Apparatus and method for measuring the kinematic characteristics of the free fall of a glass parison in a glassware molding apparatus, and method for controlling such an apparatus

Abstract

The invention relates to a device and a method for measuring the kinematic characteristics of a glass gob free fall using four different linear cameras, each linear camera having an observation linear field intercepting the theoretical free fall path at a first high interception point and a first low interception point, respectively, and intercepting the theoretical free fall path at a second high interception point and a second low interception point, respectively, the first high interception point and the first low interception point being offset from each other according to the direction of the theoretical free fall path, the second high interception point and the second low interception point being offset from each other along the direction of the theoretical free fall path, the high optical axis and the low optical axis being different from each other in the projection of the plane perpendicular to the direction of the theoretical free fall path. The invention also includes a method for controlling a glassware molding apparatus.

Description

Technical Field

[0001] This invention relates to an apparatus and method for measuring the kinematic characteristics of a glass droplet falling freely in a glass molding apparatus. The invention also relates to a method for controlling a glass molding apparatus, particularly a control method for implementing at least one kinematic characteristic of a glass droplet falling freely, measured by the measurement method and / or using the measurement apparatus according to the invention. Existing technology

[0002] Document EP2356081 / WO2010047579 describes an apparatus and method for measuring the kinematic characteristics of a glass droplet in free fall within a glass molding apparatus. This apparatus utilizes two cameras positioned along two distinct but intersecting observation axes that converge at the same intercept point at the mold inlet, representing the theoretical vertical path of free fall. These cameras must be matrix cameras, ensuring that each camera captures at least two complete views of the same droplet at two different moments in a series of images. Each view of the droplet must be a complete view encompassed within a single acquisition cycle of the cameras. It is understood that comparing the two complete images thus acquired at once makes it possible to determine the kinematic characteristics of the glass droplet's free fall. Specifically, the document teaches the necessity of determining the module and direction of the droplet velocity but does not explain how this determination is made from these images. Note that acquiring complete matrix images of the droplet within a single acquisition time requires, on the one hand, the use of very high-performance matrix cameras, and on the other hand, simply a complete view of the droplet at the intercept point from the mounting position of each camera. However, within the framework of industrial equipment, this can be problematic because of the shearing and equipment in and around the equipment, meaning that such a complete view of the observation axis that allows for accurate measurement is not necessarily possible in an industrial environment, and in any case, not necessarily. This is often the case when observing the droplet directly below the shearing machine that cuts the droplet before it enters the dispenser.

[0003] Document DE10312550 describes an apparatus and method for measuring the geometric characteristics of a freely falling glass droplet in a glass molding machine. It implements a linear camera and two phototubes positioned to detect the droplet's passage through two distinct overlapping intercept points along the theoretical vertical free-fall path at the mold inlet. The presence of a single linear camera cannot determine the kinematic characteristics. The two phototubes only determine the velocity of the droplet along the theoretical vertical free-fall path.

[0004] Document EP1418158 describes an apparatus comprising two or three CCD cameras configured such that the field of view covered by these cameras covers the spatial range between the dropper and the funnel of the dispensing system. We found the same requirement in document US20130000359. Therefore, the implemented cameras are matrix cameras, and regarding document EP2356081, it should be noted that such complete vision is not necessarily possible in industrial environments.

[0005] In document US2017121207, one or two cameras capture one or two images of a droplet, and the captured one or two two-dimensional images of the droplet are loaded into an image processor.

[0006] Document US4205973 teaches that the path of two consecutive laser beams is interrupted as the droplet falls, making it possible to measure the velocities at the beginning and end of the droplet. A horizontal scan is then performed incrementally using one (or two) linear cameras, each increment representing a segment of equal size along the vertical axis of the droplet. The scan rate is monitored by a scan clock at a scan frequency that varies over time to account for the droplet's acceleration.

[0007] Document US5434616 cites previous literature by noting that such a device requires precise synchronization between the camera scanning speed and the droplet's movement, and that even slight variations in the droplet's speed will lead to errors in the measured dimensions. Therefore, this document teaches the use of a two-dimensional CCD camera to continuously capture "frozen" two-dimensional images of the droplet's path.

[0008] Reference JP3623329 uses a CCD camera to capture an image of a droplet. The weight of the droplet is calculated from this image. We deduce from the reference that this image is a two-dimensional image. This device cannot determine the kinematic characteristics along two horizontal axes.

[0009] Document WO-2016181071 describes a glass molding apparatus comprising several forming sections, a droplet dispenser, and a detection device including at least one photodetector, the at least one photodetector being arranged to detect light information circulating along a free optical axis, the free optical axis intercepting at least two specific portions of the droplet loading path corresponding to two different forming sections.

[0010] The document “AMELIORATION SIGNIFICATIVE DE PERFORMANCES PAR CONTROLEAUTOMATIQUE DU GOB”, Institute for Glass, Paris, France, Vol. 9, No. 2, April 1, 2003 (2003-04-01), pp. 6 / 07, XP001160130, ISSN: 0984-7979 mentions another device in the prior art.

[0011] The present invention aims to provide a measuring device and method that enables the measurement of the kinematic characteristics of a glass droplet in free fall within a glass molding apparatus. These characteristics must be thin enough to allow for the operation of the apparatus, for example, a shearing machine that drives the cutting of the droplet, while taking these measurements into account. We seek to determine at least one of the following characteristics:

[0012] • For at least two distinct segments of the droplet, the segments are sets of first and second horizontal translational amounts or velocities along distinct first and second horizontal axes between a high intercept point and a low intercept point, and / or

[0013] • The set of first and second rotational amounts or velocities of the droplet around a first and second horizontal axis, which are different from each other, between the high intercept point and the low intercept point, and / or

[0014] • The amount or velocity of the material droplet between the high and low cut-off points, and / or

[0015] • The path of at least one or more segments of the droplet based on the three dimensions of space.

[0016] The measuring apparatus and method must be feasible in an industrial environment that may include many components and accessories that could interfere with the visibility of the droplet in the free fall region. Summary of the Invention

[0017] For the purposes described above, the present invention provides an apparatus and method for measuring the kinematic characteristics of a glass droplet falling freely in a glass molding apparatus, and a method for controlling such apparatus, as defined in the claims.

[0018] Therefore, this invention proposes a method for measuring the kinematic characteristics of a glass droplet in free fall within a glass molding apparatus. The method includes measuring the droplet in a free fall region along the droplet loading path between the glass source and the forming cavity, wherein the droplet has a theoretically vertical free fall path in the free fall region, and the droplet has a starting end and a tail end, as well as a droplet length between the starting end and the tail end.

[0019] - At least four series of consecutive linear digital images are acquired using four different linear cameras. Each linear camera has a linear photoelectric sensor, a lens with an optical center, and an optical axis defining the observed linear field for the camera under consideration. Each image in a given series of linear images is a linear image of the observed linear field of the corresponding linear camera. The four series include a first high series acquired by a first high camera, a first low series acquired by a first low camera, a second high series acquired by a second high camera, and a second low series acquired by a second low camera. The linear images of the four series correspond to linear images of the first high linear field, the first low linear field, the second high linear field, and the second low linear field, respectively, such that:

[0020] • The first high linear field and the first low linear field each extend along the corresponding plane defined by the corresponding optical axis and the horizontal extension direction perpendicular to the corresponding optical axis. The first high linear field and the first low linear field each intercept the theoretical free fall path at the first high intercept point and the first low intercept point, respectively. The first high intercept point and the first low intercept point are offset from each other by a perpendicular offset according to the theoretical free fall path.

[0021] - The second high linear field and the second low linear field each extend along a corresponding plane defined by the corresponding optical axis and a horizontal extension direction perpendicular to the corresponding optical axis. The second high linear field and the second low linear field each intercept the theoretical free fall path at the second high intercept point and the second low intercept point, respectively. The second high intercept point and the second low intercept point are offset from each other along the direction of the theoretical free fall path.

[0022] - Each of the four series of linear images includes images of corresponding intercept points acquired along the first high observation axis, the first low observation axis, the second high observation axis, and the second low observation axis, respectively. The observation axis of each intercept point of the corresponding linear camera is contained within the linear field of the corresponding linear camera, passes through the optical center of the lens of the camera, and passes through the corresponding intercept point of the theoretical free fall path.

[0023] -The projection of the first and second highest observation axes together on a plane perpendicular to the theoretical free fall path direction forms a high observation deviation angle around the theoretical free fall path that is different from 0 degrees and different from 180 degrees.

[0024] - The first and second low observation axes together form a low observation deviation angle around the theoretical free fall path in the projection on a plane perpendicular to the direction of the theoretical free fall path, which is different from 0 degrees and different from 180 degrees.

[0025] The time deviation between any two images from the same series and between any two images from two different series is determinable.

[0026] The method includes: computer identification of high-linearity images and low-linearity images, each comprising images of the same given end of the droplet at the beginning and end of the droplet, and computer derivation of the time deviation between the acquisition of the high-linearity image and the acquisition of the low-linearity image from the identification.

[0027] -Based on the time deviation between the acquisition of high-linearity images and low-linearity images, and by applying the laws of free fall kinematics, the computer calculates the instantaneous vertical translational velocity of the droplet's given end as it passes through one of the high and low intercept points; and

[0028] - For an intermediate linear image that includes a segment of the droplet excluding its two ends, the image of the droplet contained in the intermediate linear image is matched with the corresponding segment of the droplet by applying the kinematics of free fall as a function of the instantaneous vertical velocity of the droplet at the point corresponding to the instantaneous vertical velocity of the droplet at the given end of the droplet, and the time elapsed between the acquisition of the intermediate linear image and the passage of the droplet at the point corresponding to the instantaneous vertical velocity of the droplet at the given end of the droplet.

[0029] Furthermore, the method includes a computer determining at least one of the following:

[0030] - For at least two distinct segments of the droplet, the set of the first and second horizontal translation amounts of the segments along distinct first and second horizontal measurement axes between the high and low intercept points, or the set of the first and second average horizontal translation velocities of the segments, and / or

[0031] - The set of the first and second rotational speeds of the droplet, or the set of the first and second average rotational speeds of the droplet, relative to each other around a first and a second horizontal axis that are different from each other between the high and low interception points, and / or

[0032] - The amount of droplet deformation or the average deformation rate of the droplet between the high and low cut-off points, and / or

[0033] - The path of at least one or more segments of the droplet based on the three dimensions of space.

[0034] The method according to the invention may also include one or more of the following optional features, individually or in combination.

[0035] In some cases, the method includes: a computer determining the positional deviation between two considered segments of a droplet as a function of, wherein the respective images of the two considered segments of the droplet are contained in two consecutive linear images of a given series of linear images:

[0036] - The number of linear images in a defined series, on one side including one of the two considered segments, and on the other side, the number of linear images between the starting linear image and the tail linear image of a defined series, the starting linear image and the tail linear image including images of the beginning and end of the droplet, respectively.

[0037] - The instantaneous vertical translational velocity of the given end of the droplet when the given end of the droplet passes through the intercept points corresponding to a defined series;

[0038] - The acquisition frequency of a defined series of linear images; and

[0039] - Gravitational constant.

[0040] In some cases, the method includes: based on a first high-series linear image and a first low-series linear image, and a second high-series linear image and a second low-series linear image, respectively, a computer calculates a first instantaneous vertical translational velocity and a second instantaneous vertical translational velocity of the given end of the droplet during the passage of the given end of the droplet at one of a high intercept point and a low intercept point, and calculates the instantaneous vertical translational velocity of the given end of the droplet during the passage of the given end of the droplet at the intercept point as the average of the first instantaneous vertical translational velocity and the second instantaneous vertical translational velocity during the passage of the given end of the droplet at the intercept point.

[0041] In some cases, the method includes: calculating the droplet height between the starting and ending points by summing the positional deviations of all consecutive linear images in a given series from the starting linear image to the ending linear image.

[0042] In some cases, the method includes: for a given set of segments of a droplet, the computer determines:

[0043] - In the first high series of the first high linear image and the first low series of the first low linear image corresponding to each segment, determine the first horizontal translation amount of each segment of the set between the first high cut-out point and the first low cut-out point, and between the high linear image and the low linear image along the first horizontal measurement axis;

[0044] - In the second high series of the second high linear image and the second low series of the second low linear image corresponding to each segment, determine the second horizontal translation amount of each segment along the second horizontal measurement axis between the first high intercept point and the first low intercept point;

[0045] Furthermore, the method includes the step of computationally deriving the horizontal translation amount for each part of the set:

[0046] *Two horizontal components of the average translational velocity of the droplet along two different horizontal axes between the high and low intercept points; and / or

[0047] * The droplet rotates around two horizontal axes at two angles between the high and low cut-off points; and / or

[0048] *The deformation of the droplet during its fall between the high and low cut-off points.

[0049] In some cases, the method includes: computer measurement of at least one geometric dimension of the droplet from the following categories:

[0050] -The first diameter of the droplet along the first horizontal direction

[0051] -A second diameter along a second horizontal direction different from the first horizontal direction.

[0052] - The length or height of the droplet

[0053] - The volume of the droplet.

[0054] In some cases, determining the horizontal translation of a segment between a high cutoff point and a corresponding low cutoff point involves detecting the position of at least one point of the segment in both the corresponding high-linearity image of the high series and the corresponding low-linearity image of the low series. In some variations of this approach, the same point is an edge point of the segment, the midpoint between two edge points of the segment, or one of the points identifiable in both the high-linearity and low-linearity images.

[0055] In some cases, the first high observation axis and the first low observation axis overlap vertically in the same vertical plane.

[0056] In some cases, the first high observation axis and the first low observation axis are parallel to each other.

[0057] In some cases, the first high observation axis and the first low observation axis are perpendicular to the theoretical free fall path.

[0058] In some cases, the first high observation axis and the second high observation axis intercept the same high intercept point of the theoretical vertical free fall path, and / or the first low observation axis and the second low observation axis intercept the same low intercept point of the theoretical vertical free fall path.

