The target control function may be derived by first developing an operational model that provides an equation relating the settings of the process parameters to an estimate of the attribute of the glass article or glass tube. As previously described herein, the operational model may be developed through first principles, experimental design, or a combination of both. The operational model may be an equation in which the estimate of the attribute is the sum of an offset constant and at least one term that depends on the settings of the process parameters.

[0125] Second, a target control function may be developed using the operating model and various terms and constants from the operating model and substituted into an initial control function, such as, but not limited to, a mean squared error cost function for the transformer. The initial control function may be a function of the estimated values ​​for each attribute of the glass article 103, the glass tube 102, or both, the target values ​​for each attribute, and the settings of each process parameter. The operating model developed for the transformer 100 may be substituted into the initial control function for the estimated values ​​of the attributes. The operating model may then be solved for an offset constant to generate an offset constant function. The process of solving the operating model for an offset constant may include substituting the measured values ​​of the attributes, or a statistical characteristic that represents the distribution of the measured values ​​for each measured value of the attributes, into each estimated value of the attributes. The offset constant function may then be substituted into the modified control function for the offset constant to generate a target control function.

[0126] The derivation of an exemplary target control function according to the present disclosure will now be described in further detail. In an embodiment, the motion model for the transform unit 100 may be a linear model, such as the motion model provided in Equation 1 (Equation 1).

[0127]

number

[0128] In formula 1,

[0129]

number

[0130] is an nx1 vector of attribute estimates at iteration k+1 of the control sequence, where n is an integer equal to the number of attributes of the glass article and / or glass tube considered in the model predictive control framework, and k is an integer indicating the current iteration of the control sequence. The G term is a sensitivity matrix, where the sensitivities are obtained from developing the operating model (e.g., from a design of experiments process). Each sensitivity in the sensitivity matrix indicates the degree to which the value of the attribute changes in response to a change in a particular process parameter. The Act(k) term is an nx1 vector of process parameter settings (e.g., settings of various actuators and control valves) for iteration k of the control sequence. The α term is a vector of offset constants.

[0131] Although described herein as using a behavioral model that is a linear function, it is understood that a more complex behavioral model may be developed and substituted into the initial control function to derive the target control function. As previously described, the model predictive control framework of the present disclosure may compensate for significant errors between the behavioral model and the actual performance of the converter. Thus, it has been found that any errors in the behavioral model caused by simplifying the behavioral model to a linear model are easily compensated for by the model predictive control framework and feedback control of the converter 100 disclosed herein.

[0132] As previously described, the initial control function may be a mean squared error cost function that is a function of the attribute estimates, the attribute target values, and the process settings (e.g., actuator settings). The initial control function may also include weighting factors for the attribute differences and a penalty factor to penalize or reduce the magnitude of change in the process parameters at each iteration. In an embodiment, the initial control function may be an equation for the mean squared error cost function shown in Equation 2 (Equation 2):

[0133]

number

[0134] In Equation 2, Attrib targ The terms are vectors of target values ​​for attributes of the glass article and / or glass tube.

[0135]

number

[0136] The term Q is a vector of estimates of the attributes of the glass article and / or glass tube for iteration k+1 of the control sequence. T The Q term is a symmetric, usually diagonal, weighting matrix containing attribute weighting coefficients for the differences in attribute measurements from attribute target values. Note that it is conventional to show this weighting coefficient matrix as a self-multiplied arrangement matrix. This is not necessary, but it simplifies the formula. R T The R term is a symmetric, normally diagonal, weighting matrix of penalty coefficients for changes in process parameters (i.e., changes in actuator settings) from one iteration of the control sequence to the next. R T R acts as a closed-loop robustness adjustment lever that may be necessary if the model is substantially inaccurate. Adjusting the penalty coefficients for changes in process parameters to tune the target control function is described in further detail herein.

[0137]

number

[0138] In terms, the motion model in Equation 1 may be substituted into Equation 2 above to obtain Equation 3, which is an intermediate equation.

[0139]

number

[0140] The offset constant α may be estimated by solving Equation 1 for the offset constant (one step past, so the resulting attributes are known, not estimated), and then substituting the measured values ​​of the attributes at the current iteration of the control sequence into the estimated values ​​of the attributes. Thus, at iteration k, from Equation 1,

[0141]

number

[0142] is derived, which is an estimate obtained by measuring the attribute k times. This estimate of the offset constant is the means by which feedback is actually introduced into the control sequence to act as a form of integral control (similar to a standard PID proportional-integral-derivative controller) and eliminates steady-state errors. The equation estimating the offset constant may be substituted into Equation 3 to produce the target control function in Equation 4 (Equation 4):

[0143]

number

[0144] In Equation 4, only Act(k) is unknown. As previously described, Act(k) is an nx1 vector of process parameter settings (e.g., settings of various actuators and control valves) for iteration k of the control sequence. The control system 402 may be operable to solve Equation 4 for a value of Act(k) that minimizes a target control function J(k) for iteration k of the control sequence. The solution to Act(k) that minimizes the target control function of Equation 4 may be constrained by maximum and minimum values ​​of each setting of the process parameter. In other words, to determine a solution to Act(k) that minimizes the target control function of Equation 4, the following mathematical equation 5 (Equation 5) must also be satisfied for all values ​​of k (e.g., k and k-1, where k=1, 2, 3, 4, ...).

[0145]

number

[0146] In formula 5, Act min is the matrix of minimum values ​​of each processing parameter setting, and Act max is the matrix of maximum values ​​for each process parameter setting. The solution to this constrained optimization problem is obtained by standard quadratic programming.

[0147] Vector Attrib targ and Attrib measured is any attribute of the glass article 103 and / or glass tube 102 that can be manipulated by varying one or more process parameters, for which a sufficient behavioral model can be obtained for that manipulation. targ and Attrib measured can be any attribute of the glass article 103 and / or the glass tube 102 already described herein. targ and Attrib measured may include, but are not limited to, one or more dimensions of the glass article 103 and / or glass tube 102, one or more cosmetic attributes of the glass article 103, one or more chemical attributes of the glass, a range of any of these attributes, or a combination thereof. The range of an attribute may include a standard deviation or variance of the values ​​of the attribute. The variance of one or more attributes of the glass article and / or glass tube may be represented by Attrib targ or Attrib measured This can be accomplished by either: (1) directly modeling the variance response in a design of experiments process, or (2) modeling the maximum and minimum attribute values ​​and controlling them to desired set points.

[0148] As already mentioned, the target control function is the weighting coefficient matrix Q TThe attribute weighting coefficients may include attribute weighting coefficients such as Q. The attribute weighting coefficients may be used to weight the relative importance of various attributes of the glass article and / or the glass tubing in the target control function. Thus, deriving the target control function may further include developing an attribute weighting coefficient for each attribute of the glass article, the glass tubing, or both, and applying the attribute weighting coefficients to one or more terms of the target control function based on the average double error cost function. In an embodiment, the attribute weighting coefficients may be calculated based on the error between the attribute target value and the attribute estimate from the original average double error cost function (i.e., (Attrib est (k+1)-Attrib targ ) can be applied to the terms of the target control function representing

[0149] In an embodiment, the attribute weighting coefficients may be determined by considering the process capability index Cpk for one or more of the attributes of the glass article and the glass tube. The process capability index Cpk for an attribute may be determined by conventional calculation methods from the attribute's specification range spread, the estimated mean of the attribute's values, and the estimated variance (e.g., standard deviation, variance, etc.) of the attribute. Glass article and / or glass tube attributes with lower process capability index Cpk may have attribute weighting coefficients that weight attributes with higher relative importance in the target control function compared to attributes with higher process capability index Cpk. Similarly, glass article and / or glass tube attributes with higher process capability index Cpk may have attribute weighting coefficients that weight attributes with relatively lower importance in the target control function.

[0150] As already mentioned, the target control function is the penalty coefficient matrix R in Eq. TThe penalty coefficients may include penalty factors such as R. The penalty coefficients may be used to penalize the magnitude of change in one or more of the process parameters of the target control function for each iteration of the control sequence. Each penalty coefficient may indicate the amount of impact that a change in a particular process parameter has on the control response. In some examples, a large change in one or more process parameters may significantly impact one or more attributes, causing divergence or oscillation in the updated settings of one or more process parameters. The penalty coefficients may be used to reduce the magnitude of changes to one or more of the process parameters, which may be made in any one iteration of the control sequence. Thus, the penalty coefficients may be used to slow down the response speed of the feedback control by reducing the magnitude of changes that may occur in each iteration of the control sequence. The penalty coefficients may reduce or prevent the possibility of divergence or oscillation in the control response outside of the normal expected variation, but may slow the response time of the control method to changing operating conditions.

[0151] A penalty factor may be developed for one or more of the process parameters and applied to a term in the target control function that represents a change in the process parameter. The penalty factor may be developed based on simulation, trial and error, operator intuition and experience, or other considerations. A process parameter with a large penalty factor R will be penalized to a greater extent in the target control function, resulting in a smaller magnitude of allowable change in the process parameter with a high R for each iteration of the control sequence. Similarly, a process parameter with a small penalty factor R will be penalized to a lesser extent in the target control function, which may allow a larger magnitude of allowable change in the process parameter with a low R for each iteration of the control sequence. Thus, the penalty factor may be used as a tuning characteristic to tune the feedback control provided by the model predictive control framework.