[0059] The present invention also relates to a device for measuring the kinematic characteristics of a glass droplet in free fall in a glass molding apparatus, the type of which includes: a free fall region along the droplet loading path between the glass source and the forming cavity, wherein the droplet has a theoretical vertical free fall path in the free fall region, and the droplet has a starting end and a tail end, as well as a droplet length between the starting end and the tail end:

[0060] - At least four different linear cameras, each having a lens with an optical center and an optical axis defining the observed linear field for the camera under consideration, including a first high camera, a first low camera, a second high camera, and a second low camera, each camera having a first high optical axis, a first low optical axis, a second high optical axis, and a second low optical axis that are different from each other, each camera being capable of forming a digital image of the observed linear field (the first high linear field, the first low linear field, the second high linear field, and the second low linear field, respectively), wherein:

[0061] - The first high linear field and the first low linear field each extend along a corresponding plane defined by the corresponding optical axis and a horizontal extension direction perpendicular to the corresponding optical axis. The first high linear field and the first low linear field each intercept the theoretical free fall path at the first high intercept point and the first low intercept point, respectively. The first high intercept point and the first low intercept point are offset from each other according to the theoretical free fall path.

[0062] - The second high linear field and the second low linear field each extend along a corresponding plane defined by the corresponding optical axis and a horizontal extension direction perpendicular to the corresponding optical axis. The second high linear field and the second low linear field each intercept the theoretical free fall path at the second high intercept point and the second low intercept point, respectively. The second high intercept point and the second low intercept point are offset from each other along the direction of the theoretical free fall path.

[0063] - The high optical axes are different from each other in their vertical projections onto a plane perpendicular to the theoretical free-fall path direction, and the low optical axes are different from each other in their vertical projections onto a plane perpendicular to the theoretical free-fall path direction.

[0064] In some cases, the device includes an electronic computing unit programmed to implement a method having any of the above-described method features.

[0065] In some cases:

[0066] - The high-light axis, in its vertical projection onto a plane perpendicular to the theoretical free-fall path, forms a high-light axis deviation angle that is different from 0 degrees and different from 180 degrees, around an axis parallel to the theoretical free-fall path and passing through the convergence point of the projections of the two optical axes on the projection plane.

[0067] - The low optical axis, in its vertical projection onto a plane perpendicular to the theoretical free-fall path, forms a low optical axis acute angle deviation angle that is different from 0 degrees and different from 180 degrees, around an axis parallel to the theoretical free-fall path and passing through the convergence point of the projections of the two optical axes on the projection plane.

[0068] In some cases, the first high linear field and the first low linear field, as well as the second high linear field and the second low linear field, each intercept the theoretical free fall path of several glass droplets formed simultaneously from the same glass source.

[0069] The present invention also relates to a method for controlling a glass molding apparatus, the apparatus comprising a glass source, at least one shear, and a droplet loading path, the at least one shear being arranged at the outlet of the glass source and actuated at regular intervals to cut continuous droplets falling by gravity into a distributor, the distributor guiding the droplets toward the forming cavity of the apparatus along the at least one droplet loading path, and the droplet loading path including at least one droplet free fall region between the shear and the distributor, characterized in that the control method comprises: measuring the kinematic characteristics of the glass droplet free fall in the droplet free fall region between the shear and the distributor, said measurement comprising: determining, for a set of segments of a given droplet, the horizontal translation amount of each segment of the set, and characterized in that the control method comprises: adjusting at least one operating parameter of the shear as a function of the horizontal translation amount of each segment of the at least set.

[0070] The control method according to the invention may also include one or more of the following optional features, individually or in combination.

[0071] In some cases, adjustments may be made, including adjusting the position of the cutting point on the shearing machine.

[0072] In some cases, adjustments include at least adjusting the displacement speed of the shearing machine blades.

[0073] In some cases, adjustment involves adjusting the curve of the displacement speed of at least one blade of the shearing machine.

[0074] In some cases, adjustments may be made to the lubrication parameters of the shearing machine.

[0075] In some cases, the measurement includes determining at least two horizontal components of the horizontal translational velocity of at least one segment of the droplet between the high intercept point and the low intercept point.

[0076] In some cases, the measurement includes determining at least a first component of the rotation of the droplet about a first horizontal axis, and is characterized in that the adjustment includes: adjusting a component of the position of the cutting point of the shear, and / or adjusting a component of the displacement velocity of at least one blade of the shear, and / or adjusting a curve of the displacement velocity component of at least one blade of the shear.

[0077] In some cases, the measurement includes determining at least a second component of the rotation of the droplet about a second horizontal axis different from the first horizontal axis, and the control method includes adjusting at least one operating parameter of the shear as a function of a first component of the rotation of the droplet about the first horizontal axis and a second component of the rotation of the droplet about the second horizontal axis.

[0078] In some cases, the measurement of the kinematic characteristics of a glass droplet in free fall is implemented using a determination method that has any of the characteristics described above. Attached Figure Description

[0079] [ Figure 1 ] Figure 1 This is a schematic diagram of glass product molding equipment.

[0080] [ Figure 2 ] Figure 2 This is a schematic perspective view of the measuring device according to the present invention.

[0081] [ Figure 3 ] Figure 3 This is a schematic perspective view of a linear camera of the measuring device according to the present invention.

[0082] [ Figure 4-5A-5E Figures 4 and 5A to 5E are schematic diagrams showing different moments in the acquisition of high-series linear images and low-series linear images in the method according to the invention.

[0083] [ Figure 6 ] Figure 6 This is a schematic plan view showing the relative arrangement of two linear cameras in the same high or low group of the measuring device according to the present invention.

[0084] [ Figure 7 ] Figure 7 This is a diagram showing the vertical coordinates on a droplet associated with a series of linear images captured by a linear camera, within the framework of this invention.

[0085] [ Figure 8 ] Figure 8 It is a diagram showing an intermediate linear image of a segment of a droplet, excluding its two ends (start and end), and matching the image of the droplet contained in the intermediate linear image with the corresponding segment of the droplet.

[0086] [ Figure 9 ] Figure 9 This is a diagram illustrating the possibility of performing a determination using stereoscopic vision, based on the position in a plane determined by the optical axes of two cameras belonging to the same camera group. Detailed Implementation

[0087] Figure 1 An exemplary embodiment of a glass molding apparatus 10 is shown in part and schematically. Specifically, the apparatus 10 includes a forming machine 11, which includes forming sections 12 or several different forming sections 12, each forming section including at least one mold 14 having at least one cavity 16. The apparatus 10 may include a malleable glass source 18 (and thus a hot glass source 18) and a glass droplet distributor 20, which distributes the malleable glass droplets P into each cavity 16 of the forming machine 11 by gravity. In a known manner, the apparatus may include at least one shear 22, which is arranged at the outlet of the hot glass source 18 and actuated at regular intervals to cut extruded portions of the malleable glass from the source 18, or, for an apparatus where each section includes several cavities, several sections may operate simultaneously in parallel. The shear 22 generally includes two blades. The blades of the shear are actuated by relative movement of the cutting translation (referred to as parallel cutting) or the cutting rotation (referred to as scissor cutting) preferably utilizing two movable blades. In some devices, there may be several shearing machines 22, for example, when each part contains several blank molds and several droplets are loaded simultaneously in each cycle.

[0088] In this document, the extruded portion of malleable glass shredded by shear 22 is referred to as droplet P. This droplet is sometimes called a drip. At this stage of the forming method, droplet P is referred to as a droplet. Malleable glass generally has a temperature greater than 900°C at the cutting level of shear 22, for example, between 1,100°C and 1,300°C. The droplet is generally considered to be a solid cylinder of malleable glass of a certain length, and thus we can define a central axis corresponding to the length of the droplet. In this document, it should be understood that droplet P is not actually a perfect cylinder. Therefore, the outer surface of droplet P may be deformed relative to the cylinder. Similarly, the high end Pf and low end Pd of the droplet produced by the shearing operation of shear 22 generally do not have a planar cross section perpendicular to the central axis. However, it should be understood that the droplet generally remains as a substantially cylindrical object extending along the central axis. At least in theory, the central axis can be straight. In practice, this central axis is deformed, particularly in cases that cause the droplet to deform along its path in dispenser 20. Furthermore, in this paper, the curvature of the central axis will be referred to as the length of droplet P. The projection of this length onto the vertical axis Z, which is oriented relative to Earth's gravity, is referred to as the height of droplet P.

[0089] The time interval between two actuations of the shearing machine 22 determines the length, weight, and volume of the droplet, as the droplet cross-section and glass flow rate are determined, specifically as a function of the operating parameters of the source 18 and the shearing machine 22. The malleable glass source 18 is arranged above the forming machine 11, and therefore above the die 14 of the forming section 12, to allow for gravity-based droplet distribution. The distributor 20 generally extends along several branches between the hot glass source 18 and the die cavity of each forming section. Figure 1 This only represents one of these branches. Source 18, Distributor 20, and Shearer 22 can be of any known type, so they will not be described in more detail.

[0090] Different methods are used in machines for forming insulated glass articles to perform steps of combining filling molds, followed by continuous pressing and / or blowing. For clarity of description, examples are taken from forming bottles according to a known method known as press-blown or blow-blown.

[0091] In a bottle forming machine, each forming section 12 may include several molds, such as two molds, one of which is a preform mold and the other is a blow molding mold. Each forming section may include a preform mold group and a finishing mold group, the mold group consisting of several molds of the same forming section, involving the same forming step, and generally opening and closing simultaneously. It should be understood that in this case, a given droplet is guided by a distributor 20 toward the preform mold (e.g., the preform mold of the forming section), in which the droplet undergoes a first forming operation (referred to as a drilling operation), which is carried out by compressed air blowing or punch penetration. Then, a conveying system (not shown) is able to remove the droplet (i.e., the preform) that has undergone the first forming operation from the preform mold to carry it to a secondary mold (generally referred to as a blow molding mold), in which the preform may undergo at least a second forming operation, i.e., the final operation (referred to as a finishing operation). Generally, each mold of the forming section includes two half-molds, which are movable relative to each other in a direction perpendicular to the dividing plane, and the two half-molds contact each other in a closed position through the dividing plane. Figure 1 Only one mold 14 for each forming part 12 is shown, but the invention can certainly be applied to devices (called multi-drop devices) in which each forming part includes several molds.

[0092] The forming section 12 may include a single mold cavity 16. However, as mentioned above, especially in multi-drop devices, each different forming section 12 may include at least two different forming cavities 16, generally because they include several molds, and more rarely because they include molds with several cavities.

[0093] Generally, the hot glass source 18 simultaneously delivers the same number of droplets as the forming cavity in the forming section via the shear 22. Therefore, it can be understood that the forming section is continuously supplied with droplets one after another. Thus, we can have two forming sections, which are successively supplied through the same outlet of the hot glass source 18. Therefore, the two forming sections are not supplied simultaneously.

[0094] Therefore, the distributor 20 collects the droplets cut by the shearer 22 and guides them toward the forming cavity of the forming section along loading paths 28 corresponding to the forming cavity 16. The loading paths for different forming cavities include common portions and specific portions. Specific portions are the parts of the loading paths corresponding to the forming cavities, which are followed only by droplets guided toward that forming cavity by the distributor. Therefore, the distributor 20 includes means for guiding and deflecting the droplets. These guiding and deflecting means may include buckets, channels, deflectors, etc., some of which may be at least movable to form deflectors. In the distributor, channels, deflectors, and deflectors define the droplet loading path.

[0095] exist Figure 1 In the example, the device for guiding and directing the distributor 20 actually includes, in sequence, a bucket 20a, a channel 20b, and a deflector 20c along the loading path between the source 18 and the mold 14. Generally, the position and orientation of the bucket 20a are adjustable. Specifically, in the case where the machine includes several forming sections, the bucket 20a can be rotatable about a vertical axis to collect droplets from the source and guide them toward one of several channels corresponding to the several forming sections. Following the bucket 20a, the droplets P are supported by the channel 20b, and then the deflector 20c deflects the droplets to orient them in the direction of the cavity 16 corresponding to the deflector 20c. In some devices, a funnel and / or droplet accelerator can be provided at the inlet of the bucket (and thus at the end of free fall). For example, the droplet accelerator uses pneumatic action to bring the droplets in at an increased velocity. This action may vary depending on the forming section and / or cavity that each droplet is directed to.

[0096] In the example shown, the initial part of the droplet loading path (which is a specific part of the loading path) is a high free fall zone 24 along which the droplet is not guided and falls under the influence of gravity without vertical guidance. The high free fall zone 24 is between the shearer 22 and the bucket 20a of the distributor 20.

[0097] In the example shown, the terminal portion of the droplet loading path (i.e., a specific part of the loading path) is a low free fall region 26 along which the droplet is not guided and falls into the mold cavity 16 under the influence of gravity at the outlet of the deflector 20c of the distributor 20 without vertical guidance.

[0098] As with bucket 20a, the position and orientation of the various elements of the distributor (e.g., one or more channels 20b and one or more deflectors 20c) are adjustable so that the droplets are accurately guided toward the mold cavity 16. It is important to note that the arrangement of the deflectors 20c helps to determine the ends of the loading path, particularly the position of the path relative to the opening axis of the cavity.

[0099] In the bottle forming machine, the control and synchronization of droplet formation, shearing operations of the shearer 22, mold movement, movement of one or more punches of the hot glass source 18, blow molding, transfer, etc., were previously mechanically performed by means of grooved rollers with studs, each stud being adjustable and having the function of mechanically controlling the valves of the control cylinders. The new machine is now equipped with an electronic controller, that is, by means of a programmable logic controller, which allows movement to be driven using any type of actuator (e.g., actuators operated using pneumatic or electrical energy). As will be described in detail later, the electronic controller can exchange synchronization signals, control signals, or information about the equipment status from sensors with any internal or external components (including sensors or actuators), for example, through wired, wireless, or hybrid communication networks. However, the programmable logic controller is still a system with very specific programming and limited to cyclic sequential programs. For example, the electronic controller can implement cyclic sequential programs and can also integrate internal model control, multivariate regulation algorithms, simulation, and optimization.

[0100] According to one aspect of the invention, the apparatus 10 includes at least one device 30 for measuring the kinematic characteristics of a glass droplet P falling freely in a glass molding apparatus. The measuring device 30 is an optical device based on a linear camera, and is therefore arranged in the apparatus to observe the droplet P in the region of free fall along the droplet loading path between the glass source 18 and the forming cavity 16. As can be seen above, in this example, the apparatus has a high free fall region 24 and a low free fall region 26 along the path of the droplet P. The apparatus may be equipped with a first measuring device for the high free fall region 24 and / or a second measuring device for the low free fall region 26. The following will describe in detail the apparatus including such a device 30 for measuring kinematic characteristics, arranged in the apparatus to observe the droplet P in the high free fall region 24.

[0101] In the free-fall region, particularly in the high free-fall region 24, droplet P is only affected by Earth's gravity, and at least in the high free-fall region 24, droplet P is considered to have a theoretical free-fall path 28.24, which is vertical in the free-fall region. Furthermore, in this high free-fall region, droplet P is considered to have a theoretical vertical orientation during this vertical descent, with its central axis assumed to be vertical and coinciding with the vertical theoretical free-fall path 28.24. Therefore, in this theoretical vertical orientation during free fall, droplet P is considered to have a starting end Pd (or lower end or lower back) and a tail end Pf (or upper end or higher back) along the vertical direction, as well as a droplet length between the starting end Pd and the tail end Pf. It can be seen that this theoretical orientation is theoretical, therefore the measuring device 30 will allow the determination of the deviation between the actual orientation of the droplet and this theoretical orientation.

[0102] As in Figure 2 and Figure 3 More specifically, the measuring device 30 includes at least four different linear cameras 32.1h, 32.1b, 32.2h, 32.2b (which may be commonly and / or vaguely designated by a common reference numeral 32), each camera having a lens 33.1h, 33.1b, 33.2h, 33.2b (which may be commonly and / or vaguely designated by a common reference numeral 33), the lens having an optical center (see [reference]). Figure 6 The four linear cameras are defined by optical axes 36.1h, 36.1b, 36.2h, and 36.2b (which may be commonly and / or vaguely specified by common reference numeral 36), which define the observed linear fields 34.1h, 34.1b, 34.2h, and 34.2b for the cameras under consideration (which may be commonly and / or vaguely specified by common reference numeral 34). Therefore, the four linear cameras include a first high camera 32.1h, a first low camera 32.1b, a second high camera 32.2h, and a second low camera 32.2, each with a distinct first high optical axis 36.1h, a first low optical axis 36.1b, a second high optical axis 36.2h, and a second low optical axis 36.2b. The concepts of "high" and "low" refer to orientation perpendicular to the direction of Earth's gravity.

[0103] Each linear camera 32 includes a linear photoelectric sensor, which is thus formed by rows of photoelectric elements. The lens 33 of the linear camera 32 forms a linear field 34 on the photoelectric elements, resulting in a photographic image seen by the sensor through the lens 33. The sensor transmits image signals, continuously acquiring these signals to provide temporally continuous photographic, digital, and linear images. The sensor can be a sensitive sensor in the visible light range, allowing the use of commercially available sensors and lenses to provide sufficient performance at a reasonable cost. However, this does not preclude the use of sensors that are equally sensitive in the infrared domain, or even only sensitive in the infrared domain. For each linear camera 32, the lens 33 of the camera 32 defines the optical axis 36 of the camera. The lens 33 of the camera 32 can be a telecentric lens or a conventional lens.

[0104] Therefore, each of the linear cameras 32 can form a digital image of the linear field 34 corresponding to that camera, such that the four cameras define a total of four linear fields 34, namely a first high linear field 34.1h, a first low linear field 34.1b, a second high linear field 34.2h, and a second low linear field 34.2b. The linear field 34 corresponding to the camera extends along the direction of the plane defined by the optical axis 36 of the linear camera 32 and along the extension direction, which is perpendicular to the optical axis and corresponds to the row of photoelectric elements forming the linear photoelectric sensor.

[0105] Therefore, the first high linear field 34.1h and the first low linear field 34.1b each extend along the corresponding plane defined by the corresponding optical axes 36.1h and 36.1b and in a horizontally extending direction, which is perpendicular to the corresponding optical axes 36.1h and 36.1h of the linear camera 32 associated with the linear field 34.

[0106] The first high linear field 34.1h and the first low linear field 34.1b each intercept the theoretical free-fall path 28.24 at the first high intercept point 40.1h and the first low intercept point 40.1b, respectively. The first high intercept point and the first low intercept point are offset perpendicularly from each other according to the theoretical free-fall path 28.24. This perpendicular offset between the first high intercept point 40.1h and the first low intercept point 40.1b can advantageously be strictly smaller than the expected nominal length of the droplet, so as to allow the method described later to be implemented even when the environment does not allow the droplet to be easily seen.

[0107] The second high linear field 34.2h and the second low linear field 34.2b each extend along a corresponding plane defined by corresponding optical axes 36.2h and 36.2b in a horizontally extending direction perpendicular to the corresponding optical axis of the linear camera 32 associated with the linear field 34. The second high linear field 34.2h and the second low linear field 34.2b each intercept the theoretical free-fall path 28.24 at a second high intercept point 40.2h and a second low intercept point 40.2b, respectively. The second high intercept point and the second low intercept point are offset perpendicularly from each other along the direction of the theoretical free-fall path. This vertical offset between the second high intercept point and the second low intercept point can advantageously be strictly less than the expected nominal length of the droplet, possessing the same advantages as described above.

[0108] As can be seen from the figure, the specular axes (i.e., the first specular axis 36.1h and the second specular axis 36.2h) are different from each other in their vertical projections onto a plane perpendicular to the theoretical free-fall path, in order to allow for stereoscopic vision of the free-fall region. In other words, the first specular axis 36.1h and the second specular axis 36.2h do not overlap perpendicularly, and therefore do not coincide. Therefore, the two high-angle cameras 32.1h and 32.2h neither coincide, nor are they exactly 180° opposite each other, nor do they overlap perpendicularly, which is to obtain stereoscopic vision of the free-fall region.

[0109] As can be seen from the figure, the low optical axes (i.e., the first low optical axis 36.1b and the second low optical axis 36.2b) are different from each other in their vertical projections onto a plane perpendicular to the theoretical free-fall path, in order to allow for stereoscopic vision of the free-fall region. In other words, the first low optical axis 36.1b and the second low optical axis 36.2b do not overlap perpendicularly with each other, and therefore do not coincide. Therefore, the two low cameras 32.1b and 32.1b neither coincide, nor are they exactly 180° opposite each other, nor do they overlap perpendicularly, which is to obtain stereoscopic vision of the free-fall region.

[0110] from Figure 6 As can be seen, we preferably ensure that the high optical axes (i.e., the first high optical axis 36.1h and the second high optical axis 36.2h) together form a high optical axis deviation angle Aopth around the theoretical free fall path 28.24 (or in any case around the axis parallel to the theoretical free fall path 28.24 and passing through the convergence point of the projections of the two optical axes in the projection plane) in the vertical projection on a plane perpendicular to the direction of the theoretical free fall path 28.24. The high optical axis deviation angle Aopth is different from 0 degrees and different from 180 degrees. This is to enable a large area of ​​intersection with a linear field, in which the same droplet can be obtained from two different viewpoints. This allows the geometric and positional characteristics of the droplet along two horizontal directions to be determined by stereoscopic vision for any point in the intersection area.

[0111] Figure 6 It is a vertical projection. In Figure 2 , Figure 3 and Figure 6 In the example shown, the low-linearity cameras 32.1b and 32.2b are identical to and configured in the same way as the high-linearity cameras 32.1h and 32.2h, so that in Figure 6 In this context, all their elements are represented as overlapping the corresponding elements of the high-linearity cameras 32.1h and 32.2h. The low optical axes (i.e., the first low optical axis 36.1b and the second low optical axis 36.2b) together form a low optical axis deviation angle Aoptb around the theoretical free fall path (or in any case around the axis parallel to the theoretical free fall path and passing through the convergence point of the projections of the two optical axes in the projection plane) in a vertical projection on a plane perpendicular to the theoretical free fall path 28.24. The low optical axis deviation angle Aoptb is different from 0 degrees and different from 180 degrees, in order to be able to determine the geometric and positional characteristics of the droplet along both directions through stereoscopic vision.

[0112] Preferably, as shown in the figure, the first high optical axis 36.1h and the second high optical axis 36.2h, or at least their vertical projections on a plane perpendicular to the theoretical free-fall path 28.24, converge at a convergence point located in front of the high linearity cameras 32.1h and 32.2h (that is, on the opposite side of the camera sensors relative to their lenses). Similarly, preferably, the first low optical axis 36.1b and the second low optical axis 36.2b, or at least their vertical projections on a plane perpendicular to the theoretical free-fall path 28.24, converge in front of the low linearity cameras 32.1b and 32.2b, as shown in the figure. Parallel or divergent high optical axes and / or parallel or divergent low optical axes are possible, but not optimal.

[0113] As shown in Figure 4, during free fall, the given droplet P first passes through the first high linear field 34.1h of the first high camera 32.1h, and then through the first low linear field 34.1b of the first low camera 32.1b. Simultaneously, the same droplet first passes through the second high linear field 34.2h of the second high camera 32.2h, and then through the second low linear field 32.2b of the second low camera 32.2b. In front of the given linear camera 32, the starting end Pd of the given droplet P passes in front of the given camera, and then the tail end Pf of the same droplet P passes in front of the given camera.

[0114] In the example shown, the first high optical axis 36.1h of the first high camera 32.1h and the first low optical axis 36.1b of the first low camera 32.1b overlap vertically in the same vertical plane. In the example shown, the second high optical axis 36.2h of the second high camera 32.2h and the second low optical axis 36.2b of the second low camera 32.2b overlap vertically in the same vertical plane.

[0115] In the example shown, the low-axis deviation angle Aoptb between the low optical axes 36.1b and 36.2b is the same as the high-axis deviation angle Aopth between the high optical axes 36.1h and 36.2h. This layout is by no means mandatory, but is beneficial for the calculations described below.

[0116] In some embodiments, the plane defined by the first high optical axis 36.1h and the first low optical axis 36.1b is also a plane that includes the theoretical vertical path 28.24 of the droplet in the considered free fall region. This layout can be used particularly when the measuring device 30 is implemented for a single, given theoretical free fall path of the droplet.

[0117] also, Figure 6 The following configuration is illustrated schematically: including four linear cameras 32 as described above. Figure 6 Only the first high camera 32.1h and the second high camera 32.2h are visible in the image (but the same image can be reproduced to represent the case of the first low camera 32.1b and the second high camera 32.2b). The same measuring device 30 allows at least two theoretical vertical free fall paths 28.24 and 28'.24 of at least two droplets from the same glass source 18 to be observed in the same free fall region 24 in the directions of at least two different forming cavities. At least in the considered free fall region 24, the two theoretical vertical free fall paths 28.24 and 28'.24 are parallel and offset from each other in a horizontal direction perpendicular to their common vertical direction. Figure 6This means that the optical axes 36 of the four linear cameras 32 are positioned to intercept the first theoretical free-fall path 28.24 in the sense that the optical axes 36.1h, 36.2h, ... intersect the first theoretical free-fall path 28.2 at their respective high and low intercept points. On the other hand, since the two theoretical vertical free-fall paths 28.24 and 28'.24 are horizontally offset relative to each other, it can be seen that the optical axes 36.1h, 36.2h, ... of the four cameras 32.1h cannot intersect the second theoretical vertical free-fall path 28'.24. However, the four linear fields 34.1h, 34.2h, ... corresponding to the four linear cameras 32.1h, 32.2h, ... of the same measuring device 30 also capture the second theoretical vertical free fall path 28'.24 at the high intercept point and low intercept points 40'.1h, 40'.1b, 40'.2h, 40'.2b of each of the four linear fields 34.1h, 34.2h, ... without modifying the orientation of the four linear cameras 32.1h, 32.2h, ... relative to the device. Therefore, in general, the measuring device 30 can be arranged such that the first high linear field 34.1h and the first low linear field 34.1b, as well as the second high linear field 34.2h and the second low linear field 34.2b, each capture the theoretical free fall path of several glass droplets formed simultaneously from the same glass source. In fact, "multi-drop" dispensers are quite common. They generally use two, three, or four independent plungers, sometimes with two, three, or four can holes, and two, three, or four pairs of synchronous or asynchronous shears to deliver two, three, or four drops in parallel.

[0118] Therefore, in this case, for each given theoretical free-fall path and for each of the four linear cameras 32.1h, 32.2h, ... of the measuring device 30, we can define the observation axis 37.1h, 37.2h, ... of the given theoretical free-fall path by means of the camera under consideration. In the sense that the observation axis 37.1h, 37.2h, ... of the given theoretical free-fall path given by the linear camera under consideration intersects the given theoretical free-fall path (e.g., the second theoretical free-fall path 28'.24) at the corresponding intercept point, the observation axis is the axis contained in the linear field 34.1h, 34.2h, ... of the linear cameras under consideration 32.1h, 32.2h, ..., which passes through the optical center of the lens 33 of the camera under consideration and intercepts the given theoretical free-fall path 28'.24. Under these conditions, the measuring device 30 is capable of measuring the kinematic characteristics of a glass droplet falling along a first theoretical free-fall path 28.24, and also of measuring the kinematic characteristics of a glass droplet falling along a second theoretical vertical free-fall path 28'.24, which is also captured by the linear fields 34.1h, 34.2h, ... of the four linear cameras 32.1h, 32.2h, ... of the measuring device 30. Note that when the optical axes 36.1h, 36.2h, ... of the linear cameras 32 capture a given theoretical free-fall path (which is specifically in... Figure 6 When considering the first theoretical free-fall path 28.24 (as shown in the figure), the optical axes 36.1h, 36.2h, ... of the linear camera 32 under consideration coincide with the observation axes 37.1h, 37.2h, ... of the lens that connects the given theoretical free-fall path 28.24 to the optical center of the linear camera 32.1h, 32.2h, ... under consideration.

[0119] exist Figure 6In the example shown, it can be understood that the optical axes 36.1h, 36.2h, ... of each of the linear cameras 32.1h, 32.2h capture the same theoretical free fall path among multiple theoretical free fall paths, which are captured by the linear fields 34.1h, 34.2h, ... of the four linear cameras 32.1h, 32.2h, ... of the measuring device 30. However, as a variation, we can arrange the measuring device 30 such that, although its four linear cameras are configured such that the linear fields 34.1h, 34.2h, ... of the four linear cameras 32.1h, 32.2h, ... capture multiple theoretical free-fall paths, the optical axes 36.1h, 36.2h, ... of the linear cameras 32.1h, 32.2h do not capture any of the multiple theoretical free-fall paths, or the optical axes 36.1h, 36.2h, ... of the linear cameras 32.1h, 32.2h do not all capture the same theoretical free-fall path among the multiple theoretical free-fall paths. Therefore, we can have the following configuration: the optical axes of the first high camera and the first low camera capture the first theoretical free-fall path (without capturing the second theoretical free-fall path), while the optical axes of the second high camera and the second low camera capture the second theoretical free-fall path (without capturing the first theoretical free-fall path).