[0152] Referring again to FIG. 11 , the target control function may be used in a method for providing feedback control of the conversion unit 100 for producing glass articles 103 from glass tubes 102. The control method for feedback control of the conversion unit 100 may include providing target values ​​during or after conversion for one or more attributes of the glass tubes 102, the glass articles 103, or both. The control method may include operating the conversion unit 100 to produce a plurality of glass articles 103 from a plurality of glass tubes 102, and measuring attributes of the plurality of glass articles 103 and / or the plurality of glass tubes 102. The control method may include recording settings of operating parameters of the conversion unit 100 for the attributes to generate a data set including measurements of the attributes and settings of the operating parameters. The control method may further include processing the data set to generate statistical characteristics of the distribution of the measurements for each attribute, and determining updated settings of each operating parameter from the statistical characteristics of the distribution of the measurements for each attribute, the target values ​​for each attribute, and the settings of the operating parameters. The updated settings for the operating parameters may be the values ​​of the settings that minimize the target control function developed for the transformer. The control method may further include adjusting each of the operating parameters of the transformer 100 to the updated settings.

[0153] The control method may further include repeating the steps of measuring the attribute, recording the setting for each processing parameter, processing the data set, determining an updated setting for each processing parameter, and adjusting each processing parameter to the updated setting. In an embodiment, the control method may be repeated continuously while the transform unit 100 is operating. In an embodiment, the control method may be repeated at least until the updated settings of the processing parameters have converged. Convergence of the processing parameters refers to adjusting the processing parameters to a range within which further changes in the processing parameters are within the normal expected variation of the process.

[0154] 12, a graphical illustration of a control sequence 500 for feedback control of the converter 100 is shown illustrating a method described herein. During one iteration of the control sequence 500 for feedback control of the converter 100, a conversion process 510 is performed to produce glass articles from glass tubes using the converter 100. During the conversion process 510, one or more attributes of the glass tubes, glass articles, or both are measured for a number of glass articles. Measurement values ​​512 for each attribute and process parameter settings 514 are output from the conversion process 510 and received by the control system 402. The measurements 512 for each attribute are processed in a data processing module 520 to remove outlying data points and to calculate one or more statistical characteristics 522 of the distribution of the measurements 512.

[0155] Following processing of the data including the attribute measurements 512, the process parameter settings 514 and at least one statistical characteristic 522 representative of the attribute measurements 512 are sent to a control module 530 along with a target value 524 of the attribute. The control module 530 includes a target control function, which in an embodiment may be based on a mean squared error cost function for the transform. The control module 530 receives the process parameter settings 514, the statistical characteristics 522 representative of the multiple attribute measurements 512, and the target value 524 of the attribute. The control module 530 then finds values ​​of updated settings of the process parameters that minimize the target control function. The updated settings 532 are then output from the control system 402 to the transform process 510 and the transform 100, where the process parameters are adjusted to the updated settings 532. The control sequence 500 may be repeated any number of times.

[0156] The control methods disclosed herein may control a single attribute by considering one or more process parameters, or may control multiple attributes by considering multiple process parameters. The control methods may be implemented fully or partially through the control system 402. In embodiments, the control methods and control sequences may be fully automated through the control system 402. For example, the control system 402 may be communicatively coupled to controllers that control each actuator and process parameter, and the control system 402 may make adjustments to change the process parameters to updated settings at the end of each iteration of the control sequence in a fully automated manner.

[0157] In other embodiments, the control methods and sequences may include some degree of interaction with an operator of the conversion unit 100, while other actions are performed using the control system 402. In an embodiment, the control system 402 may be operative to receive attribute measurements and process parameter settings from the conversion unit 100, process the data including the attribute measurements and process parameter settings, and determine updated values ​​for the operating parameters by minimizing a target control function. Following determination of the updated settings for the operating parameters, the control system 402 communicates the updated settings for the process parameters to an operator, who may then decide to adjust the process parameters of the conversion unit 100 to the updated settings. As previously noted, a human operator may have difficulty simultaneously considering more than a maximum of about three different operating parameters and about three different attributes. Thus, for conversion units that must consider multiple operating conditions and attributes in their control strategies, the control system 402 must be capable of performing the trial and error solutions required to minimize complex target control functions for multiple attributes of the glass article to determine updated settings for the operating parameters. 11 , in an embodiment, the control system 402 may include a display 430 including a graphical user interface 432 operable to facilitate interaction between the control system 402 and an operator of the converter 100. The control system 402 may enable an operator to input parameters, such as attribute weighting coefficients, penalty coefficients, target values ​​for attributes, operating model information, or other information, into the control system 402 for use in the model predictive control framework.

[0158] The target values ​​for the attributes of the glass article 103 and / or the glass tube 102 may be provided as a single target value, a range of target values ​​(e.g., by providing a maximum and minimum specification for the attribute or by providing a single target value and a tolerance), or both. Examples of a single target value may include a target mean, a target median, a target maximum, a target minimum, a target spread, or other single target value. A range of target values ​​may be represented by a target minimum and a target maximum, or by a combination of a single target value and a tolerance. The target value may refer to a dimension or cosmetic feature of the finished glass article 103, such as, but not limited to, those from ISO 8362-1 in Table 1. The target value may also include a target value for an intermediate state of the glass article 103, such as the temperature, dimensions, or cosmetic features of the glass tube 102, a glass preform at the working end 150 of the glass tube 102, a feature of the partially formed glass article at the working end 150 of the glass tube 102, or a combination thereof. The target values ​​for each attribute may be manually entered into the control system 402 by an operator, such as using a graphical user interface 432, or may be automatically obtained by the control system 402 over the network 410 from an external system, which may be a business management system or other system capable of storing specifications for the glass articles and / or glass tubes. In an embodiment, the control system 402 may be operable to store the target values ​​for the attributes in a memory module 406 of the control system 402.

[0159] 9 and 11, as previously described, a number of attributes of the glass tube 102, the glass article 103, or both may be measured by one or more measurement devices 360 during operation of the converter 100. The measured values ​​of the attributes and the settings of the process parameters may be stored in one or more memory modules 406 of the control system 402, such as a relational database stored in the memory module 406 of the control system 402. In an embodiment, the control system 402 includes machine-readable and executable instructions 408 that, when executed by the processor 404, may cause the control system 402 to automatically measure each attribute of the glass tube 102, the glass article 103, or both, and record the settings of the process parameters to generate a data set including the measured values ​​and the settings of the process parameters for each attribute. The machine-readable, executable instructions 408, when executed by the processor 404, may cause the control system 402 to automatically store the dataset in the memory module 406 or export the dataset to an external computing device 420.

[0160] For each iteration of the control sequence, attributes of the glass tube 102 and / or glass article 103 may be measured for a number of glass articles 103 produced by the converter 100 to generate a data set including a number of measurements of each attribute at the settings of the process parameters. The number of glass articles 103 measured at each iteration of the control sequence may be a statistically relevant number of glass articles 103. A statistically relevant number of glass articles 103 may be a number of glass articles 103 large enough to generate a distribution of measurements that represents the true underlying distribution (i.e., the distribution of measurements of all glass articles produced at the process parameters when the converter operates with uncertainty at those parameters). The number of glass articles 103 that constitute a statistically relevant number may be a function of the number of different process parameters incorporated into the model predictive control framework. In general, the more glass articles that are measured, the more accurate the data set and the more accurate the statistical characteristics of the distribution of measurements of each attribute. Determining the number of glass articles required to generate a statistically relevant data set is within the capabilities of one of ordinary skill in the art who understands the natural process variations by attribute.

[0161] The number of glass articles to be measured may also be balanced by the desired response time of the control sequence. Increasing the number of glass articles 103 to be measured increases the time allotted to measuring the attributes to generate a data set, thereby increasing the response time of the feedback control. Similarly, reducing the number of glass articles to be measured may reduce the response time of the feedback control sequence to control the conversion unit 100, but at the risk of compromising accuracy. Thus, the number of glass articles 103 or articles 103 for measuring the attributes of the glass articles 103 or glass tubes may be selected to balance accuracy and response speed of the feedback control. In some embodiments that consider multiple attributes and multiple process parameters in a model predictive control framework, each iteration of the control sequence or method may include measuring the attributes of the glass articles for at least one full tube length of glass tube conversion for all holding units 130 of the conversion unit 100. For example, in a converter 100 having 16 processing stations in the main turret, for a glass tube long enough to produce 20 to 30 glass articles 103 from each glass tube 102, the number of glass articles measured per iteration of the control sequence may be at least 320 to 480. However, the number of glass articles 103 measured may be more or less depending on the level of inherent attribute variability.

[0162] Upon obtaining the attribute measurements and the transformer processing parameter settings, the data set is processed to generate one or more statistical characteristics representative of the distribution of the measurements. The data set obtained from the transformer 100 may include measurements for each attribute and settings for each processing parameter. The method disclosed herein may include removing outlying data points from the data set of measurements for at least one attribute. After removing the outlying data points, the control method may include calculating a statistical characteristic of the distribution of the attribute measurements from the data set. The statistical characteristic for each attribute is a calculated value representative of the attribute measurements. The statistical characteristic of the distribution of the data set may be a mean, median, range, standard deviation, variance, or a combination thereof. In an embodiment, the statistical characteristic may be the mean or median of the distribution of the measurements. The control system 402, and in particular the data processing module 520 of the control system, includes computer readable and executable instructions 406 which, when executed by the processor 404, may cause the control system 402 to automatically process the data of measurements 512 to remove outlying data points and calculate one or more statistical properties of the distribution of measurements for each attribute. With reference to Figure 11, processing of the data set may be performed using the control system 402 or may be exported to an external computing device 420 which processes the data set and returns statistical properties.

[0163] 12, once the data set has been processed to generate a statistical characteristic of the distribution of the measurements of each attribute, the control method may then include determining updated settings for the process parameters from a target control function. Determining updated settings from the target control function may include sending settings for the target values ​​of the attributes, settings for the process parameters used to generate the measurements of the attributes, and the statistical characteristic representing the distribution of the measurements of each attribute to the control module and substituting these values ​​into the target control function. In an embodiment, the target control function may be the function provided herein as Equation 4. In the target control function of Equation 4, the target values ​​of the attributes are expressed as Attrib targ , and the statistical characteristics that represent the measurements of the attributes are stored in the vector Attrib measured(k), and further assign process parameter settings to a vector in Act(k-1).

[0164] Determining updated settings for the process parameters may further include solving the target cost function for values ​​of the process parameters that minimize the target control function. The control system 402 may be operable to determine values ​​of the process settings that minimize the target control function. The updated settings for the process parameters are the settings that minimize the target control function. In an embodiment, the control system 402 includes computer readable and executable instructions 406 that, when executed by the processor 404, may cause the control system 402 to automatically substitute target values ​​of the attributes, statistical characteristics representative of the measured values ​​of the attributes, and settings of the process parameters into the target control function, and further determine updated settings for the process parameters by finding values ​​of the settings that minimize the target control function. The updated settings may be determined using the control system 402, or the control system 402 may be operable to export the statistical characteristics, target values, process parameter settings, or a combination thereof to an external computing device 420, which may be operable to determine the updated settings and return the updated settings to the control system 402.

[0165] As previously described, each process parameter may have a maximum and minimum value based on the operable range of an actuator, control valve, servo motor, or other control device, or based on values ​​of the process parameter known to produce non-standard glass articles. Additionally, as previously described herein with respect to Equations 4 and 5, the solution to minimize the target control function may be constrained by maximum and minimum values ​​of each process parameter. In an embodiment, determining updated settings for the process parameters may include providing a target control function with constraints including maximum and minimum values ​​for each process parameter. In an embodiment, the control method may include providing maximum and minimum settings for each process parameter, and maintaining the updated settings for each process parameter in a range between the minimum and maximum settings for the process parameter.

[0166] Once updated settings for the process parameters are determined, the control method may include adjusting the process parameters of the conversion unit 100 to the updated settings. In an embodiment, the control system 402 includes machine-readable, executable instructions 408 that, when executed by the processor 404, cause the control system 402 to automatically send control signals indicative of the updated settings for each process parameter to one or more controllers of the conversion unit 100, where the control signals may cause the controllers to adjust the process parameters to the updated settings. The control signals may be indicative of the updated settings or may be changes in the settings of the process parameters compared to the previous settings at k-1. In an embodiment, the machine-readable, executable instructions 406, when executed by the processor 404, may cause the control system 402 to automatically display the updated settings on the display 430. In an embodiment, instead of automatically changing the process parameters using the control system 402, an operator may view the updated settings and change the process parameters.

[0167] As previously described, the control method may include repeating the control sequence one or more times. In an embodiment, the control sequence may be repeated continuously while the conversion unit 100 is operated to produce the glass article 103. In this example, the model predictive control framework and control system 402 continues to provide feedback control of the conversion unit 100 during operation. In an embodiment, the control sequence may be repeated a sufficient number of times to cause the updated settings to converge to values ​​within an acceptable range. The acceptable range of the updated settings refers to a range of process parameters that represent the normal expected variation in the process. If the iteration of the control sequence results in the updated settings converging, the final updated settings may be maintained and the iteration of the control sequence may end. In an embodiment, the control method is performed on a periodic control basis throughout the operation of the conversion unit 100, where "periodic control basis" refers to alternating operation of the conversion unit 100 between control periods and non-control periods. The non-control periods are periods of operation of the conversion unit during which the control method is not performed. In an embodiment, the non-controlled period may have a length equal to or greater than the length of a single iteration of the controlled sequence. In an embodiment, the non-controlled period may be at least as long as necessary to use up one set of glass tubes loaded into the converter 100.

[0168] As previously described, the model predictive control framework disclosed herein may compensate for some errors in the operating model or changes in the conversion process, materials, or operating environment that result in a mismatch between the operating model developed for the conversion unit 100 and the actual operation of the conversion unit. In some examples, including but not limited to process changes, tool wear, changes in input materials, process drift over time, the mismatch between the operating model embedded in the target control function and the actual operation of the conversion unit 100 may be large enough to cause a lack of control of the conversion process. The lack of control may be manifested in a divergence in the updated settings of the process parameters at each iteration of the control sequence, or in an oscillation in the updated settings beyond the effects of normal expected variations in the conversion process.

[0169] As previously mentioned, in some cases the target control function may be adapted to correct for mismatches that result in a degradation of the ability to control the transform. For example, in some cases a penalty coefficient matrix (R T Control may be regained by varying the penalty coefficients in (R). Changing the penalty coefficients for changes in one or more process parameters, such as by penalizing changes in the process parameters to a greater extent, may reduce the magnitude of any changes to those process parameters in each iteration of the control sequence, reducing the likelihood of overshooting that leads to divergence or oscillation. However, varying the penalty coefficients to reduce the magnitude of changes to the process parameters may slow the control response, thereby increasing the time required to compensate for any further changes in the process.

[0170] In embodiments, the methods disclosed herein may include repeating a control method (i.e., a control sequence) for multiple iterations and identifying conditions indicative of a reduced ability to control one or more aspects of the transform unit. The conditions indicative of a reduced ability to control the transform unit may include divergence or oscillation of updated settings for one or more of the processing parameters through multiple iterations of the control sequence. After identifying the conditions indicative of a reduced ability to control the transform unit, the method may include adjusting a penalty factor for one or more of the processing parameters. Adjusting the penalty factor may reduce the magnitude of change in the one or more processing parameters with each iteration of the method, thereby reducing the divergence or oscillation of the updated settings and restoring control of the transform unit.

[0171] At some point during operation of the converter 100, whether due to gradual changes in the conversion process over time or due to anticipated abrupt changes in the conversion process, a mismatch between the operating model for the converter and the actual operation of the converter may become large enough to require redevelopment of one or more of the operating models for the converter 100. In some cases, gradual changes in the process, such as equipment wear or other gradual changes, may eventually lead to a mismatch between the operating model and the actual converter performance, necessitating redevelopment of the operating model. The need for model redevelopment may be indicated by a degradation in the ability of the model predictive control framework to control the converter, such as exhibiting divergence in settings for the process variables, oscillations in the process variables beyond the expected natural process variations in the process, or excessive time required to converge updated settings for the process parameters (e.g., uneconomical number of iterations of the control sequence). Additionally, the operational model may be redeveloped to anticipate or in response to anticipated changes in the operation of the conversion unit 100, such as, but not limited to, changes in size, shape, or type of glass articles being manufactured, changes in product specifications and targets, known changes in glass composition or quality, anticipated changes in fuel gas composition, scheduled replacement of worn equipment (e.g., worn formers, burners, control valves, actuators, etc.), other scheduled maintenance activities, or other anticipated changes. In an embodiment, the redevelopment of the conversion unit's operational model is performed on a regular, periodic schedule, which may help improve the mismatch between the operational model and the actual conversion unit's performance before a degradation in the ability to control the conversion unit is experienced.

[0172] The behavioral model of the transform unit 100 may be redeveloped through first principles studies or by performing the design of experiments process described previously herein for developing the behavioral model. Following redevelopment of the behavioral model through first principles, a design of experiments process, or both, the target control function may be modified to incorporate the new behavioral model, such as by incorporating an updated sensitivity matrix G into Equation 4. In some embodiments, updating the target control function involves modifying the attribute weighting coefficient matrix (Q TQ), the penalty coefficient matrix (R T R), or a combination thereof. The attribute weighting coefficient matrix and the penalty coefficient matrix may be updated in situations involving expected changes in the process, such as, but not limited to, changes in the size, shape, or type of glass article being manufactured, replacement of worn equipment, changes in glass composition, and other expected changes. Once the operating model and target control functions have been updated, the control method may be restarted.

[0173] The development and / or redevelopment of an operating model in response to gradual or anticipated changes in the operation of the conversion unit 100 will now be described in more detail. Using the experimental design process previously described herein, an operating model of the conversion unit 100 may be developed and / or redeveloped to achieve rapid and efficient steady-state operation of the conversion unit 100 during start-up of a new conversion unit 100, during start-up of a new type, size or shape of glass article 103 with an existing conversion unit 100, during switching from one type or size of glass article 103 to another type or size of glass article 103, or combinations thereof. The experimental design process disclosed herein may also be used to develop or redevelop an operating model to retune the operation of the conversion unit 100 after replacement of one or more components of the conversion unit 100, such as the former 324 or burner 302, after a change in the fuel gas composition, after a change in one or more external inputs to the conversion unit 100, or combinations thereof.

[0174] Starting and / or switching the converter Referring again to FIG. 1 , the system 400 and method of the present disclosure may be used to quickly and efficiently determine the initial operating settings of the conversion unit 100 upon start-up or after a changeover of the conversion unit 100 to achieve steady-state operation. Start-up may refer to the initial operation of a new conversion unit 100 or the start of operation of an existing conversion unit 100 to produce a new type or size of glass article 103 not previously produced by the existing conversion unit 100. Changeover may refer to changing the size, shape or type of glass article 103 from one previously produced size, shape or type of glass article 103 to another previously produced size, shape or type of glass article 103. Changeover may also include changing the dimensions of the glass tube 102, changing the glass composition of the glass tube 102, changing the coefficient of thermal expansion (CTE) of the glass tube 102, or other changes in the physical or chemical properties of the incoming glass tube 102. As already mentioned, a change in the type, shape or size of the glass article 103 being manufactured, or a change in the type or size of the glass tube 102, can result in significant lost production time, typically due to the setting and adjustment of hundreds of processing parameters of the conversion unit 100 for the new type or size of glass article.