[0120] In the example shown in the figure, the first high optical axis 36.1h and the first low optical axis 36.1b of the first high camera 32.1h overlap in the same vertical plane and are also parallel to each other. This parallelism will improve the accuracy of the measurement and will help determine the kinematic equations involved. However, this parallelism is not mandatory. We will also advantageously stipulate that, as in the example shown, the second high optical axis 36.2h and the second low optical axis 36.2b of the second high camera 32.2h overlap in the same vertical plane, and even more advantageously, they are also parallel to each other.

[0121] Preferably, the optical axes 36 of the four linear cameras 32 of the measuring device 30 are all horizontal, and therefore orthogonal to the direction of one or more theoretical vertical free-fall paths. This also applies to cases where one of these optical axes intersects with one of these theoretical vertical paths, and cases where one of these optical axes does not intersect with the theoretical vertical paths. In this case, the observation axes 37.1h, 37.2h... connecting the intercept point of the given theoretical free-fall path 28.24 to the optical center of the lens of the corresponding linear cameras 32.1h, 32.2h, ... are also horizontal, regardless of the theoretical free-fall path.

[0122] Preferably, for a given theoretical free-fall path 28.24, 28'.24, the first high intercept points 40.1h, 40'.1h and the second high intercept points 40.2h, 40'.2h will coincide. In other words, in this case, the first high camera 32.1h and the second high camera 32.2h are arranged such that they see the same point on the theoretical vertical free-fall path of the droplet through their lenses. Preferably, in the same manner, for a given theoretical free-fall path 28.24, 28'.24, the first low intercept points 40.1b, 40'.1b and the second low intercept points 40.2b, 40'.2b will coincide.

[0123] As can be understood from the foregoing, the measuring device 30 therefore includes four linear cameras 32, which together can be considered as forming two sets of cameras, each consisting of two cameras: a high camera set including a first high camera 32.1h and a second high camera 32.2h, and a low camera set including a first low camera 32.1b and a second low camera 32.2b.

[0124] Preferably, the two cameras in the high camera group acquire linear images of the same high intercept point 40.1h containing the theoretical vertical free-fall path 28.24, and thus acquire linear images of the same segment of the droplet as it passes this point. Since the two cameras in the high camera group are positioned at different points in space relative to the high intercept point 40.1h, we can obtain linear images of the same segment of the droplet from two different viewpoints, which allows us to determine the geometric and positional features of the segment along two different horizontal directions using stereo vision. The same applies to the two cameras in the low camera group. In this example, the two cameras in the high camera group are configured such that their respective optical axes together form a non-zero angle Aopth different from 180 degrees around the theoretical vertical free-fall path 28.24. Typically, the two cameras in the high camera group are configured such that their respective optical axes together form an angle Aopth around the theoretical vertical free-fall path 28.24, which includes an angle ranging from 5 to 90 degrees.

[0125] Furthermore, the four linear cameras 32 of the measuring device can be considered together as two sets of cameras, each consisting of a high camera and a low camera. Using these sets of cameras, including both high and low cameras, we can determine some kinematic characteristics of the free fall of the glass droplet. In the case of two sets of cameras, each including a high camera and a low camera, but where the optical axes of the cameras belonging to two different sets together form a non-zero angle different from 180 degrees, we can determine the kinematic characteristics of the free fall of the glass droplet in three dimensions.

[0126] In practice, the measuring device 30 includes or is connected to an electronic computing unit 42, which is programmed to implement a method for measuring the kinematic characteristics of the free fall of the glass droplet when the free fall path of the glass droplet intercepts four linear fields of the four linear cameras 32 of the measuring device 30. The electronic computing unit 42 may be made, for example, in the form of a computer and / or acquisition card within the device 11, and it collects a series of consecutive linear digital images acquired by each of the four cameras 32 of the measuring device 30. The electronic processing unit 42 may be dedicated solely to the measuring device 30 or may be shared with other devices within the device 10.

[0127] Figure 1 The example shows that device 10 includes a control and monitoring computer system 44, with an electronic computing unit 42 that may form part of it or communicate with it. In fact, in such a device, the operation of all components must be coordinated, and the operation of all components follows a cyclical pattern. For example, the distributor 20 is monitored by the device's control and monitoring central unit according to the cycle of the forming machine 11. The forming machine 11 generally also includes one or more control units that monitor, for example, the sequential opening and closing of different parts of the mold 14, control one or more forming operations, possibly control components for transferring blanks to finishing molds or devices for transferring containers formed by blow molding to container conveyors, control the feeding trough for conveying molten glass at a selected temperature to the glass source 18, and also control the operation of one or more plungers 17 and / or cylinders 19 of the glass source 18, control one or more shears 22, etc. The electronic control units of the machine and / or device may be disjoint or overlapping, but are part of the device's control and monitoring computer system 44, and therefore necessarily have information related to the operating cycles of the machine and its various forming parts. Computer system 44 may include at least one standard computer, and therefore includes at least one microprocessor, one or more electronic memory units, and one or more displays (screen, projector, holographic display, etc.), input devices (keyboard, mouse, touchpad, touch screen, etc.) and / or communication (USB, etc.). (etc.) interface. The computer system may include, for example, an interface via the Internet or... The protocol shares data with one or more other computers on the network or with other computer networks. In addition to its apparent connection to the linear camera 32, the computer system 44 may also connect to sensors that provide information about the status of the device and / or actuators (conveyors, ejectors, etc.) connected to the device. The computer system implements one or more software stored locally or remotely and / or executed, including one or more software on one or more remote computer servers. This or such software preferably includes one or more software programmed to implement a method for measuring the kinematic characteristics of a droplet in free fall, as described below. Specifically, the electronic computing unit 42 may be programmed to implement this method.

[0128] First, the method may include, for example, using four different linear cameras 32 of the measuring device 30 as described above to acquire at least four series of continuous linear digital images, each of the given series of linear images being a linear image of the observation field of the free fall region acquired by one of the four linear cameras 32.

[0129] Therefore, the four series of linear images include a first high series acquired by the first high camera 32.1h, a first low series acquired by the first low camera 32.1b, a second high series acquired by the second high camera 32.2h, and a second low series acquired by the second low camera 32.2b. These four series of linear images correspond to the linear images of the first high linear field 34.1h, the first low linear field 34.1b, the second high linear field 34.2h, and the second low linear field 34.2b in the free-fall region (e.g., the high free-fall region 24), respectively.

[0130] For a given theoretical free-fall path, the images in the four series of linear images each include images of corresponding intercept points 40.1h, 40.1b, 40.2h, and 40.2b, and are therefore acquired along the first high observation axis 37.1h, the first low observation axis 37.1b, the second high observation axis 37.2h, and the second low observation axis 37.2b, respectively. These observation axes are different from each other and are determined by the corresponding linear cameras in the manner described above. The observation axes pass through the intercept points and through the optical centers of the cameras corresponding to the series under consideration.

[0131] For a given theoretical free fall path (e.g., Figure 6The two theoretical free-fall paths (either of 28.24 and 28'.24) together form high observation deviation angles Aobsh and Aobsh' around the theoretical free-fall path in their vertical projections onto a plane perpendicular to the theoretical free-fall path. These angles differ from 0° and 180°. Therefore, acquiring two linear images of the same segment of the droplet from two different viewpoints and in two misaligned observation directions allows for the determination of the droplet's geometric and positional features along two horizontal directions based on stereoscopic vision principles. Preferably, the high observation deviation angles Aobsh and Aobsh' are within an angular range of 10° to 170° around the theoretical free-fall path. With a minimum angle of 10 degrees between the directions of the two observation axes, we ensure sufficient difference between viewpoints to achieve adequate accuracy in determining the droplet's geometric and positional features along two mutually perpendicular horizontal directions. Note that the observation axes do not need to be perpendicular to each other to obtain the droplet's geometric and positional features along two mutually perpendicular horizontal directions. Note that high observation deviation angles Aobsh and Aobsh' can advantageously be included in the range of 10° to 45°, or even in the range of 10° to 25°, to improve the compactness of the device and facilitate its integration into equipment. This allows the high optical axis deviation angle Aopth to be included in the range of 10° to 45°, or even in the range of 10° to 25°.

[0132] Similarly, for a given theoretical free fall path (e.g., Figure 6 The first low observation axis and the second low observation axes 37'.1b, 37'.1h preferably form low observation deviation angles Aobsb, Aobsb' around the theoretical free fall path, which are different from 0° and different from 180°, in their vertical projections onto a plane perpendicular to the direction of the theoretical free fall path, having the same advantages. Similarly, the low observation deviation angles are preferably included in the angular range of 10° to 170° around the theoretical free fall path to obtain the geometric and positional characteristics of the droplet in two horizontal directions perpendicular to each other with sufficient accuracy. Similarly, the low observation deviation angles Aobsb, Aobsb' can advantageously be included in the range of 10° to 45°, or in the range of 10° to 25°, to improve the compactness of the device and facilitate its integration into equipment. This allows the low optical axis deviation angle Aoptb to be included in the range of 10° to 45°, or even in the range of 10° to 25°.

[0133] The time deviation between the acquisition of any two images from the same series (and therefore acquired by the same linear camera 32), and the time deviation between the acquisition of any two images from two different series (and therefore acquired by two different linear cameras of the measuring device 30), is determinable. For example, the time deviation between the acquisition of two consecutive images from the same series is predetermined, preferably constant, and at least has a known variation or is based on a known function, such that knowing the time of acquisition of images in the series is sufficient to determine the time deviation between the acquisition of any two consecutive or non-consecutive images from that series. Similarly, for example, knowing the time deviation between the acquisition of a reference image from a first series (e.g., an image containing an identifiable point of a droplet) and the acquisition of a reference image from another series (e.g., an image containing the same identifiable point of a droplet) is sufficient to determine the time deviation between the acquisition of any two images from two different series.

[0134] Figure 4-5A-5E The diagram shows the different consecutive positions of the droplet on its theoretical free-fall path 28.24 as it passes in front of the measuring device 30.

[0135] Figure 4 shows droplet P at time td1h, where its starting end Pd is located at the high intercept point 40.1h of the set of linear cameras 32 (here, for example, the two linear cameras 32.1h and 32.1b of the first set). Figure 5A shows the same droplet P at a later time, where its starting end Pd has passed the high intercept point 40.1h of the same set of linear cameras 32, but has not yet reached the low intercept point 40.1b. Figure 5B shows the same droplet P at a later time td1b, where its starting end Pd has reached the low intercept point 40.1b. Note that the tail end Pf of droplet P has not yet reached the high intercept point 40.1h, reflecting the fact that the vertical offset between the high and low intercept points of the set of linear cameras (here, the first set of linear cameras) is strictly less than the length of the droplet. Although not shown, there are subsequent times when the two linear cameras in the set of linear cameras observe two different segments of droplet P, which are different from the end segment of the droplet. Figure 5C shows the same droplet P at a later time tf1h, where its tail end Pf is located at the high intercept point 40.1h of the linear camera 32. Figure 5D shows the same droplet P at a later time, where its tail end Pf has passed the high intercept point 40.1h of the same linear camera 32, but has not yet reached the low intercept point 40.1b. At this time, the low camera (here, the first low camera 32.1b) still observes the section of droplet P between its starting end Pd and its tail end Pf. Figure 5E shows the same droplet P at a later time tf1b, where its tail end Pf reaches the low intercept point 40.1b. To recap, the device 30 enables the measurement of droplet characteristics in devices that could not possibly have an observation field covering the entire droplet. Note that the vertical offset between the high and low intercept points of the linear camera is selected, for example, within a range from 10 mm to 100 mm. Therefore, in most applications, this vertical offset is less than the nominal length of the droplet, which can be, for example, in the range of 100 to 250 mm, such as 150 mm. The nominal length of the droplet is adjustable. Clearly, the described apparatus 30 can operate even if the actual droplet length is less than the vertical offset between the high and low intercept points of the camera assembly. For example, we can actually adjust the device to produce droplets shorter than the nominal length (e.g., 80 mm droplets), while the vertical offset between the high and low intercept points of the linear camera assembly can be greater than the droplet length, for example, equal to 100 mm, if the environment allows. The described measurement method and apparatus still operate in different sequences at the times when the ends coincide with the high and low intercept points, in the following order: td1h, tf1h, td1b, tf1b.

[0136] In a given series, the time deviation between the acquisition of two consecutive images by the linear camera 32 corresponding to that series (i.e., the acquisition period of the linear camera 32) is sufficiently reduced so that the linear camera 32 acquires several consecutive images of the droplet in a time space that separates the passage of the starting end Pd of the droplet in the linear field 34 of the linear camera 32 relative to the passage of the tail end Pf of the same droplet P. Preferably, in the time interval that separates the passage of the starting end Pd of the droplet P in the linear field 34 of the linear camera 32 relative to the passage of the tail end Pf of the same droplet P, the linear camera 32 acquires, for example, 1 to 100 linear images of a 1 mm droplet, for example, 10 images of a 1 mm droplet. In other words, for a 18 cm high droplet, the linear camera 32 acquires approximately 1,800 linear images of the droplet. Therefore, we will preferably select a linear camera with an acquisition frequency of at least 5 kHz, preferably at least 10 kHz, for example, 40 kHz.

[0137] Typically, we can use a camera whose linear photoelectric sensor consists of at least 512 photoelectric elements in a row (preferably at least 1,024 or 2,048 or more), provided that for good accuracy of vertical measurements, the acquisition frequency is kept greater than 5 kHz, preferably greater than 10 kHz.

[0138] Note that if fields 34.1h and 34.2h (correspondingly 34.1b and 34.1h) are parallel to each other, thus defining the same viewing angle relative to the horizontal, then for a given segment captured by two linear cameras in the same set, the linear image of that segment captured by the two linear cameras in the same set is an image of that same segment, particularly for the starting or ending segment of the droplet. Similarly, if fields 34.1h and 34.1b (correspondingly 34.2h and 34.2b) are parallel to each other, defining the same viewing angle relative to the horizontal, then for a given segment captured by two linear cameras in the same set, the linear image of that segment captured by the two linear cameras in the same set is an image of that same segment, based on the same viewing angle relative to the horizontal. In both cases, this allows the timing of the passing of such a starting or ending point of the droplet to be detected independently of the shape of the end. This is currently desirable because the shape of this end depends on the cutting operation performed by the shearing machine, and depending on the viewing angle of that end, and therefore, the angle relative to the horizontal, and thus on the linear camera group or set whose image is considered. More advantageously, the four fields are parallel to each other. The device is designed to provide the best possible parallelism for fields 34.1h, 34.2h, 341b, and 34.1h, because the spatial positioning accuracy of the sensors and lenses is limited by the accuracy of the mechatronic components.

[0139] Figure 8 The diagram shows a set of linear images acquired by a given linear camera during the time a droplet P traverses within the linear field of the camera. This set of images is part of a larger series of images (i.e., a series of linear images acquired by the given linear camera during the camera's operating time). This set of images is determined by computationally identifying the linear images of the series (referred to as the droplet start image I1 (the image where the starting end Pd of the droplet appears) and the droplet linear tail image IL (the image where the tail end of the same droplet appears)). Therefore, the set of images consists of L consecutive linear images acquired by the linear camera 32: I1, I2, I3, ..., I(i-1), Ii, ..., I(L-2), I(L-1), IL. Within each image, the presence of a segment of the image containing the droplet can be determined computationally, for example, by detecting the contrast between the droplet (recalling its brightness in the visible light range) and the background.

[0140] It should be understood that for the same droplet P, we thus obtain four image sets, each corresponding to one of the linear cameras 32. For each image set, we can easily identify, in each corresponding series, the first image of that image set (i.e., the droplet's initial image I1), in which the image of the droplet's initial end Pd appears, for example as the first image including the droplet, followed by several images in the series that do not include the droplet. We can also easily identify, in each series, the linear tail image IL of that image set, in which the image of the droplet's tail end Pf appears, for example as the last image in the series including the droplet, followed by several consecutive images in the series that do not include the droplet. On the other hand, for the intermediate images between the droplet's initial image I1 and the droplet's end image IL, we cannot accurately determine which droplet segment the image it contains through simple analysis of the image. In fact, throughout the entire descent, we assume that the droplet is only affected by Earth's gravity, which means that it has a constant acceleration equal to Earth's gravity, and therefore its velocity changes constantly. Furthermore, the method then provides an intermediate linear image of the image of the droplet, excluding its two ends, the beginning and the end, and matches the image of the droplet contained in the intermediate linear image with the corresponding segment of the droplet.

[0141] From two series of linear images corresponding to two cameras in the same set of linear cameras 32 (i.e., corresponding to the high camera and the low camera), we can determine the average vertical velocity (Vz_moy_d and / or Vz_moy_f) of one end of the droplet (and therefore the end segment) between two intercept points corresponding to the high and low of that set of linear cameras. This involves the computer identification of the high linear image and the low linear image, each comprising images of the droplet at the same given end in the beginning and end of the droplet, and involves deriving the time deviation between the acquisition of the high linear image and the acquisition of the low linear image.

[0142] Therefore, we can determine the time deviation between the acquisition of these two images, for example, from the determination of the initial image I1.1h (correspondingly, the tail image IL.1h) of the droplet in the high linear image series corresponding to the high camera and the determination of the initial image I1.1b (correspondingly, the tail image IL.1b) of the droplet in the low linear image series corresponding to the low camera. For two synchronized high and low linear cameras with the same acquisition constant Ta (i.e., acquisition frequency Fa = 1 / Ta), this time deviation is equal to Od / Fa (correspondingly, Of / Fa), where Od is the number of image acquisition cycles between the acquisition of the initial image I1.1h of the droplet in the high linear image series and the acquisition of the initial image I1.1b of the droplet in the low linear image series corresponding to the low camera (correspondingly, Of is the number of image acquisition cycles between the acquisition of the tail image IL.1h of the droplet in the high linear image series and the acquisition of the end image IL.1b of the droplet in the low linear image series).

[0143] We can also obtain the actual distance D between the two intercept points of the high and low linear cameras of the measuring device, respectively, by constructing, but preferably by, for example, through the previous steps of calibrating the measuring device. D is a function of the set of cameras under consideration, but may also depend on the position of the droplet's path in the high and low images of the set of cameras.

[0144] The average vertical velocity Vz_moy_d at the beginning of the droplet (and correspondingly, the average vertical velocity Vz_moy_f at the end of the droplet) is the actual distance D between the two intercept points (high and low) divided by the time deviation between the two initial images I1.1h and I1.1b of the droplet in the two series (and correspondingly, the time deviation between the two end images IL.1h and IL.1b of the droplet). Returning to the examples in Figures 4 and 5A to 5E, we have, for example:

[0145] Vz_moy_d = D / (td1b–td1h); and / or

[0146] Vz_moy_f=D / (tf1b–tf1h)

[0147] It can also be written as:

[0148] Vz_moy_d = D * Fa / Od; and / or

[0149] Vz_moy_f=D*Fa / Of

[0150] Note that this average vertical velocity can be obtained using only two of the four series of linear images (corresponding to a set of linear cameras including a high camera and a low camera (we indicate the use of the first set of cameras in the example above)). However, we can perform the same calculation using two other series of linear images (corresponding to another set of linear cameras including another high camera and another low camera). If the two sets of cameras are arranged such that they have the same high intercept point and the same low intercept point, the result should be the same. In reality, measurement errors may produce slightly different results. In this case, we can choose to determine the average vertical velocity of the droplet at a given end between the high intercept point and the low intercept point as the arithmetic mean of two average vertical velocities at that end, calculated separately using either of the two sets of linear cameras.

[0151] Furthermore, the method provides a computer-calculated instantaneous vertical translational velocity of at least one given end of a droplet as it passes through one of a high intercept point and a low intercept point. This instantaneous velocity can actually be derived from the time deviation between the acquisition of a high linearity image and the acquisition of a low linearity image of the given end, and by applying the kinematic laws of free fall.

[0152] In fact, for the sake of simplicity, we arbitrarily choose t0 = 0 as the time origin, at which point droplet P begins its vertical free fall along a vertical path, influenced only by the Earth's gravitational force "g", with an initial vertical velocity of zero. For example, this moment can correspond to the moment droplet P is cut by shearing machine 22 and begins to fall. Therefore, starting from this moment (as the measurement time origin), the instantaneous vertical velocity of any point on the droplet evolves as a function of time t according to the following relationship:

[0153] Vz_inst_par(t)=g*t

[0154] Note that choosing a time origin t0 = 0 corresponding to the zero velocity moment is not mandatory. The time origin t0 can be chosen arbitrarily, but preferably after the moment when droplet P is cut by shearing machine 22 and begins to fall. Therefore, we can choose a time origin t0 between the moment droplet P is cut by shearing machine 22 and the moment when at least one of the two high-angle cameras captures a first image containing droplet P. Under this convention, at this time origin t0 = 0, the velocity of the droplet is the velocity v0 of the droplet at the time origin, which is not zero. However, even in this case, the droplet velocity v0 at the time origin can be determined in four different ways. Using the droplet's starting point and the first set of cameras, the droplet velocity v0 at the time origin is given by the following relationship: v0 = D / (td1b - td1h) - 0.5 * g * (td1b + td1h). Using the droplet's tail point and a second set of cameras targeting both the droplet's starting and tail points, we can determine a total of four values ​​of v0 using the following relationship:

[0155] v0=D / (tf1b-tf1h)-0,5*g*(tf1b+tf1h)

[0156] v0=D / (td2b-td2h)-0,5*g*(td2b+td2h)

[0157] v0=D / (tf2b-tf2h)-0,5*g*(tf2b+tf2h).

[0158] Please note that, still at the time origin t0 after the moment when the droplet P is cut by the shearing machine 22 and begins to fall, and assuming that the velocity v0 of the droplet at the time origin is arbitrarily chosen, we can choose to use the arithmetic mean of the values ​​obtained through the above four relationships to average the unavoidable measurement error.

[0159] In both cases, if we consider, for example, the starting end Pd of droplet P and the series of linear images acquired by the first set of linear cameras 32.1h and 32.1b, then between the time td1h when the starting end Pd of droplet Pd passes the high intercept point 40.1h and the time td1b when the starting end Pd of droplet Pd passes the low intercept point 40.1b, the instantaneous vertical velocity of the starting end Pd of droplet Pd from the first high intercept point is:

[0160] Vz_inst_d.1h=Vz_inst_par(td1h)

[0161] The change is at the first low cutoff point:

[0162] Vz_inst_d.1b=V_inst_par(td1b),

[0163] in,

[0164] Vz_inst_par(td1b)-Vz_inst_par(td1h)=g*(td1b–td1h)

[0165] in,

[0166] td1b–td1h=Od / Fa

[0167] Make

[0168] Vz_inst_d.1b-Vz_inst_d.1h=g(td1b–td1h)=g*Od / Fa

[0169] But due to

[0170] Vz_moy_d = [Vz_inst_par(td1b) + Vz_inst_par(td1h)] / 2 = D*Fa / Od makes:

[0171] Vz_inst_d.1h=Vz_moy_d–[Vz_inst_d.1b-Vz_inst_d.1h] / 2 Therefore, we obtain the instantaneous vertical velocity Vz_inst_d.1h of the droplet's initial end Pd at the first high intercept point 40.1h as:

[0172] Vz_inst_d.1h=(D*Fa / Od)–(g*Od / 2*Fa)

[0173] Furthermore, the instantaneous vertical velocity Vz_inst_d.1b at the initial point Pd of the droplet at the first low intercept point 40.1b is equal to:

[0174] Vz_inst_d.1b=(D*Fa / Od)+(g*Od / 2*Fa)

[0175] Similarly, we can prove that the instantaneous vertical velocity Vz_inst_f.1h of the droplet's tail end Pf at the first high intercept point 40.1h is equal to:

[0176] Vz_inst_f.1h=(D*Fa / Of)–(g*Of / 2*Fa)

[0177] Furthermore, the instantaneous vertical velocity Vz_inst_f.1b at the tail end Pf of the droplet at the first low intercept point 40.1b is equal to:

[0178] Vz_inst_f.1b=(D*Fa / Of)+(g*Of / 2*Fa)

[0179] Therefore, we use computer calculations to determine the instantaneous vertical translational velocity of the given end of the droplet when it passes through the intercept point (here, the high intercept point) (e.g., the instantaneous vertical translational velocity Vz_inst_d.1h of the starting end Pd), which corresponds to the instantaneous vertical translational velocity of the given end of the droplet.

[0180] Please note that here, for a given end, we work using a set of two image series, one high and one low, acquired by a given set of cameras including a high camera and a low camera. Therefore, based on the first high linear image series and the first low linear image series, we calculate the first instantaneous vertical translational velocity of the given end of the droplet during its passage at one of the high and low intercept points (this velocity is given for both the starting and ending ends, and for both the high and low intercept points). For the same given end (and thus involving the starting and / or ending ends, this is for the end passing at the level of the high and / or low intercept points), we can of course perform the same computational calculation using another set of two image series, one high and one low, acquired by another set of cameras including another high camera and another low camera. We can also calculate the second instantaneous vertical translational velocity of the given end of the droplet during its passage at one of the high and low intercept points based on the second high linear image series and the second low linear image series. The result should be the same. In reality, measurement errors may produce slightly different results. In this case, we can choose to determine the instantaneous vertical translational velocity of the given end of the droplet during its passage at one of the high intercept point and the low intercept point as the arithmetic mean of the first instantaneous vertical velocity and the second instantaneous vertical velocity of that end, and calculate these two average velocities using either of the two sets of linear cameras.

[0181] Therefore, for an intermediate linear image of the image including the intermediate segment of the droplet excluding its starting end Pd and tail end Pf, the image of the droplet contained in the intermediate linear image can be matched with the corresponding segment of the droplet. For this purpose, the kinematic laws of free fall, as a function of the following: the instantaneous vertical translation velocity of the droplet when its given end passes through a point corresponding to the instantaneous vertical translation velocity of the given end of the droplet, and the time elapsed between the acquisition of the considered intermediate linear image and the passing of the droplet's end at the point corresponding to the instantaneous vertical translation velocity of the given end of the droplet.

[0182] This matching is in Figure 7 and Figure 8 It is symbolically shown in the middle. As mentioned above, Figure 8This diagram shows a set of linear images acquired by a given linear camera during the time a droplet traverses within the linear field of the camera. Here, the image set consists of L consecutive linear images acquired by the linear camera 32: I1, I2, I3, ..., I(i-1), Ii, ..., I(L-2), I(L-1), IL. Furthermore, Figure 7 and Figure 8 An axis S representing the vertical direction is shown, on which the vertical coordinate S1 corresponds to the starting end Pd of the droplet. This vertical coordinate S1 is a relative coordinate associated with the droplet P, more specifically, with the starting end Pd of the droplet, and it does not necessarily correspond to the absolute coordinates of the starting end Pd of the droplet in space or on the theoretical free-fall path. Therefore, we can arbitrarily adopt the convention that S1 = 0. Similarly, we arbitrarily define the value of S to increase along the direction from the starting end Pd of the droplet to the tail end Pf of the droplet. Thus, axis S is a reference frame associated with the droplet, which is a single vertical scale axis. As mentioned above, we know how to computationally identify the starting image I1 corresponding to the starting end Pd of the droplet in a linear image set of the droplet, so we can match the starting image I1 with the vertical coordinate S1 representing the starting end Pd of the droplet.

[0183] Next, we point out that the time deviation between the acquisition of any two images in the same series is deterministic. For example, the time deviation between the acquisition of two consecutive images in the same series is constant or follows a known function. Thus, for all intermediate linear images Ii after the initial image I1 in the set (i.e., in the example, images I2, I3, ..., I(i-1), Ii, ..., I(L-2), I(L-1), IL), we can determine the acquisition time ti of the intermediate linear images.

[0184] To perform this matching, we can first determine the positional deviation hi between the two considered segments of the droplet, the images of which are contained in two consecutive linear images I(i-1) and Ii of a given linear image series.