[0175] The development of a setup for a new converting section 100 or a new glass article 103 not previously produced can be a relatively long iterative process that relies heavily on a human operator to execute. The setup for a new converting section 100 or a new glass article 103 not previously produced can include a two step process: (1) burner setup, and (2) forming station setup. In particular, the burner setup for the converting section 100 typically begins with a suitable preform being produced at the working end 150 of the glass tube 102, followed by the setup of the forming station 204 to produce the final glass article 103. During the burner setup phase, the part speed of the converting section 100, the position of the burner 302, and the heat output of the burner 302 can be adjusted in the order of movement around the converting section 100 from the separation station 206 to the last heating station 202 before forming. These adjustments during burner setup may be based on general guidelines and targets for various dimensions of the glass tube 102, which may help evaluate the impact of these individual burner 102 parameters and interactions. For the most part, however, burner setup involves an iterative process to produce a glass preform suitable for forming.

[0176] Once the glass preform shape is developed, the setup of the forming station 204 is then developed. The relative vertical and horizontal placement of the former 324 affects the shape of the glass article 103 features. For glass vials, these features may include any of the attributes listed in Table 1 for ISO 8362-1, including but not limited to flange height, neck outside diameter, flange inside diameter, flange outside diameter, shoulder radius, or other features. The features and shapes are dependent on the type of glass article 103 being produced. There are general guidelines and targets for these various shapes and dimensions. However, meeting the dimensional tolerances required for pharmaceutical glass articles often requires significant iteration time to find the appropriate forming setup, which may include backtracking and even changing the burner setup.

[0177] As previously described, the system 400 and method of the present disclosure may include systematically operating the conversion unit 100 under multiple condition sets and developing an operating model based on the condition sets and the measured attributes of the glass tube 102, glass article 103, or both produced using the condition sets. The operating model developed for the conversion unit 100 to produce a particular glass article 103 may then be used to determine initial operating settings for the conversion unit 100 to produce a glass article 103 having desired attributes. This provides a systematic and automated setup process, reducing or eliminating reliance on machine operator judgment and speeding up the setup and changeover process of the conversion unit 100. The developed operating model for the conversion unit may then be incorporated into a target control function of a model predictive control framework to provide feedback control of the conversion unit 100.

[0178] In an embodiment, the system 400 may receive an input indicating a change in the type of glass article 103 produced by the conversion unit 100 from a first glass article to a second glass article. The change in the type of glass article 103 may include a change in size, shape, or both of the glass article 103, the glass tube 102, or both. In an embodiment, the second glass article may be a new type or size of glass article 103 not previously produced by the conversion unit 100. The input indicating the change in type of glass article 103 may be a manual input or an electronic signal received over the network 410 from an external system, such as a business management system. In an embodiment, the electronic signal includes an initial recipe for the second glass article, and multiple condition sets may be developed starting from the initial recipe. The system 400 may operate the conversion unit 100 to convert the glass tube 102 into a plurality of second glass articles, measure one or more attributes of the plurality of second glass articles using the measurement device 360, adjust the plurality of process parameters to operate the conversion unit 100 under each of the plurality of condition sets, associate each of the plurality of second glass articles with the condition set and the measured attributes used to produce the second glass article, and further develop one or more operating models for the second glass articles based on the measured one or more attributes and the plurality of condition sets for each of the plurality of second glass articles. The system 400 may further determine initial operating settings for each of the plurality of process parameters based on the one or more second operating models, and further operate the conversion unit 100 with each of the plurality of process parameters set to the initial operating settings determined from the one or more second operating models. Although described in terms of switching from a first glass article to a second glass article, or from one type or size of glass tubing to another type or size of glass tubing, the system 400 and methods described in this paragraph may be equally applicable when starting up a new conversion unit 100 (e.g., setting up a first glass article on a new machine).

[0179] In an embodiment, the converting unit 100 may be set up in stages. The first stage of the setup may be a burner setup to determine initial operational settings of process parameters for the heating station 202 to produce a glass preform having target attributes such as temperature, viscosity, dimensions, or a combination thereof. Once the converting unit 100 is set up to produce a suitable glass preform at the working end 150 of the glass tube 102, the second stage of the setup may be a forming setup to determine initial operational settings of process parameters for the forming station 214 to form the glass preform into the features of the glass article 103.

[0180] The disclosed method includes the steps described below, which may be performed by the system through execution by the processor 404 of the machine readable and executable instructions 408. The method may include providing a first subset of the set of conditions that may include process parameters related to producing a glass preform at the working end 150 of the glass tube 102 at one or more heating stations 202. The method may further include providing a second subset of the set of conditions that may include process parameters related to forming the glass preform at the working end 150 of the glass tube 102 into one or more features of the glass article 103 at one or more forming stations 204. The set of conditions may be developed in any manner disclosed herein, including, but not limited to, by inputting various process parameters into a design of experiments statistical software program. In an embodiment, the first subset of condition sets and the second subset of condition sets may be developed outside of the system 400 and then uploaded to the control system 402 as “recipes” that the control system 402 may then run automatically through the conversion unit 100.

[0181] The method may further include operating the conversion unit 100 at each of the first subset of condition sets, measuring one or more preform attributes of the glass preform at the working end 150 of the glass tube 102, and developing one or more preform operating models based on the measured one or more preform attributes and the first subset of condition sets. The preform operating models may be developed by any of the operating model development methods disclosed herein. The preform operating models may describe relationships of various burner input parameters (burner flow rate, position, and rotational speed of the holding unit 130) to one or more dimensions and / or attributes of the glass preform, such as a temperature profile. The method may further include determining initial operating settings for each of the plurality of processing parameters in the heating station 202 based on the preform operating models, and operating the conversion unit 100 with each of the plurality of processing parameters in the heating station 202 set to the initial operating settings determined from the preform operating models.

[0182] In an embodiment, the preform operating model may be developed using an external computing system 420 (FIG. 11) outside of the control system 402 using a statistical software program such as “JMP” or other statistical software. Once developed by the external computing system 420, the preform operating model may be used to determine the best initial operating settings for the heating station 202, and the initial operating settings may be uploaded to the control system 402 as a recipe. In an embodiment, the preform operating model may be developed internal to the control system 402. The control system 402 includes machine readable and executable instructions that, when executed by the processor 404, may cause the control system to perform a statistical analysis of a set of conditions and measured attributes of the glass preform to develop the preform operating model. The machine-readable, executable instructions, when executed by the processor 404, may further cause the control system 402 to automatically determine initial operating settings for the heating station 202 from the preform operating model and automatically adjust processing parameters of the heating station to the initial operating settings developed from the preform operating model.

[0183] Once the appropriate glass preforms are set up, a similar process may be performed on the forming setup to determine initial operating settings for the plurality of forming stations 204. The method of the present disclosure may further include operating the converting unit 100 at each of the second subset of condition sets including processing parameters for the forming stations 204 while operating the converting unit 100 with the processing parameters of the heating station 202 set to the initial operating settings. The method may further include measuring one or more attributes of the manufactured plurality of glass articles 103, developing one or more forming operating models based on the measured attributes of the plurality of glass articles 103 and the second subset of condition sets, determining initial operating settings for each of the plurality of processing parameters at the forming stations 204 based on the forming operating models, and operating the converting unit 100 with each of the plurality of processing parameters of the forming stations 204 set to the initial operating settings determined from the one or more forming operating models. Similar to the preform motion model, the forming motion model may be developed in an external computing system 420 or internally in the control system 402 .

[0184] The control system 402 further includes machine-readable and executable instructions that, when executed by the processor, may cause the control system 402 to automatically update the target control function with information from the operating model developed during start-up or switchover of the conversion unit 100. In embodiments, switching the conversion unit 100 includes changing operation from a first glass article to a second glass article, where the second glass article may be an article from a previous run of the conversion unit 100. In these embodiments, the operating model developed for the previous campaign may be pulled and the target control function updated with the previously developed operating model without having to redevelop the operating model. The ability to use previously developed operating models is made possible by the ability of the model predictive control framework to compensate for mismatches between the operating model and the actual operation of the conversion unit 100.

[0185] Component replacement In an embodiment, the control system 402 is operable to redevelop one or more operating models in response to a change in one of the components of the converter, such as, but not limited to, replacement of a worn former, burner, actuator, control valve, converter drive motor, drive train operatively connecting the converter drive motor to the main turret, or other components. As previously described, replacement of one or more components of the converter 100, or other scheduled or unscheduled maintenance activities, may introduce significant errors into the operating models, slowing the response time of the model predictive control framework or causing poor control of the converter by the model predictive control framework. After replacement of one or more components of the converter 100, updated operating settings for one or more of the process parameters calculated from minimization of the target control function may no longer produce glass articles 103 that meet quality specifications. For example, after replacing one or more formers 324 in the forming station 204 due to wear or malfunction of the formers 324, it may be necessary to adjust the placement of the new formers 324 since the replacement formers 324 have different dimensions compared to the worn formers 324. Replacement of the converter drive motor, components of the drive train operatively connecting the converter drive motor to the main turret, former actuators, burner actuators, burner control valves, burners, or other components of the converter 100 may result in further mismatches between the actual operation of the converter 100 and the attributes predicted from the operating model. High yields may be lost from this mismatch between the previous operating model and the actual operation of the converter 100 after replacement of one or more components.

[0186] In these cases, the system 400 and method of the present disclosure may automatically redevelop one or more of the operating models of the conversion unit 100 after replacement of one or more components of the processing station 106 of the conversion unit 100. Method steps for automatically redeveloping one or more operating models of the conversion unit 100 after replacement of a component will now be described for the case of replacement of the former 324. However, it should be understood that the method steps may be used to achieve similar effects in redeveloping one or more operating models of the conversion unit 100 after replacement of any other components, such as, but not limited to, the burner 302, the burner arrangement 318, the control valves (fuel gas control valve, oxygen control valve, air control valve), the former actuator 326, the holding unit 130, the conversion unit drive motor, drive train components, or other components.

[0187] The method disclosed herein for automatically redeveloping one or more operating models of the conversion unit 100 after replacing a component of the conversion unit 100 may include identifying one or more defective components in one or more processing stations 106 of the conversion unit 100. A defective component may refer to a component that is malfunctioning or worn out and no longer produces a glass article 103 that meets the dimensional and / or cosmetic criteria for the glass article 103. The method may further include replacing the defective components in the one or more processing stations 106 with one or more replacement components. The defective components may be any components in any processing station 106 previously described herein. In an embodiment, the method may include identifying one or more defective formers in one or more forming stations 204 of the conversion unit 100. A defective former may refer to a former 324 that no longer produces a glass article 103 that meets the dimensional criteria for the glass article 103. A defective former may be one that is worn out or broken. The method may further include replacing the defective former with one or more replacement formers.

[0188] After replacing the defective component with a replacement component, the method may further include providing a plurality of condition sets for the conversion unit 100, operating the conversion unit 100 to convert the glass tube 102 into a plurality of glass articles 103, measuring one or more attributes of the plurality of glass articles 103, the glass tube 102, or both, adjusting one or more of the plurality of processing parameters of the condition sets to operate the conversion unit 100 at each of the plurality of condition sets, and associating each of the plurality of glass articles 103 with the condition set and the measured one or more attributes used to produce the glass articles 103. The method may further include updating the operating model based on the measured one or more attributes and the plurality of condition sets for each of the plurality of glass articles 103 to generate a set of updated operating models that take into account the replacement component. Initial operating settings for each of the plurality of processing parameters may be determined from the updated operating model. The processing parameters may be adjusted to the initial operating settings, and the conversion unit 100 may be operated at the new initial operating settings to produce the glass articles 103. The method further includes updating the target control function with information from the updated motion model, which may be, but is not limited to, updating the sensitivity matrix G of the target control function of Equation 4.

[0189] In an embodiment, replacement of a component, such as the former 324, may not require adjustment of all of the process parameters or operational models of the conversion unit 100. For example, replacement of the former 324 does not affect the glass preform, and operational models and process parameters associated with the upstream heating station 202 may not require redevelopment and adjustment. Thus, automatic redevelopment of the operational model of the conversion unit 100 after a component replacement may require adjustment of only a subset of the operational model for the conversion unit 100. In an embodiment, the multiple condition sets for the conversion unit 100 may include settings for a subset of multiple process parameters of the conversion unit 100. The subset of process parameters may include process parameters for the process station 106 in which the component was replaced, and optionally any downstream process stations 106. In an embodiment, the component exchange is a former exchange, and the subset of process parameters of the conversion section 100 may include the location (in all three dimensions of the coordinate system of the drawing) of the one or more replacement formers, the timing of contact of the one or more replacement formers with the glass tube 102, or both. The timing of contact may include the total contact time that the replacement formers are in contact with the glass tube 102.

[0190] In an embodiment, a method for automatically redeveloping one or more operating models of the conversion unit 100 following a component replacement or other machine malfunction may be performed using a system 400 including a control system 402. The control system 402 may include machine readable and executable instructions 408 that, when executed by the processor 404, may further cause the control system 402 to automatically receive input indicative of a change in one or more components of at least one processing station 106 of the conversion unit to develop a subset of the set of conditions. The subset of the set of conditions may include settings for a subset of the processing parameters of the conversion unit 100 that are related to or affected by the change in the component of the conversion unit 100. In an embodiment, the change in the component is a replacement of one or more formers 324, and the subset of the set of conditions may include settings for a subset of the processing parameters for the forming station 204 that replaced the former and / or for the processing stations 106 downstream of the forming station 204.

[0191] The machine-readable, executable instructions 408, when executed by the processor 404, may further cause the control system 402 to operate the conversion unit 100 to convert the glass tube 102 into a plurality of glass articles 103, measure one or more attributes of the plurality of glass articles 103, the glass tube 102, or both, using at least one measurement device 360, adjust one or more of the plurality of process parameters, operate the conversion unit 100 under each of the subset of condition sets to associate each of the plurality of glass articles 103 with the set of conditions used to produce the glass articles 103 and the measured attributes for each glass article 103, and further update one or more operating models based on the measured attributes for each of the plurality of glass articles 103 and the subset of condition sets to generate one or more updated operating models for the conversion unit 100. The updated operating models may be used by the control system 402 to determine initial operating settings for the plurality of process parameters after a component replacement. The control system 402 may then operate the converter 100 with process parameters set to the updated initial operating settings determined from the operational model. The control system 402 also includes machine-readable, executable instructions 408 that, when executed by the processor 404, may cause the control system 402 to automatically update the target control function with information from the updated operational model.

[0192] Identifying Malfunctioning Conditions In an embodiment, the control system 402 is operable to identify one or more malfunctions with the conversion unit 100, which may be, but are not limited to, malfunctions of components of the conversion unit 100, the wrong size glass tube 102 loaded into the conversion unit 100, a missing glass tube 102, an interruption in burner gas flow, an interruption in operation of the ventilation system, other malfunctions, or combinations thereof. Methods disclosed herein may include measuring one or more attributes for each of the plurality of glass articles 103 while operating the conversion unit 100 at operational settings for each of the plurality of process parameters, determining one or more expected attributes for each of the plurality of glass articles 103 from the plurality of process settings and operating models, and comparing the one or more measured attributes for each of the plurality of glass articles 103 to the one or more expected attributes to identify deviations from normal steady-state operation of the conversion unit 100. The method may further include identifying a malfunction of the conversion unit 100 based on a comparison of one or more measured attributes of each of the plurality of glass articles 103 with one or more expected attributes based on the operating model, correcting the malfunction of the conversion unit 100, and repeating the development of the model after correcting the malfunction.

[0193] In any of the methods disclosed herein, the conversion portion 100 may include a plurality of holding portions 130, and the methods disclosed herein may include the steps of securing one of the plurality of glass tubes 102 to each of the plurality of holding portions 130, and advancing each of the plurality of holding portions 130, and the glass tubes 102 disposed therein, through a plurality of processing stations 106.

[0194] In any of the methods disclosed herein, each of the plurality of processing stations 106 of the conversion unit 100 may be a fixed position, and the method may include indexing the glass tube 102 sequentially through each processing station 106. Alternatively, in an embodiment, the conversion unit 100 in any of the methods disclosed herein may be a continuous conversion unit, and the method may include advancing the glass tube 102 sequentially through the plurality of processing stations, where each of the plurality of processing stations may move in coordination with the translation of the glass tube 102 during the active time.

[0195] Referring to FIG. 11, in an embodiment, a system 400 includes a distributed computing environment, which may include the conversion unit 100, a control system 402, a network 410, and one or more external computing devices 420. The control system 402 may communicate with the external computing device 420 via the network 410. One or more steps of the method disclosed herein may be implemented using the external computing device 420 alone or in combination with the control system 402. Although shown in FIG. 11 as being directly communicatively coupled to the conversion unit 100, it should be understood that the control system 402 may further communicate with the conversion unit 100 through the network 410. The network 410 may be a wired or wireless network. In an embodiment, the network 410 may be a cloud network.

[0196] The embodiments of the present disclosure may be implemented in hardware and / or software (including firmware, resident software, microcode, etc.). The control system 402 of the converter 100 and / or other controllers for the converter 100 may include at least one processor and a computer-readable storage medium (i.e., memory module) as previously described herein. The control system 402 may be communicatively coupled to one or more system components (e.g., converter 100, burner positioner 318, burner control valve, former actuator 326, measuring device 360, converter drive system, etc.) via any wired or wireless communication path. The computer usable or computer readable storage medium or memory module 406 is any medium that can contain, store, communicate, propagate, or transport a program, which may be used by or connected to an instruction execution system, apparatus, or device.

[0197] The computer usable or computer readable storage medium or memory module 406 may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, device, or propagation medium. More specific examples (non-exhaustive list) of the computer readable storage medium or memory module 406 may include an electrical connection having one or more wires, a portable computer diskette, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, and a portable compact disk read-only memory (CD-ROM). It should be noted that the computer usable or computer readable storage medium or memory module 406 may be paper or other suitable medium on which the program is printed, since the program may be captured electronically, for example, via optical scanning of the paper or other medium, and then compiled, interpreted, or otherwise processed as appropriate, and then stored in the computer memory.

[0198] The computer readable storage medium or memory module 406 may include machine readable and executable instructions 408 for performing operations of the present disclosure. The machine readable and executable instructions 408 may include computer program code, which may be written in a high level programming language, such as C or C++, which is convenient for development. Additionally, computer program code for performing operations of the present disclosure may also be written in other programming languages, such as, but not limited to, interpreted languages. Some modules or routines may also be written in assembly language or microcode to enhance performance and / or memory usage. However, software embodiments of the present disclosure are not dependent on the use of any particular programming language. It will be appreciated that the functionality of any or all of the program modules may be implemented using discrete hardware components, one or more application specific semiconductor circuits (ASICs), or a programmed digital signal processor or microcontroller. EXAMPLES

[0199] The following examples illustrate the operation of the disclosed systems and methods for producing a plurality of glass articles from a glass tube at a conversion station. The following examples are not intended to limit the scope of the disclosure.

[0200] Example 1: Development of a conversion unit operation model The following Example 1 illustrates the development of a working model of the conversion unit and the use of the disclosed system and method to initially set up a conversion unit 100 that converts a glass tube into a glass article. For a typical conversion unit, hundreds of different process parameters may be adjusted, and for illustrative simplicity, the examples herein involve only a subset of these process parameters. The glass tube in these examples was an aluminosilicate glass tube, such as VALOR® glass manufactured and sold by Corning Incorporated. The aluminosilicate glass tube may be further processed by annealing and / or ion-exchanging the converted glass tube. Although the examples use an aluminosilicate glass, the effectiveness of the systems and methods disclosed herein is not dependent on the type or composition of the glass.