[0185] These two images were taken at two consecutive acquisition times t(i-1) and ti.

[0186] We can define, at time ti, as the segment Pi of the droplet at the cutoff point corresponding to the series under study (therefore, if we consider the first high cutoff point series, it is the first high cutoff point), with the vertical coordinate of segment Pi on the droplet being Si = S(ti). Therefore, segment Pi is the segment appearing in image Ii. Similarly, the vertical coordinate S[i-1] = S(t(i-1)) is the vertical coordinate of segment P(i-1) of the droplet at time t at the same cutoff point corresponding to the series under study (therefore, also the first high cutoff point). Therefore, segment P(I-1) is the segment appearing in image I(i-1).

[0187] Therefore, under this convention, the positional deviation hi between the two considered segments of the droplet can be written as: hi = S(ti) – S(t(i-1)), where the respective images of the two considered segments of the droplet are contained in two consecutive linear images I(i-1) and Ii of a given series of linear images. The positional deviation hi can also be considered as the relative displacement of the droplet with respect to the linear field of the considered camera between times ti-1 and ti, the velocity of which is the instantaneous velocity of the droplet during free fall, for example, Vz_inst_f.1h at time ti.

[0188] We can take the acquisition time t1 corresponding to the initial image I1 of the considered series as a reference. However, for the first high series, this time t1 is the time td1h discussed above, corresponding to the moment when the droplet's initial end Pd passes through the first high intercept point 40.1h. As can be seen above, S1 refers to the vertical coordinate corresponding to the droplet's initial end Pd. Therefore, we have S(t1) = S1. As can be seen above, we can adopt the convention that S(t1) = S1 = 0.

[0189] However, generally speaking, we assume that for any point on the droplet, the instantaneous vertical velocity of the droplet is:

[0190] Vz_inst_par(ti)=g*(ti-t1)+Vz_inst_par(t1)

[0191] By integrating over time, we determined the vertical coordinate S(ti) of the segment Pi of the droplet at time ti, corresponding to the intercept point of the studied series, on the droplet. Therefore, this segment appears in the corresponding series image Li:

[0192] S(ti)=(1 / 2)*g*(ti–t1)^2+Vz_inst_par(t1)*(ti-t1)+S1

[0193] Therefore, for the first high series, when t1 = td1h and denoted as Vz_inst_par(td1h) = Vz_inst_d.1h, we have:

[0194] Vz_inst_par(ti)=g*(ti–td1h)+Vz_inst_d.1h

[0195] =g*(ti–td1h)+(D*Fa / Od)–(g*Od / 2*Fa)

[0196] Furthermore, by integrating over time, we determined the vertical coordinate S(ti) of the droplet segment at the first high intercept point at time ti on the droplet. Therefore, this segment appears in the image Li of the first high series:

[0197] S(ti)=(1 / 2)*g*(ti–td1h)^2+Vz_inst_d.1h*(ti-td1h)+S1

[0198] In this way, we can easily calculate the positional deviation mentioned above: hi = S(ti) – S(t(i-1)).

[0199] We can also easily calculate the height H of droplet P in a given series using the formula H = SL – S1 = S(tL) – S(t1). The height H of droplet P is the difference between the vertical coordinate SL corresponding to the tail end Pf of the droplet (obtained at time tL, so SL = S(tL)) and the vertical coordinate S1 corresponding to the starting end Pd of the droplet.

[0200] In the example shown, the time deviation between the acquisition of two consecutive images in the same series is constant, such that for the first high series, ti–td1h=i*Ta=i / Fa, therefore we can obtain:

[0201] S(ti)=(i^2)*g / (2*Fa^2)+i*(D / Od)-(g*Od / 2*Fa)*i / Fa+S1

[0202] =(i^2)*g / (2*Fa^2)+i*[(D / Od)–(g*Od / (2*Fa^2)]+S1

[0203] Therefore, we can express the positional deviation hi of the droplet segment Pi at the intercept point corresponding to the series under study at time ti as a function of the following terms:

[0204] - The number of linear images determined in the series, on the one hand, is a continuous linear image including one of the two considered segments, and on the other hand, is the number of linear images between the starting linear images of the series determined in the above example, including the image of the starting end of the droplet.

[0205] - The instantaneous vertical translational velocity of the droplet's starting end as it passes through a defined series of intercept points (high intercept points in the example);

[0206] - The acquisition frequency "Fa" of a defined series of linear images;

[0207] - Gravitational constant "g".

[0208] We can express the positional deviation hi of the droplet segment Pi at the intercept point corresponding to the series under study at time ti as a function of the following terms:

[0209] - The number of linear images determined in the series, on the one hand, is one Ii of a continuous linear image including one of the two considered segments, and on the other hand, is the number of linear images between the tail linear images of the series determined in the other example above, including the tail end of the droplet.

[0210] - The instantaneous vertical translational velocity of the droplet's tail end as it passes through a defined series of cut-off points (high cut-off points in the example).

[0211] - The acquisition frequency "Fa" of a defined series of linear images;

[0212] - Gravitational constant "g".

[0213] Based on other examples, we can use a low-linearity image series.

[0214] Therefore, for each linear image in a set of linear images acquired by a given linear camera during the linear field time of the droplet passing through the camera, we are guided to determine the position of the imaged droplet segment on the droplet. In this way, we can determine the geometric features (segment width, absolute position in space) at the intercept point corresponding to the given linear camera by analyzing all or part of the image set via computer.

[0215] By performing this matching operation on several segments of the droplet, which is for two series of linear cameras (including high and low cameras) corresponding to the first set of devices, we can determine the horizontal displacement characteristics of these segments in the segment between the high intercept point and the low intercept point along the first horizontal measurement direction.

[0216] By performing this matching operation on several segments of the droplet, which is for two series of the second set of linear cameras (including high and low cameras) corresponding to the device (the second set of cameras is different from the first set of cameras and observes its corresponding intercept points along an observation axis different from that of the first set), we can determine the horizontal displacement characteristics of these segments in the segment between the high intercept point and the low intercept point along the second horizontal measurement direction.

[0217] Therefore, using the measuring device 30, we can perform this matching operation on several sections of the droplet, which is for four series of linear images corresponding to the first and second sets of linear cameras of the device 30.

[0218] Specifically, we can therefore, for example, provide a computationally determined range for at least two distinct sections of the droplet:

[0219] - The segment is the set of first and second horizontal translations along different first and second horizontal measurement axes between the high and low intercept points, or

[0220] - The segment is a set of first and second level horizontal translation velocities along different first and second level horizontal measurement axes between the high intercept point and the low intercept point.

[0221] The horizontal translation of a segment along the horizontal measurement axis corresponds to the horizontal offset, or the difference in the horizontal position of the segment in the space between the high intercept point and the low intercept point.

[0222] Determining the horizontal translation of a segment between a high cutoff point and a corresponding low cutoff point can, for example, include detecting the position of at least the same point of the segment in the corresponding high and low linear images of the same set of cameras.

[0223] The same point in the segment determined in the corresponding high and low linear images of the high and low series can be the edge point of the segment, thus corresponding to the edge of the segment of the droplet as seen in the linear image, and thus corresponding to the point on the outer wall of the droplet. Here, it is worth recalling that the droplet is generally cylindrical or substantially cylindrical.

[0224] The same point in a segment can be the midpoint between two edge points of the segment, so these two edge points correspond to the two edges of the segment of the droplet as seen in the linear image, and thus correspond to two opposite points on the outer diameter of the droplet when the droplet is cylindrical or generally cylindrical.

[0225] Figure 9The illustration schematically demonstrates the possibility of determining the position of a point in a droplet segment within a plane defined by the optical axes of two cameras belonging to the same set of cameras, using the device 30 according to the invention. More specifically, we will explain below a method for determining the position of a point in a droplet segment using two images of the same segment belonging to two series of linear images, each series of linear images being acquired by a first and a second camera of two linear cameras belonging to the same set (high or low) of the measuring device 30. Therefore, this method for determining the position of a point in a droplet segment allows the position of the point along two different horizontal measuring axes to be obtained by applying the principles of stereo vision. This example will be described for the high camera set, but variations (decline) may be made for the low camera set “as necessary.” It will be understood that those skilled in the art, by applying the principles of stereo vision, know how to calculate the coordinates of any point seen by the two cameras relative to any reference system of the plane through in-plane triangulation.

[0226] Figure 9 Let 37.1h and 37.2h be the first and second highest observation axes, respectively, of the first high camera 32.1h and the second high camera 32.2h. Note that we have defined the observation axis of the theoretical free-fall path 28.24 given by the linear cameras 32.1h and 32.2h as the axis passing through the optical centers C1h and C2h of the lenses 33.1h and 33.2h of the cameras under consideration, and intersecting the theoretical free-fall path 28.24 (assuming at the corresponding intercept point, here the high intercept point 40.1h). If the intersection of the two optical axes of the two cameras is the high intercept point 40.1h, then the observation axis of the camera coincides with its optical axis. The two observation axes 37.1h and 37.2h are not parallel, but they meet at the high intercept point 40.1h, and therefore they are coplanar in a preferably horizontal observation plane. By definition, the high intercept point belongs to the high horizontal observation plane defined by the two observation axes. The high intercept point 40.1h belonging to the theoretical free fall path 28.24 under consideration is known because it coincides with the intersection of the two optical axes of the two cameras under consideration, and / or, for example, because of previous calibration operations.

[0227] Figure 9 The first observation beam 39.M1h passing through point M is shown, and the position of point M in the observation plane will be determined. Point M is a point belonging to segment Pi of droplet P, located at the height of the high intercept point at the moment when both high and low linear images are acquired. Point M can be, for example, the midpoint between two edge points of segment Pi of the droplet, as seen in the linear image of that segment acquired by the first high camera. In the example, in the observation plane defined by the two observation axes, point M is offset by a distance D from the high intercept point 40.1h.

[0228] Figure 9 A second observation beam 39.M2h is shown passing through the same point M, and the position of point M in the observation plane will be determined. The second observation beam 39.M2h passes through the optical center C2h of the lens of the second camera (here, the second high camera 32.2h) and passes through point M. Point M can be, for example, the midpoint between two edge points of a droplet segment, as seen in a linear image of that segment acquired by the second camera. In the example of the midpoint, we make a simplifying assumption: the midpoint between the two edge points of the droplet segment is the same when viewed from two different angles. In this case, point M can be considered as the center of the cross-section of the droplet segment Pi.

[0229] For each of the two linear cameras 32.1h and 32.2h, we calculate the image deviation between the image of point M in the linear image of the segment acquired by that camera and a reference point in the linear image. In the example, for each camera, the reference point is the image point of that camera's observation axis, but we can also use another reference point, such as the image point of the optical axis. In the example, the combination of the reference point in the linear image acquired by the first camera and the reference point in the linear image acquired by the second camera corresponds to the intersection of the observation axes. The position of the reference point in the observation plane is obtained by calculating the intersection of the observation bundle passing through that point.

[0230] Therefore, for each camera, the image deviation is initially determined in pixels. In a known manner, we can derive from this image deviation the actual deviation of point M relative to a reference point in the linear image under consideration, i.e., in this example, the actual deviation between point M and the observation axis of the camera under consideration. For the first camera, this actual deviation is measured along a first measurement axis, and for the second camera, it is measured along a second measurement axis. The two measurement axes are respectively associated with the observation axis of the camera under consideration, here perpendicular to it. By knowing the layout and characteristics of the measuring device (e.g., the position of the optical center of the camera in the observation plane, the orientation of the optical axis, the position of the sensor, or the focal length), we know how to derive the relationship between the image deviation and the actual deviation. Thus, in this example, we obtain the actual deviation “ra” between point M and the observation axis of the first camera along a measurement axis X1 perpendicular to the first observation axis, and the actual deviation “rb” between point M and the observation axis of the second camera along a measurement axis X2 perpendicular to the second observation axis. Since the two measurement axes X1 and X2 are not parallel, the actual deviations “ra” and “rb” define the position of the midpoint M in the observation plane relative to the intercept point (here, the first highest intercept point). Furthermore, since the observation plane can be determined through calibration, we can determine the position of point M in 3D space.

[0231] Based on these two actual deviations, we can define the projected coordinates (dx, dy) of point M on a reference frame of orthogonal axes X and Y on the observation plane (which is therefore horizontal in this example), where the orthogonal axes X and Y intersect at the preceding point of the reference point in the linear image (i.e., the high intercept point 40.1h in this example). The Y-axis of this orthogonal reference frame forms an angle "a" with the observation axis 31.1h of the first linear camera 32.1h and an angle "b" with the observation axis 37.2h of the second linear camera 32.2h.

[0232] The projected coordinates (dx, dy) of point M on the reference frame with orthogonal axes X and Y are given by the following formula:

[0233] dx=(ra*sin(b)–rb*sin(a)) / (cos(a)*sin(b)–cos(b)*sin(a))

[0234] dy=(ra*cos(b)–rb*cos(a)) / (cos(a)*sin(b)–cos(b)*sin(a))

[0235] Therefore, we have demonstrated that we can computationally determine the location of at least one representative point for any segment Pi of a droplet, for which we have two high-linearity images acquired along non-parallel high observation axes 37.1h and 37.2h. This computational determination can be repeated in a similar manner for the same segment Pi for which we have two low-linearity images acquired along non-parallel observation axes 37.1b and 37.2b.

[0236] Thus, for a given but arbitrary segment Pi of a droplet P, we can determine the horizontal translation of the segment between the high and low intercept points along the horizontal measurement axes by the following: the simple difference between the positions of at least one representative point of the given segment Pi of the droplet (for which we have two high linearity images acquired along the non-parallel high observation axes 37.1h and 37.2h and two low linearity images acquired along the non-parallel low observation axes 37.1b and 37.2b). This horizontal translation of the segment can be determined along two different horizontal measurement axes. Therefore, for a given but arbitrary segment Pi of a droplet P, we obtain a set of first and second horizontal translations of the segment between the high and low intercept points along different first and second horizontal measurement axes, respectively.

[0237] In the above method, for any segment, we determine the position of points in the segment in two horizontal dimensions based on a first linear image and a second linear image from the same high or low camera group. We then point out that we can repeat the same operation based on two low linear images acquired from another high or low camera group, which allows us to obtain the position of a reference point for the same segment in two horizontal dimensions at the level of that other camera group. We then derive the first and second displacements.