[0201] The glass tubes were converted into glass articles using a converter that included 18 processing stations in a main circuit. Each processing parameter may have a preset value for producing a particular type or size of glass article. Development of a condition set may begin with the preset value as a starting point.

[0202] Burner Setup First, using the methodology of the present disclosure, operational settings were developed for three process parameters related to the burner setup of the conversion section 100. The three process parameters in Example 1 were the burner position of the first heating station upstream of the forming station, the burner position of the second heating station located between the first heating station and the forming station, and the rotational speed (chuck speed) of the holding section 130. A subset of 18 condition sets were developed using "JMP" experimental design statistical analysis software to model these three process parameters. The subset of 18 condition sets is shown in Table 2.

[0203] [Table 2]

[0204] The converter 100 was run under each set of conditions in Table 2 to produce 60 glass articles under each set of conditions. Five different dimensions / angles were measured for each glass article produced under each set of conditions. Each glass article was assigned an identifier / part number and the measured dimension / angle and set of conditions were further associated with each glass article in a relational database. Referring to FIG. 13, data for one of the dimensions measured for the glass article of Example 1 is plotted as a function of part number. As shown in FIG. 13, the data points are grouped by the set of conditions used to produce the glass article 103.

[0205] The data for the measured dimensions / angles and the set of conditions for each glass article were then analyzed using "JMP" statistical analysis software to develop several operating models that relate process parameters to each measured dimension / angle. Referring to Figure 14, a graph of the operating models developed from the data of Example 1 is shown for illustrative purposes. It should be understood that the control system 402 uses an empirical mathematical representation of the models, rather than a graphical representation, to determine the initial operating settings for the transforms.

[0206] Forming station setup The method of the present disclosure was then used to develop an operational model and initial operating settings for five process parameters for a forming setup including a third heating station and one forming station located immediately downstream of the third heating station. The five process parameters for the forming station setup of Example 1 included the burner relative gas flow rate at the third heating station, the burner position at the third heating station, the position of the first former at the forming station, the end time of contact of the first former with the glass tube, and the end time of contact of the second former with the glass tube at the forming station. In this example, the position of the second former is entirely dependent, through first principles, on the position of the first former. Therefore, only the position of the first former was modeled. The burner relative gas flow rate is the flow rate of the fuel gas delivered to the burner divided by the burner reference gas flow rate. A subset of 24 condition sets was developed using "JMP" experimental design statistical analysis software for modeling these five process parameters. The subset of 24 condition sets for the forming station setup of Example 1 is shown in Table 3.

[0207] [Table 3]

[0208] The converter 100 was run under each condition set in Table 3, and 60 glass articles were produced under each condition set. Two dimensions and two appearance attributes were measured for each glass article produced under each condition set. Additionally, the overall desirability characteristics of each glass article were determined from the measurements of the two dimensions and two appearance attributes. Each glass article was assigned a unique identifier / part number, and the measured attributes and condition sets were then associated with each glass article in a relational database with the unique identifier. Referring to FIG. 15, data for one dimension measured for the glass article of Example 1 is plotted as a function of part number. As shown in FIG. 15, the data points are grouped by the condition set used to produce the glass article 103.

[0209] The measured dimension / angle data and the condition set for each glass article were then analyzed using "JMP" statistical analysis software to develop several operating models relating process parameters to each measured dimension, measured cosmetic attributes, and overall desirability. Referring to Figure 16, a graph illustrating the operating models developed from the data of Example 1 is illustrated.

[0210] Example 2: Feedback control using a model predictive control framework In Example 2, a motion model for the conversion unit was developed and incorporated into the target control function of Equation 4, which was then used to provide feedback control of the conversion unit 100. For Example 2, glass vials were produced from glass tubes using the conversion unit described in Example 1 with 18 processing stations in the main turret. The glass tubes were the same as those in Example 1.

[0211] In the first step of Example 2, an operating model was developed through a Design of Experiments process. The operating model was developed for nine process parameters that considered thirteen attributes of the glass vials. The process parameters modeled included the vertical position of the former at the first forming station, the horizontal position of the former at the first forming station, the gas flow to the burner at the first heating station prior to the first forming station, the vertical position of the burner at the first heating station, the vertical position of the burner at the second heating station after the second forming station, the vertical position of the pin former at the second forming station after the second heating station, the working position of the BB former at the second forming station, the working position of the IB wheel former at the second forming station, and the working position of the OB wheel former at the second forming station. The glass tube attributes measured for each set of conditions included maximum flange outer diameter, minimum flange outer diameter, maximum flange inner diameter, minimum flange inner diameter, maximum flange height, minimum flange height, maximum top height, minimum top height, average eccentricity, average lower flange angle, average top flange angle, average shoulder angle, and average shoulder radius.

[0212] As described in Example 1, the converter was run under each set of conditions and the measured attributes and process parameters for each vial were associated with a unique identifier and stored in a relational database. Once all sets of conditions were run, the data was then analyzed using "JMP" statistical analysis software to develop multiple operating models that associated process parameters with each measured attribute of the glass vial. In Example 2, the operating model was a linear operating model in the form of Equation 1 to simplify incorporation of the operating model into the target control function. Now, referring to FIG. 17, normalized values ​​of the flange outer diameter, flange inner diameter, top height, and shoulder angle of the glass article were derived from the measured data and plotted against the predicted values ​​of those attributes using the operating model and process parameter settings. As shown in FIG. 17, the R values ​​were calculated with a high value of more than 0.9. 2The linear behavioral model developed in Example 2 provided good prediction of the attributes, as indicated by the values ​​of and the very low P values ​​of less than 0.001.

[0213] For Example 2, the converter was operated without feedback control for five months, during which mechanical wear and long-term changes created at least some additional mismatch between the operating model and the actual operation of the converter. In step 2 of Example 2, after five months, the original operating model was incorporated into the target control function of Equation 4.

[0214] Once the operating model was incorporated into the target control function, the control method of the present disclosure was repeated multiple times to provide feedback control of the converter. Initially, the process parameters were set to initial operating settings determined from the operating model. For each iteration of the control method, the converter was operated at the current settings of the nine process parameters previously described in Example 2. Thirteen attributes were measured for each glass article produced during operation of the converter. The attribute measurements and the operating settings for the nine process parameters for each glass article produced were stored in a database. For each iteration of the control method, the attribute measurements and the process parameter settings were recorded for a number of vials corresponding to exhausting the entire length of glass tubing at each holding section of the converter (e.g., 18 holding sections).

[0215] The data, including the attribute measurements and process settings, were then processed to remove outlying data points and to calculate statistical properties of the distribution of the measurements for each attribute. In Example 2, the statistical properties were the average of the measurements for each measured attribute. The average of the attribute measurements, the target value of the attribute, and the current settings of the process parameters were then fed into the target control function of Equation 4 as Attrib measured , Attrib target , and Act(k-1) terms were substituted. During data processing and solving the target control function, the converter was operated at the previous operating settings to keep the converter thermal environment constant.

[0216] The target control function was then solved for the updated settings of the process parameters (e.g., Act(k)) that minimized the target control function. The process parameters of the transform unit were then changed to the updated settings of the process parameters and the control method was repeated. In Example 2, the control method was repeated for four iterations.

[0217] Now, referring to FIG. 18, the sum of squares error (e.g., the value of the minimized target control function) is plotted as a function of the number of iterations of the control method. Reference numeral 1802 refers to the value of the target control function for Example 2. As shown in FIG. 18, the largest change occurs in the first iteration. Thus, the model predictive control framework disclosed herein may be capable of compensating for the mismatch between the behavioral model and the actual behavior of the conversion unit in as little as one iteration of the control method. In FIG. 18, for iterations 2 to 4, the variation in the value of the target control function may account for the natural variation in the conversion process.

[0218] 19, there is shown the normalized average of each of the 13 attribute values ​​measured in Example 2 as a function of the number of iterations of the control method. Overall, the values ​​of the attributes converged within four iterations of the control method, further demonstrating the effectiveness of the model predictive control framework disclosed herein.

[0219] 20 and 21, normalized measurements of the thirteen attributes studied in Example 2 are plotted as a function of time during four iterations of the control method. In FIG. 20, the normalized values ​​of flange outside diameter minimum 2002, flange outside diameter maximum 2004, flange inside diameter minimum 2012, flange inside diameter maximum 2014, flange height minimum 2022, flange height maximum 2024, top height minimum 2032, and top height maximum 2034 are plotted. As shown in FIG. 20, initially during iteration zero (k-0), the flange inside diameter maximum 2014 and top height maximum 2034 of the glass vial were both greater than the upper specification limit (USL) for these parameters. As time progresses through each iteration, the process parameters are adjusted to change the inside diameter and top height such that the flange inside diameter maximum 2014 and top height maximum 2034 decrease, returning them to the acceptable range between the upper specification limit (USL) and lower specification limit (LSL). In Figure 20, the corresponding adjustments in the flange outside diameter minimum 2002, flange outside diameter maximum 2004, flange inside diameter minimum 2012, and flange height may also be observed.

[0220] Referring to FIG. 21, a graph shows normalized values ​​of the average bottom flange angle 2102, the average top flange angle 2104, the average shoulder angle 2106, and the average shoulder radius 2108. As shown in FIG. 21, at the beginning during iteration zero (k-0), the average shoulder angle 2106 was less than the lower specification limit (LSL) for the average shoulder angle. As time progresses through each iteration, the process parameters are adjusted to change the shoulder angle to maintain the average shoulder angle 2106 in an acceptable range between the USL and the LSL. FIGS. 20 and 21 show that the model predictive control framework and control method disclosed herein can provide feedback control to the transformer to identify glass articles with out-of-specification attributes and return those out-of-specification attributes to and maintain them in the acceptable specification range.