[0238] Note that we can obtain the same result by first determining, in the first high series and first low series acquired by a first set of cameras including a high camera and a low camera, respectively, a first high linear image and a first low linear image corresponding to a given segment of the droplet, to determine the first horizontal translation of the given segment along the first horizontal measurement axis between the high linear image and the low linear image (and therefore between the first high intercept point and the first low intercept point). Then, we can achieve the same operation by determining, in the second high series and second low series acquired by a second set of cameras including a high camera and a low camera, a second high linear image and a second low linear image corresponding to the same given segment of the droplet, to determine the second horizontal translation of the given segment along the second horizontal measurement axis between the high linear image and the low linear image (and therefore between the first high intercept point and the first low intercept point).

[0239] These operations can be performed on any two segments of a droplet, or even on a set of segments of a given droplet, which may include more than two segments of the given droplet, and therefore may include at least three different segments of the given droplet.

[0240] Of course, it is then easy to derive the set of first and second horizontal translation velocities of the segment along different first and second horizontal measurement axes between the high and low intercept points. For this purpose, it is sufficient to determine these quantities by dividing the first and second horizontal translation amounts of the segment by the time elapsed between the acquisition of the high and low linear images of the segment.

[0241] Because for any segment, we know the set of first and second horizontal translations of the segment along different first and second horizontal measurement axes between the high and low intercept points, we can deduce the set of first and second rotations of the droplet around different first and second horizontal axes between the high and low intercept points.

[0242] In this sense, the first method requires the assumption that the deformation of the droplet during free fall is negligible. In this case, it is sufficient to select two distinct segments of the droplet (preferably two distinct segments far apart from each other along the length of the droplet). Then, we can define the droplet's directrix as a straight line passing through a first representative point belonging to the first segment of the droplet and a second representative point belonging to the second segment. We can determine the first orientation of this droplet directrix based on the positions of these two representative points in the high observation plane after passing through them, for which we have two high linearity images acquired along the non-parallel high axis at 37.1h and 37.2h for each of the two representative points in the two segments. We can also determine the second orientation of this droplet directrix based on the positions of these two representative points in the two segments, for which we have two low linearity images acquired along the non-parallel low axis at 37.1b and 37.2b for each of the two representative points in the two segments. By comparing the two orientations of the droplet's straight line, we can deduce the amount of rotation of the droplet around a first horizontal axis and a second horizontal axis that are different from each other between the high intercept point and the low intercept point.

[0243] More generally, to determine the first and second rotations of the droplet around two different horizontal axes, we can also define the droplet guideline as a straight line defined by a linear regression of representative points of several segments Pi of the droplet. We then use more than two segments. We can use all segments of the droplet, for which we have two high-linearity images and two low-linearity images acquired along the non-parallel high observation axes 37.1h and 37.2h and along the non-parallel low observation axes 37.1b and 37.2b, respectively. We can also choose to use a set of segments, including more than two segments but fewer than all segments. Therefore, we can determine the orientation of this droplet guideline for the droplet passing through the high intercept point. For this purpose, we use the positions of representative points from the two high-linearity images acquired along the non-parallel high observation axes 37.1h and 37.2h for these segments. We can also determine the orientation of the droplet guideline line for the droplet's passage at the low intercept point by using the positions of representative points from two low-linearity images acquired along the non-parallel low observation axes 37.1b and 37.2b for the aforementioned segments. By comparing the two orientations of the droplet guideline line, we can deduce the amount of rotation of the droplet around a first and a second horizontal axis, which are different from each other, between the high and low intercept points.

[0244] Regardless of the method used to determine the droplet's rotation, we can derive the set of first and second average rotational velocities of the droplet around different first and second horizontal axes, respectively, between the high and low intercept points. To this end, we divide the first and second rotational velocities of the droplet by the time elapsed between the acquisition of the high and low linearity images of the segment to determine this amount.

[0245] By treating the droplet as indeformable, we can determine the total horizontal translation of the droplet along two different directions in the same way as for rotation. For example, by determining the horizontal translation of each of at least two segments, if all segments have the same horizontal translation, then the rotation of the droplet is zero, and we determine the total translation of the droplet. Therefore, we can determine the kinematics of an indeformable droplet by calculating the average horizontal translational velocities along two different directions and the rotation about two horizontal axes.

[0246] However, the droplet may also deform during free fall. Therefore, we can consider the horizontal translation of at least three distinct segments of the droplet. If the horizontal translations of the three segments are the same, then there exists a pure overall translation of the droplet. Otherwise, we can calculate the rotation of the droplet portion included between two segments by considering the entire set of 2x2 segments. If at least two of these rotations are different, then we derive the deformation. If we divide the horizontal translation by the travel time, we obtain the average deformation velocity, expressed as the horizontal translation of several segments. In other words, we can assign a two-dimensional average horizontal translation velocity between the high and low observation planes for each segment (at least three segments). By determining the two-dimensional average horizontal translation velocity between the high and low observation planes for different segments of the same droplet, we obtain the velocity field. Analysis of the two-dimensional average horizontal translation velocity field can determine the kinematics (translation and rotation) of the droplet between the high and low intercept points, in addition to the vertical axis of rotation, and also allows us to observe and determine the amount and velocity of deformation during free fall.

[0247] Using the elements identified above, we can determine the path of at least one or more segments of the droplet based on the three dimensions of space. This path can be determined not only between the high and low intercept points, but also outside the low intercept point within the considered free-fall region. In fact, it can be seen that at the low intercept point, we can determine the position of at least one representative point for each segment based on the three dimensions of space. We can consider that in the free-fall region, the only force applied to the free-falling droplet is gravity acting vertically. Therefore, any initial horizontal translational velocity caused, for example, by shear cutting is maintained until it enters the dispenser 20. Therefore, we can predict the shape and position (centering and inclination) of the droplet when it enters the dispenser 20 using extrapolation.

[0248] Furthermore, based on the above factors, it will be possible to perform measurements of one or more geometric dimensions of the droplet.

[0249] Therefore, we can measure the first diameter of the droplet along a first horizontal direction and / or the second diameter along a second horizontal direction different from the first horizontal direction. This can be very easily achieved by pointing out the image deviation between two edge points of a segment in a series of linear images in a linear image series, these two edge points thus corresponding to the two edges of the droplet segment as seen in the linear images, and therefore, when the droplet is cylindrical or substantially cylindrical, to points opposite each other on the two diameters of the outer wall of the droplet. As shown above, we can derive the actual deviation between the two edge points of the segment from this image deviation. For the first camera, this actual deviation is measured along the first horizontal measurement axis, while for the second camera, it is measured along a second horizontal measurement axis different from the first horizontal measurement axis. The two measurement axes are here associated with the observation axis of the camera under consideration, respectively, by being perpendicular to the observation axis of the camera under consideration. By knowing the layout and characteristics of the measuring device (e.g., the position of the camera, the orientation of the observation axis, the distance of the camera to the corresponding intercept point, the focal length of the camera lens, etc.), we know how to derive the relationship between the image deviation and the actual deviation.

[0250] We can still measure the volume of the droplet, which can be calculated as the sum of the individual volumes of each imaged droplet segment. This volume will inevitably be an estimate that includes errors, depending on whether the droplet is perfectly cylindrical.

[0251] If each horizontal segment of the droplet is a cylinder of revolution, then the volume can be calculated as the sum of the individual volumes of each imaged camera segment. For this purpose, we can assign a separate volume Vi to each segment Pi, which is determined as a function of the segment's diameter Di (which can be determined as described above) and the positional deviation hi between the considered segment of the droplet and the segment whose image is contained in the preceding linear image of the given series of images as previously defined. Therefore, the basic volume can be written as:

[0252] Vi = hi * PI * Di (PI is the number PI)

[0253] As a result of the present invention, we will be able to implement a method for controlling a glass molding apparatus having a device including a glass source, at least one shear 22 and a droplet loading path, the at least one shear 22 being arranged at the outlet of the glass source 18 and actuated at regular intervals to cut continuous droplets that fall by gravity into a distributor 20, the distributor 20 guiding the droplets toward a forming cavity 16 of the apparatus along at least one droplet loading path, the droplet loading path including at least one free fall region of the droplets between the shear 22 and the distributor 20.

[0254] This method for controlling the equipment will include measuring the kinematic characteristics of the glass droplet in free fall within the droplet's free fall region between the shearer 22 and the dispenser. As described above, this measurement will include determining the amount of horizontal translation for each segment of a given droplet. The control method will advantageously include adjusting at least one operating parameter of the shearer 22 as a function of the horizontal translation of at least each segment of the set.

[0255] Adjustments may include adjusting the position of the cutting point of the shear 22. The cutting point of the shear can be defined as the point where the two blades of the shear are completely closed.

[0256] Adjustment may include adjusting the displacement speed of at least one blade of the shearing machine 22, or adjusting the displacement speed of both blades of the shearing machine 22.

[0257] Adjustments may include profiled changes to the displacement speed of at least one blade of the shear 22. Thus, one or both blades may have non-uniform displacement speeds during the droplet cutting operation between the open and closed configurations of the shear. During the displacement of at least one blade of the shear 22 between the open and closed configurations, the blade speed may, for example, have a first phase that is slower or faster than the second phase, to the extent that it is adjustable based on one or more kinematic characteristics of the glass droplet falling freely in the free-fall region.

[0258] Adjustments may include adjusting the lubrication parameters of the shearing machine 22.

[0259] In this control method, measuring the kinematic characteristics of a glass droplet falling freely in the free fall region may include, as described in the measurement method above, determining at least two horizontal components of the horizontal translational velocity of at least one segment of the droplet between a high intercept point and a low intercept point.

[0260] In this control method, measuring the kinematic characteristics of a glass droplet in free fall in the free fall region may include, as in the measurement method described above, determining at least a first component of the droplet's rotation about a first horizontal axis. In this case, adjustment may include adjusting a component of the position of the cutting point of the shearing machine, and / or adjusting a component of the displacement velocity of at least one blade of the shearing machine, and / or adjusting a curve of the component of the displacement velocity of at least one blade of the shearing machine in a direction perpendicular to the first horizontal axis. In this control method, the measurement may further include determining at least a second component of the droplet's rotation about a second horizontal axis different from the first horizontal axis, and may then further include adjusting at least one operating parameter of the shearing machine as a function of the first component of the droplet's rotation about the first horizontal axis and the second component of the droplet's rotation about the second horizontal axis.

[0261] More generally, a method for controlling equipment may include measuring the kinematic characteristics of a glass droplet in free fall in the region of free fall between the shearer and the dispenser, having one or more characteristics of the measurement method described above.

[0262] In this method for controlling the equipment, we can therefore drive any operating parameter of the equipment, particularly one or more of the movement and / or position parameters of one or more plungers 17 and / or cylinders 19 of the glass source 18, by taking into account the measurement results of the kinematic characteristics of the droplet in free fall. This control method aims to improve the consistency of the droplet and its entry into the distributor 20.

[0263] As with any method sensor, the role of measurement is clearly to monitor the method and thus adjust it to achieve the desired results in terms of quality and productivity. Measurement results can be displayed to the operator adjusting the equipment, particularly the droplet dispenser 20, or provided via a computer link to a method calculator that automatically corrects the components of the glass source 18 and / or dispenser 20, thereby enabling servo control of the method.

[0264] Time-based monitoring of these measurements can also predict and correct for method drift.

[0265] By learning, the measurement results of the device according to the invention can be easily correlated with the qualified results of the manufactured articles. These measurements of the droplets can then be determined statistically, thereby ensuring good container quality. This information can then be readily used to create servo control of the glass source 18 and / or the distributor 20 directly based on the measurements of the droplets directly below the shearer 22, upstream of the distributor 20, or at the outlet of the distributor 20.

[0266] We mentioned above the possibility of having at least two devices for measuring the kinematic characteristics of glass droplet free fall, one device installed for the high free fall region 24 and another device installed for the low free fall region 26 of the same droplet loading path between a given glass source and a given forming cavity. Therefore, we are able to obtain the kinematic characteristics of droplet free fall at two locations along the droplet path. By combining these characteristics, we can determine information about the droplet's velocity loss or gain, the droplet's deformation, and the droplet's elongation in the distributor 20. We can also improve the operation of loading the droplet into the mold by determining the kinematics of free fall at the mold inlet and monitoring the droplet using known common settings of the distributor (accelerator, guide position, lubrication, cooling).

[0267] Generally, the acquisition period Ta of the linear camera and therefore the acquisition frequency Fa are preferably considered fixed and known. If they change, they are considered to be known at any time by the computing unit. The acquisition period is the period between the start of the integration of two acquisitions of two consecutive images. The integration time is generally less than the acquisition period Ta. The linear camera can be of any known type, including cameras called Tdi (time delay and integration) cameras. The acquisition performed by the camera occurs during the free fall of the droplet. This condition is satisfied when the distance of all camera fields 34 from the shear 22 is greater than the length of the droplet. The camera can be driven using signals indicating the shearing moment of the shear 22 (e.g., the moment of command closure or the moment when the shear 22 is fully closed) and a time window is determined during which the droplet is expected to be observed.