[0221] Example 3: Feedback control using the model predictive control framework when further mismatches occur In Example 3, the behavioral model of Example 2 was incorporated into the target control function of Equation 4. The resulting target control function was then used in a model predictive control framework to provide feedback control to a converter different from the converter used to generate the initial behavioral model. The converter in Example 3 had the same number and same arrangement of processing stations as the converter in Example 2. The use of a different converter in Example 3, despite having the same number and arrangement of processing stations, was expected to result in larger errors and mismatches between the behavioral model of the converter in Example 2 and the actual behavior of the converter in Example 3. Using a behavioral model from one converter for feedback control of a different converter with the same number and configuration of processing stations as in Example 3 shows that the model predictive control framework described herein can compensate for larger mismatches between the behavioral model and the actual behavior of the converter.

[0222] The control method for Example 3 was the same as described for Example 2. Referring again to FIG. 18, reference numeral 1803 shows the value of the target control function for Example 3 as a function of the number of iterations of the control method. Although the operating model was developed for a completely different converter, as shown in FIG. 18, the model predictive control framework disclosed herein was able to compensate for the mismatch between the operating model and the actual operation of the converter in just one iteration of the control method. Thus, the model predictive control framework disclosed herein is robust enough to compensate for significant errors in the operating model without losing the ability to control the conversion process.

[0223] Although various embodiments of conversion units and systems and methods for providing feedback control of the conversion units for producing a plurality of glass articles from a glass tube have been described herein, it should be understood that it is contemplated that each of these embodiments and techniques may be used separately or in combination with one or more of the embodiments and techniques.

[0224] It will be apparent to those skilled in the art that various modifications and variations can be made to the embodiments described herein without departing from the spirit and scope of the claimed subject matter. Accordingly, this specification is intended to cover modifications and variations of the various embodiments described herein, provided such modifications and variations come within the scope of the appended claims and their equivalents.

[0225] Preferred embodiments of the present invention will be described below in detail.

[0226] EMBODIMENT 1 1. A method for controlling a conversion unit for producing a glass article from a glass tube, comprising: operating a conversion section including a plurality of processing stations to produce a plurality of glass articles from a plurality of glass tubes, the step of operating the conversion section including translating the glass tube through each of the plurality of processing stations in succession; providing a target value during or after the transformation for at least one attribute of the plurality of glass articles or the plurality of glass tubes; measuring the at least one attribute of the plurality of glass articles or the plurality of glass tubes during or after conversion; recording settings of at least one processing parameter of the transformer for the at least one attribute to generate a data set comprising measurements of the at least one attribute and the settings of the at least one processing parameter of the transformer; processing the data set to generate statistical characteristics of the distribution of the measurements of the at least one attribute; determining an updated setting for each of the at least one process parameter from statistical characteristics of the distribution of the measured at least one attribute, the target value of the at least one attribute, and the setting of the at least one process parameter, the updated setting of the at least one process parameter being the value of the setting that minimizes a target control function for the at least one attribute; adjusting each of said at least one processing parameter of said transform unit to said updated setting; The method includes:

[0227] EMBODIMENT 2 repeating the steps of measuring the at least one attribute, recording the settings for each of the at least one processing parameter, processing the data set, determining the updated settings for each of the at least one processing parameter, and adjusting each of the at least one processing parameter until the updated settings for each of the at least one processing parameter converge; 2. The method of embodiment 1, further comprising:

[0228] EMBODIMENT 3 The step of processing the data set comprises: removing outlying data points from the data set of measurements of the at least one attribute; calculating the statistical characteristics of the distribution of the measurements of the at least one attribute from the data set after removing the outlying data points; 2. The method of embodiment 1, comprising:

[0229] EMBODIMENT 4 4. The method of embodiment 3, wherein the statistical characteristic of the distribution of the dataset is a mean, a median, a range, a standard deviation, a variance, or a combination thereof.

[0230] EMBODIMENT 5 providing a specification range for each of the at least one attribute, the specification range for an attribute comprising: a minimum value for said attribute below which the glass article is deemed out of specification; and comprising a maximum value for said attribute above which said glass article is deemed out of specification; applying, to each of the at least one attribute in the target control function, an attribute weighting factor based on a spread of the specification range for each of the at least one attribute; 2. The method of embodiment 1, further comprising:

[0231] EMBODIMENT 6 determining said attribute weighting coefficients from said specification ranges for each of said at least one attribute; 6. The method of embodiment 5, further comprising:

[0232] EMBODIMENT 7 The process capability index C for each of the at least one attribute pk from the specification range; The attribute weighting coefficient for each of the at least one attribute is calculated as the process capability index C for the at least one attribute. pk determining based on 7. The method of embodiment 6, further comprising:

[0233] EMBODIMENT 8 developing a penalty coefficient for each of the at least one processing parameter; applying said penalty factor to each of said at least one process parameter in said target control function, said penalty factor acting to reduce the magnitude of changes to process parameters that have a greater impact on one or more of said at least one attribute; 2. The method of embodiment 1, further comprising:

[0234] EMBODIMENT 9 Repeating the method according to embodiment 8 for a number of iterations; identifying a divergence or oscillation of the updated settings for one or more of the at least one processing parameter indicative of reduced performance and controlling one or more aspects of the transformer; adjusting the penalty factor for one or more of the at least one processing parameter, the penalty factor being adapted to reduce the magnitude of change to the one or more processing parameters at each iteration of the method, thereby reducing divergence or oscillation of the updated settings; 9. The method of embodiment 8, further comprising:

[0235] EMBODIMENT 10 10. The method of embodiment 9, wherein the divergence is indicated by an oscillation or consistent increase in the target control function with each iteration of the method.

[0236] EMBODIMENT 11 providing a maximum setting and a minimum setting for each of said at least one processing parameter; maintaining the updated setting of each of the at least one process parameter within a range between a minimum setting and a maximum setting for that process parameter; 2. The method of embodiment 1, further comprising:

[0237] EMBODIMENT 12 The target control function includes a mean squared error cost function according to the following equation:

[0238]

number

[0239] where J(k) is the mean squared error cost function as a function of k, k is an integer indicating the current iteration that minimizes the mean squared error cost function; G is a sensitivity coefficient matrix representing the extent to which a change in each of said at least one process parameter causes a change in each of said at least one attribute; Attrib measured (k) is the vector of the statistical characteristics of the variance of the measurements of the at least one attribute during iteration k; Act(k) is the setting of the processing parameters at iteration k; Act(k-1) is the setting of the processing parameter at 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 coefficients for the error of the measured value of the attribute from the target value of the attribute; R T R is a symmetric weighting matrix of penalty coefficients for said variation of the at least one process parameter; 2. The method of embodiment 1.

[0240] EMBODIMENT 13 developing the target control function, 13. The method of embodiment 12, further comprising:

[0241] EMBODIMENT 14 The step of developing a target control function comprises: developing at least one model for a predicted value of the at least one attribute for each setting of the at least one process parameter, the at least one model including an equation where the predicted value of the at least one attribute is the sum of at least one term that depends on the setting of the at least one process parameter and an offset constant; providing an initial mean squared error cost function that is a function of the predicted value for each of the at least one attribute, the target value for each of the at least one attribute, and the setting for each of the at least one processing parameter; plugging the at least one model into the initial mean squared error cost function for the predicted value of the at least one attribute; solving the at least one model for the offset constant to generate an offset constant function, wherein solving the at least one model for the offset constant comprises substituting the statistical characteristics of the measured values ​​of each of the at least one attribute into the predicted values ​​of each of the at least one attribute; substituting the offset constant function for the offset constant of the initial mean squared error cost function to generate the target control function; 14. The method of embodiment 13, comprising:

[0242] EMBODIMENT 15 15. The method of embodiment 14, wherein developing the at least one model comprises performing an experimental design process, deriving the at least one model from first principles, or a combination thereof to generate the at least one model.

[0243] EMBODIMENT 16 periodically redeveloping the at least one model to account for process changes over time or in response to known changes in the operation of the converter; 15. The method of embodiment 14, further comprising:

[0244] EMBODIMENT 17 applying an attribute weighting factor for each of the at least one attribute to one or more terms of the target control function. 15. The method of embodiment 14, further comprising:

[0245] EMBODIMENT 18 applying a penalty coefficient for each of the at least one process parameter to one or more terms of the target control function; 15. The method of embodiment 14, further comprising:

[0246] EMBODIMENT 19 2. The method of embodiment 1, wherein determining an updated setting for each of the at least one processing parameter further comprises providing constraints on the at least one processing parameter including maximum and minimum values ​​for the at least one processing parameter.

[0247] EMBODIMENT 20 providing target values ​​for a plurality of attributes of the plurality of glass articles, the plurality of glass tubes, or both; measuring the attributes of the plurality of glass articles, the plurality of glass tubes, or both for the plurality of glass articles; recording settings of a plurality of processing parameters of the transformer for the plurality of attributes to generate a plurality of data sets, each of the plurality of data sets including measurements for each of the plurality of attributes over a period of time and the settings of each of the plurality of processing parameters of the transformer; processing each of the plurality of data sets to generate statistical characteristics of distributions for each of the plurality of attributes; determining updated settings for each of the plurality of process parameters from the statistical characteristics of the distribution of each of the plurality of attributes, the target values ​​for each of the plurality of attributes, and the settings for each of the plurality of process parameters, the updated settings for the plurality of process parameters being values ​​of the updated settings that minimize the target control function; adjusting each of the plurality of processing parameters of the transform unit to the updated settings for each of the plurality of processing parameters; 2. The method of embodiment 1, further comprising:

[0248] EMBODIMENT 21 A step of displaying a user interface on a display unit, 2. The method of embodiment 1, further comprising:

[0249] EMBODIMENT 22 accepting one or more user inputs from the user interface; varying the updated setting of the at least one processing parameter based on the one or more user inputs; 22. The method of embodiment 21, further comprising:

[0250] EMBODIMENT 23 2. The method of claim 1, wherein the at least one attribute includes one or more attributes of the glass article after transformation, one or more attributes of one or more features of the partially formed glass article at the working end of the glass tube, one or more attributes of the glass tube, one or more attributes of a preform at the working end of the glass tube, or a combination thereof.