Claims

1. A method for measuring the kinematic characteristics of a glass droplet (P) in free fall in a glass molding apparatus, comprising measuring the free fall region (24, 26) of the droplet along the droplet loading path between the glass source and the forming cavity, wherein, The droplet has a theoretical vertical free-fall path (28.24, 28'.24) in the free-fall region, and the droplet has a starting end (Pd) and a tail end (Pf), as well as the droplet length between the starting end and the tail end: - Acquire at least four series of consecutive linear digital images using four different linear cameras (32.1h, 32.1b, 32.2h, 32.2b). Each linear camera has a linear photoelectric sensor, a lens with an optical center (33.1h, 33.1b, 33.2h, 33.2b), and an optical axis (36.1h, 36.1b, 36.2h, 36.2b) defining the observation linear field (34.1h, 34.1b, 34.2h, 34.2b) for the camera under consideration. Each image in a given series of linear images is a corresponding linear camera image. The head observes linear images of linear fields, the four series including a first high series acquired by a first high camera (32.1h), a first low series acquired by a first low camera (32.1b), a second high series acquired by a second high camera (32.2h), and a second low series acquired by a second low camera (32.2b), the linear images of the four series respectively corresponding to the linear images of the first high linear field (34.1h), the first low linear field (34.1b), the second high linear field (34.2h), and the second low linear field (34.2b), such that: The first high linear field and the first low linear field each extend along a corresponding plane defined by the corresponding optical axis and a horizontal extension direction perpendicular to the corresponding optical axis. The first high linear field and the first low linear field each intercept the theoretical free fall path at the first high intercept point (40.1h) and the first low intercept point (40.1b), respectively. The first high intercept point and the first low intercept point are offset from each other by a perpendicular offset according to the theoretical free fall path. The second high linear field and the second low linear field each extend along a corresponding plane defined by the corresponding optical axis and a horizontal extension direction perpendicular to the corresponding optical axis. The second high linear field and the second low linear field each intercept the theoretical free fall path at the second high intercept point (40.2h) and the second low intercept point (40.2b), respectively. The second high intercept point and the second low intercept point are offset from each other along the direction of the theoretical free fall path. The images (I1, ..., Ii, ..., IL) in the four series of linear images each include images of corresponding clipping points acquired along the first high observation axis (37.1h), the first low observation axis (37.1b), the second high observation axis (37.2h), and the second low observation axis (37.2b), respectively. The observation axis of each clipping point of the corresponding linear camera is contained in the linear field (34.1h, 34.1b, 34.2h, 34.2b) of the corresponding linear camera (32.1h, 32.1b, 32.2h, 32.2b), passes through the optical center of the lens of the camera, and passes through the corresponding clipping point of the theoretical free fall path. The projections of the first and second highest observation axes (37.1h and 37.2h) onto a plane perpendicular to the theoretical free fall path together form high observation deviation angles (Aobsh and Aobsh') around the theoretical free fall path, which are different from 0° and different from 180°. The first and second low observation axes (37.1b, 37.2b) together form low observation deviation angles (Aobsb, Aobsb') around the theoretical free fall path in the projection of the first low observation axis and the second low observation axis (37.1b, 37.2b) onto a plane perpendicular to the direction of the theoretical free fall path. These angles are different from 0 degrees and different from 180 degrees. The time deviation between any two images acquired in the same series and between any two images from two different series is determinable. The measurement method includes: computer identification of high-linearity images and low-linearity images, both of which include images of the same given end of the droplet at the beginning and end of the droplet; and computer derivation of the time deviation between the acquisition of the high-linearity image and the acquisition of the low-linearity image from the identification. Based on the time deviation between the acquisition of high-linearity images and low-linearity images, and by applying the laws of free fall kinematics, the computer calculates the instantaneous vertical translational velocity of the droplet's given end as it passes through one of the high and low intercept points; and For an intermediate linear image (Ii) of the image of a segment of a droplet excluding its starting and ending ends, the image of the droplet contained in the intermediate linear image is matched to the corresponding segment (Pi) of the droplet by applying the kinematics of free fall as a function of the instantaneous vertical velocity of the droplet at the intercept point corresponding to the instantaneous vertical velocity of the droplet at the given end, and the time elapsed between the acquisition of the intermediate linear image and the passage of the droplet at the intercept point corresponding to the instantaneous vertical velocity of the droplet at the given end. Furthermore, the measurement method includes a computer determining at least one of the following: For at least two distinct segments of the droplet, the set of the first horizontal translation and the second horizontal translation of the segment along a first horizontal measurement axis (X1) and a second horizontal measurement axis (X2) that are distinct from each other between the high intercept point and the low intercept point, or the set of the first horizontal translation velocity and the second horizontal translation velocity of the segment, and / or The set of the first rotational amount and the second rotational amount of the droplet, or the set of the first average rotational velocity and the second average rotational velocity of the droplet, respectively, around a first horizontal axis (X1) and a second horizontal axis (X2) that are different from each other between the high interception point and the low interception point, and / or The amount of deformation of the droplet or the average deformation rate of the droplet between the high cut-off point and the low cut-off point, and / or The path of at least one or more segments of the droplet based on the three dimensions of space.

2. The measurement method according to claim 1, characterized in that, The measurement method includes a computer determining a positional deviation (hi) between two considered segments (Si, S[i-1]) of a droplet as a function of the following, wherein the respective images of the two considered segments of the droplet are contained in two consecutive linear images (Ii, I(i-1)) of a given series of linear images: - The number of linear images between one of the two considered segments in a defined series, which includes one of the continuous linear images (Ii, I(i-1)) on one side, and one of the starting linear image (I1) and the tail linear image (IL) in a defined series on the other side, where the starting linear image (I1) and the tail linear image (IL) respectively include images of the beginning and end of the droplet. - The instantaneous vertical translational velocity of the given end of the droplet when the given end of the droplet passes through the intercept points corresponding to a defined series; - The acquisition frequency of a defined series of linear images; and - Gravitational constant.

3. The measurement method according to claim 1, characterized in that, The measurement method includes, based on a first high-series linear image and a first low-series linear image, as well as a second high-series linear image and a second low-series linear image, calculating by computer a first instantaneous vertical translational velocity and a second instantaneous vertical translational velocity of the given end (Pd, Pf) of the droplet during its passage through one of the high intercept point and the low intercept point (40.1h, 40.1b, 40.2h, 40.2b). The method is characterized in that the instantaneous vertical translational velocity of the given end (Pd, Pf) of the droplet during its passage through the intercept point is determined by calculation as the average of the first instantaneous vertical translational velocity and the second instantaneous vertical translational velocity during the passage of the given end of the droplet at the intercept point.

4. The measurement method according to claim 2, characterized in that, The measurement method includes using a computer to calculate the droplet height between the starting end (Pd) and the tail end (Pf) by summing the positional deviations (hi) of all consecutive linear images in a defined series from the starting linear image (I1) to the tail linear image (II).

5. The measurement method according to claim 1, characterized in that, The measurement method includes: for a set of several segments of a given droplet, the computer determines: In the first high series of the first high linear image and the first low series of the first low linear image corresponding to each segment, determine the first horizontal translation amount of each segment of the set between the first high intercept point and the first low intercept point (40.1h, 40.1b), and between the high linear image and the low linear image, along the first horizontal measurement axis (X1). In the second high series of the second high linear image and the second low series of the second low linear image corresponding to each segment, the second horizontal translation amount of each segment along the second horizontal measurement axis (X2) between the first high intercept point and the first low intercept point (40.2h, 40.2b) is determined; Furthermore, the measurement method includes the step of deriving the horizontal translation amount of each segment of the set in a computational manner: The two horizontal components of the average translational velocity of the droplet along two different horizontal axes between the high and low intercept points; and / or The droplet rotates around two horizontal axes at two angles between the high and low cut-off points; and / or The deformation of the droplet during its fall between the high and low cut-off points.

6. The measurement method according to claim 1, characterized in that, The measurement method includes: computer measurement of at least one geometric dimension of the droplet from the following categories: - The first diameter of the droplet along the first horizontal direction - A second diameter along a second horizontal direction different from the first horizontal direction. - The length or height of the droplet. - The volume of the droplet.

7. The measurement method according to claim 1, characterized in that, The computer determines the horizontal translation of a segment (Pi) between a high cutoff point and a corresponding low cutoff point by detecting the position of at least one point (M) of the segment (Pi) in the corresponding high-linearity image of the high series and the corresponding low-linearity image of the low series.

8. The measurement method according to claim 7, characterized in that, The same point (M) is an edge point of a segment, the midpoint between two edge points of a segment, or one of the points whose image is identifiable in both high-linearity and low-linearity images.

9. The measurement method according to claim 1, characterized in that, The first high observation axis (37.1h) and the first low observation axis (37.1b) overlap vertically in the same vertical plane.

10. The measurement method according to claim 1, characterized in that, The first high observation axis (37.1h) and the first low observation axis (37.1b) are parallel to each other.

11. The measurement method according to claim 1, characterized in that, The first high observation axis (37.1h) and the first low observation axis (37.1b) are perpendicular to the theoretical free fall path (28.24).

12. The measurement method according to claim 1, characterized in that, The first high observation axis (37.1h) and the second high observation axis (37.2h) intercept the same high intercept point (40.1h) of the theoretical vertical free fall path, and / or the first low observation axis (37.1b) and the second low observation axis (37.2b) intercept the same low intercept point (40.1b) of the theoretical vertical free fall path.

13. A measuring device for measuring the kinematic characteristics of a glass droplet (P) in free fall in a glass molding apparatus (10), said measuring device comprising measuring the free fall region (24, 26) of the droplet along the droplet loading path between the glass source (18) and the forming cavity (16), wherein, The droplet has a theoretical vertical free-fall path (28.24, 28'.24) in the free-fall region, and the droplet has a starting end (Pd) and a tail end (Pf), as well as the droplet length between the starting end and the tail end: - At least four different linear cameras (32.1h, 32.1b, 32.2h, 32.2b), each linear camera having a lens with an optical center (33.1h, 33.1b, 33.2h, 33.2b) and an optical axis (36.1h, 36.1b, 36.2h, 36.2b) defining the observation linear field for the camera under consideration, said at least four different linear cameras including a first high camera (32.1h), a first low camera (32.1b), a second high camera (32.2h), and a third low camera (32.1b). Two low-light cameras (32.2b) are configured, each having a distinct first high optical axis (36.1h), first low optical axis (36.1b), second high optical axis (36.2h), and second low optical axis (36.2b). Each camera can generate a digital image (I1, ..., Ii, ..., IL) of the observed linear field, which are respectively the first high linear field (34.1h), the first low linear field (34.1b), the second high linear field (34.2h), and the second low linear field (34.2b). The first high linear field (34.1h) and the first low linear field (34.1b) each extend along a corresponding plane defined by the corresponding optical axis and a horizontal extension direction perpendicular to the corresponding optical axis. The first high linear field and the first low linear field each intercept the theoretical free fall path at the first high intercept point (40.1h) and the first low intercept point (40.1b), respectively. The first high intercept point and the first low intercept point are offset from each other according to the theoretical free fall path. The second high linear field (34.2h) and the second low linear field (34.2b) each extend along a corresponding plane defined by the corresponding optical axis and a horizontal extension direction perpendicular to the corresponding optical axis. The second high linear field and the second low linear field each intercept the theoretical free fall path at the second high intercept point (40.2h) and the second low intercept point (40.2b), respectively. The second high intercept point and the second low intercept point are offset from each other along the direction of the theoretical free fall path. The high optical axes (36.1h, 36.2h) are different from each other in their vertical projections onto a plane perpendicular to the theoretical free fall path, and the low optical axes (36.1b, 36.2b) are different from each other in their vertical projections onto a plane perpendicular to the theoretical free fall path.

14. The measuring device according to claim 13, characterized in that: The projections of the high optical axes (36.1h, 36.2h) onto a plane perpendicular to the theoretical free fall path together form a high optical axis deviation angle (Aopth) around an axis parallel to the theoretical free fall path that passes through the convergence point of the projections of the two optical axes in the projection plane. This angle is different from 0 degrees and different from 180 degrees. The projections of the low optical axes (36.1b, 36.2b) onto a plane perpendicular to the theoretical free-fall path together form a low optical axis acute angle deviation angle (Aoptb) around an axis parallel to the theoretical free-fall path that passes through the convergence point of the projections of the two optical axes in the projection plane. This angle is different from 0 degrees and different from 180 degrees.

15. The measuring device according to claim 13, characterized in that, The measuring device includes an electronic computing unit (42) programmed to implement the measuring method according to any one of claims 1 to 12.

16. The measuring device according to claim 13, characterized in that, The first high linear field (34.1h) and the first low linear field (34.1b), as well as the second high linear field (34.2h) and the second low linear field (34.2b), each intercept the theoretical free fall paths (28.24, 28'.24) of several glass droplets formed simultaneously from the same glass source (18).

17. A control method for a glass product molding apparatus (10), the glass product molding apparatus (10) comprising a glass source (18), at least one shear (22), and a droplet loading path, said at least one shear (22) being arranged at the outlet of the glass source and actuated at regular intervals to cut continuous droplets (P) falling by gravity into distributors (20a, 20b, 20c), the distributors guiding the droplets toward the forming cavity (16) of the glass product molding apparatus along at least one droplet loading path, the droplet loading path including at least one droplet free fall region (28.24, 28'.24) between the shear (22) and the distributor (20), characterized in that, The control method includes measuring the kinematic characteristics of the free fall of the glass droplet in the free fall region (24) of the droplet between the shearer (22) and the dispenser (20), the measurement including determining the horizontal translation amount of each segment of a given set of segments of the droplet, and characterized in that the control method includes adjusting at least one operating parameter of the shearer (22) as a function of the horizontal translation amount of at least each segment of the set.

18. The control method according to claim 17, characterized in that, Adjustments include adjusting the position of the cutting point of the shearing machine (22).

19. The control method according to claim 17, characterized in that, Adjustments include adjusting the displacement speed of at least the blade of the shearing machine (22).

20. The control method according to claim 17, characterized in that, The adjustment includes adjusting the curve of the displacement speed of at least one blade of the shearing machine (22).

21. The control method according to claim 17, characterized in that, Adjustments include adjusting the lubrication parameters of the shearing machine (22).

22. The control method according to claim 17, characterized in that, The measurement includes determining at least two horizontal components of the horizontal translational velocity of at least one segment of the droplet between the high intercept point and the low intercept point.

23. The control method according to claim 17, characterized in that, The measurement includes determining at least a first component of the rotation of the droplet about a first horizontal axis, and is characterized in that the adjustment includes: adjusting the component of the position of the cutting point of the shear (22), and / or adjusting the component of the displacement velocity of at least the blade of the shear (22), and / or adjusting the curve of the component of the displacement velocity of at least one blade of the shear (22).

24. The control method according to claim 23, characterized in that, The measurement includes determining at least a second component of the rotation of the droplet around a second horizontal axis different from the first horizontal axis, and is characterized in that the control method includes adjusting at least one operating parameter of the shearing machine (22) as a function of a first component of the rotation of the droplet around the first horizontal axis and a second component of the rotation of the droplet around the second horizontal axis.

25. The control method according to claim 17, characterized in that, The measurement of the kinematic characteristics of the free fall of the glass droplet is achieved according to any one of claims 1 to 12.

Images