[0251] EMBODIMENT 24 1. A system for producing a plurality of glass articles from a plurality of glass tubes, comprising: A conversion unit; Control system and Including, The conversion unit is a plurality of holders, each of which is operable to hold a glass tube and rotate the glass tube about a central axis of the glass tube; a plurality of processing stations including at least one heating station, at least one forming station, and at least one separating station; at least one measurement device operable to measure one or more attributes of each of the glass articles manufactured from the glass tube, each of the glass tubes, or both; Including, the conversion unit is operable to translate the glass tube through each of the plurality of processing stations in succession to produce the plurality of glass articles; The control system includes: a processor communicatively coupled to the converter, the processor communicatively coupled to the one or more processors, one or more memory modules communicatively coupled to the one or more processors, and machine-readable, executable instructions stored in the one or more memory modules; The machine-readable, executable instructions, when executed by the one or more processors, cause the control system to automatically: measuring at least one attribute of the glass article, the glass tube, or both with the at least one measurement device; recording settings of at least one processing parameter of the transformer for the at least one attribute to generate a data set including measurements of the at least one attribute and settings of each of the at least one processing parameter of the transformer; processing the data set to generate statistical characteristics of a distribution of the measurements of the at least one attribute; determining updated settings for the at least one process parameter from the statistical characteristics of the distribution of the measurements of the at least one attribute, the target value of the at least one attribute, and the settings for each of the at least one process parameter, the updated settings for each of the at least one process parameter being a setting value that minimizes a target control function for the at least one attribute; and adjusting the at least one processing parameter of the transform unit to the updated setting. A system that is. [Explanation of symbols]

[0252] 100 Converter 102 Glass tube 103 Glass items 106 Processing Station 112 Secondary Processing Station 130 Holding part 302 Burner 324 Forming tool 360 Measuring Equipment 402 Control System 430 Display section 520 Data Processing Module 530 Control Module

Claims

1. 1. A method for controlling a conversion unit for producing a glass article from a glass tube, comprising: operating a conversion section including a plurality of processing stations to produce a plurality of glass articles from a plurality of glass tubes, the step of operating the conversion section including translating the glass tube through each of the plurality of processing stations in succession; providing a target value during or after the transformation for at least one attribute of the plurality of glass articles or the plurality of glass tubes; measuring the at least one attribute of the plurality of glass articles or the plurality of glass tubes during or after conversion; recording settings of at least one processing parameter of the transformer for the at least one attribute to generate a data set comprising measurements of the at least one attribute and the settings of the at least one processing parameter of the transformer; processing the data set to generate statistical characteristics of the distribution of the measurements of the at least one attribute; determining an updated setting for each of the at least one process parameter from statistical characteristics of the distribution of the measured at least one attribute, the target value of the at least one attribute, and the setting of the at least one process parameter, the updated setting of the at least one process parameter being the value of the setting that minimizes a target control function for the at least one attribute; adjusting each of said at least one processing parameter of said transform unit to said updated setting; Including, The target control function includes a mean squared error cost function according to the following equation: [0010] where J(k) is the mean squared error cost function as a function of k, k is an integer indicating the current iteration that minimizes the mean squared error cost function; G is a sensitivity coefficient matrix representing the extent to which a change in each of said at least one process parameter causes a change in each of said at least one attribute; Attrib ｍｅａｓｕｒｅｄ (k) is the vector of the statistical characteristics of the variance of the measurements of the at least one attribute during iteration k; Act(k) is the setting of the process parameter at iteration k; Act(k-1) is the setting of the process parameter at 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 coefficients for the error of the measured value of the attribute from the target value of the attribute; R Ｔ The method of claim 1, wherein R is a symmetric weighting matrix of penalty coefficients for said variation of said at least one process parameter.

2. repeating the steps of measuring the at least one attribute, recording the settings for each of the at least one processing parameter, processing the data set, determining the updated settings for each of the at least one processing parameter, and adjusting each of the at least one processing parameter until the updated settings for each of the at least one processing parameter converge; The method of claim 1 further comprising:

3. The step of processing the data set comprises: removing outlying data points from the data set of measurements of the at least one attribute; calculating the statistical characteristics of the distribution of the measurements of the at least one attribute from the data set after removing the outlying data points; The method of claim 1 , comprising:

4. The method of claim 3 , wherein the statistical characteristic of the distribution of the data set is a mean, a median, a range, a standard deviation, a variance, or a combination thereof.

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

6. determining said attribute weighting coefficients from said specification ranges for each of said at least one attribute; The method of claim 5 further comprising:

7. The process capability index C of each of the at least one attribute ｐｋ from the specification range; The attribute weighting coefficient for each of the at least one attribute is calculated as the process capability index C for the at least one attribute. ｐｋ determining based on The method of claim 6 further comprising:

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

9. Repeating the method of claim 8 a number of iterations; identifying a divergence or oscillation of the updated settings for one or more of the at least one processing parameter indicative of reduced performance and controlling one or more aspects of the transformer; adjusting the penalty factor for one or more of the at least one processing parameter, the penalty factor being adapted to reduce the magnitude of changes to the one or more processing parameters at each iteration of the method, thereby reducing divergence or oscillation of the updated settings; Further comprising: The method of claim 8 , wherein the divergence is indicated by an oscillation or a consistent increase in the target control function with each iteration of the method.

10. providing a maximum setting and a minimum setting for each of said at least one processing parameter; maintaining the updated setting of each of the at least one process parameter within a range between a minimum setting and a maximum setting for that process parameter; The method of claim 1 further comprising:

11. The target control function is developing at least one model for a predicted value of the at least one attribute for each setting of the at least one process parameter, the at least one model including an equation where the predicted value of the at least one attribute is the sum of at least one term that depends on the setting of the at least one process parameter and an offset constant; providing an initial mean squared error cost function that is a function of the predicted value for each of the at least one attribute, the target value for each of the at least one attribute, and the setting for each of the at least one processing parameter; plugging the at least one model into the initial mean squared error cost function for the predicted value of the at least one attribute; solving the at least one model for the offset constant to generate an offset constant function, wherein solving the at least one model for the offset constant comprises substituting the statistical characteristics of the measurements of each of the at least one attribute into the predicted values ​​of each of the at least one attribute; substituting the offset constant function for the offset constant of the initial mean squared error cost function to generate the target control function; The method of claim 1 , further comprising the step of developing by:

12. 12. The method of claim 11 , wherein developing the at least one model comprises performing a design of experiments process, deriving the at least one model from first principles, or a combination thereof to generate the at least one model.

13. periodically redeveloping the at least one model to account for process changes over time or in response to known changes in the operation of the transformer; The method of claim 11 further comprising:

14. providing target values ​​for a plurality of attributes of the plurality of glass articles, the plurality of glass tubes, or both; measuring the attributes of the plurality of glass articles, the plurality of glass tubes, or both for the plurality of glass articles; recording settings of a plurality of processing parameters of the transformer for the plurality of attributes to generate a plurality of data sets, each of the plurality of data sets including measurements for each of the plurality of attributes over a period of time and the settings of each of the plurality of processing parameters of the transformer; processing each of the plurality of data sets to generate statistical characteristics of distributions for each of the plurality of attributes; determining updated settings for each of the plurality of process parameters from the statistical characteristics of the distribution of each of the plurality of attributes, the target values ​​for each of the plurality of attributes, and the settings for each of the plurality of process parameters, the updated settings for the plurality of process parameters being values ​​of the updated settings that minimize the target control function; adjusting each of the plurality of processing parameters of the transform unit to the updated settings for each of the plurality of processing parameters; The method of claim 1 further comprising:

15. 1. A system for producing a plurality of glass articles from a plurality of glass tubes, comprising: A conversion unit; Control system and Including, The conversion unit is a plurality of holders, each of which is operable to hold a glass tube and rotate the glass tube about a central axis of the glass tube; a plurality of processing stations including at least one heating station, at least one forming station, and at least one separating station; at least one measurement device operable to measure one or more attributes of each of the glass articles manufactured from the glass tube, each of the glass tubes, or both; Including, the conversion unit is operable to translate the glass tube through each of the plurality of processing stations in succession to produce the plurality of glass articles; The control system includes: a processor communicatively coupled to the converter, the processor communicatively coupled to the one or more processors, one or more memory modules communicatively coupled to the one or more processors, and machine-readable, executable instructions stored in the one or more memory modules; The machine-readable, executable instructions, when executed by the one or more processors, cause the control system to automatically: measuring at least one attribute of the glass article, the glass tube, or both with the at least one measurement device; recording settings of at least one processing parameter of the transformer for the at least one attribute to generate a data set including measurements of the at least one attribute and settings of each of the at least one processing parameter of the transformer; processing the data set to generate statistical characteristics of the distribution of the measurements of the at least one attribute; determining updated settings for the at least one process parameter from the statistical characteristics of the distribution of the measurements of the at least one attribute, the target value of the at least one attribute, and the settings for each of the at least one process parameter, the updated settings for each of the at least one process parameter being a setting value that minimizes a target control function for the at least one attribute; and adjusting said at least one processing parameter of said transform unit to said updated setting. A system that is.

Images