Method for monitoring a facility for manufacturing glass containers in a hot sector by obtaining an image

Abstract

The invention relates to a method for monitoring a facility for manufacturing glass containers (2) having a ring, implemented by a monitoring system, the method comprising, downstream of the system for manufacturing the containers and upstream of the passage of the containers through an annealing lehr (6) of the facility, for at least one container moving on a conveyor; - obtaining at least one image comprising at least the surface of the ring of the container; - detecting, by processing the at least one image, at least one defect of the ring of the container included in a list of container ring defects comprising at least the seam defect and the unrendered ring defect.

Description

Description Title of the invention: Method for controlling a glass container manufacturing installation in the hot sector by obtaining an image Technical Field

[0001] The present invention relates to the field of manufacturing containers formed using molds, for example bottles, jars, or flasks, and more specifically to the detection of defects affecting these containers. The invention also relates to the control of manufacturing parameters to avoid these defects. Prior art

[0002] Glass container manufacturing plants consist of at least two sectors, called the cold sector and the hot sector. These two sectors are delimited by an annealing arch, with the containers passing from the hot sector to the cold sector through this arch. In fact, in the cold sector, the containers cool after passing through the annealing arch.

[0003] Containers are generally inspected in the cold sector, approximately between 35 minutes and 2.5 hours after their formation (in particular, because it is easier to handle cooled containers). When the detection of a defect appears to be linked to the container manufacturing process, the modification of the process parameters is therefore carried out at least after this period of 35 minutes to 2.5 hours. All containers manufactured during this period may be affected by the same defect (for example, all containers manufactured within the same mold, if the mold is defective and causes the defect), which results in numerous losses. The use of inspections implemented in the cold sector is therefore unsatisfactory for limiting losses.

[0004] We will now present the known methods of manufacturing glass containers. It may be noted that, as in the rest of this description, the term container is used to designate a finished or even unfinished container (for example, a glass parison loaded into a blank mold may be designated as a container).

[0005] A first known process is the blow-blow process (designated by the English acronym "BB: Blow-Blow"). This process is used for example for containers with a narrow opening, narrower than the body or barrel itself (typically bottles). In this process, a glass parison ("gob" in English) is formed and is loaded into a blank mold. Compression is then implemented in which air is used to compress the parison in the blank mold so that the container ring is formed at the bottom of the blank mold. Piercing is then implemented in which the container is pierced so that compressed air is blown into the container, the walls of which conform to those of the blank mold. A blank container ("parison" in English) is then obtained.This blank is then transferred from the blank mold to a finishing mold where the blank is stretched before air is blown through the container ring to produce the final container. The container is then extracted to be transported on a conveyor line.

[0006] Another known process is the press-and-blow process, in its variant designated by the English acronym "PB: Press and Blow", adapted for wide-mouthed containers (typically pots). In this process, a glass parison is formed and loaded into a blank mold. Pressing is then carried out in which a wide punch forms the opening of the ring, the glass conforming to the surface of the punch and the walls of the blank mold. The resulting container blank is then transferred to the finishing mold for air blowing. The container is then extracted to be transported on a conveyor track.

[0007] A variant of the press-and-blow process is also known, suitable for narrow-necked containers (this variant is designated by the English acronym "NNPB: Narrow Neck Press and Blow"). In this process, a narrow punch is used to form the container neck. The other steps are similar to those of the PB process described above.

[0008] Containers manufactured by these processes can be affected by various defects. While it is known to detect defects in the hot sector, the detection of certain defects in the container ring has no solutions in prior techniques.

[0009] From the prior art, document EP3204759 is known. This document concerns the cold sector inspection of containers. It proposes inspection by light transmission for any type of container (PET, glass). In this document, a flat light source is used positioned opposite the cameras. The light source is a panel divided into two or even three lighting zones that can be controlled depending on the part of the bottle to be inspected, for example, the bottom of the panel is switched off during an inspection of the container's ring.

[0010] This document proposes a solution limited to the cold sector and which does not allow certain types of faults to be detected.

[0011] Document EP1118854 is also known, which relates to the inspection of containers in the hot sector or in the cold sector, and more specifically the inspection of the surface of container rings. In this document, a light line is projected onto the surface of the ring and the reflection of this line is observed on a chord of the ring by scanning during several increments of translation or rotation of the container. This document has the disadvantage of requiring particularly precise conveying, which is ultimately difficult to implement in the hot sector.

[0012] It may be noted that the surface of the container ring is called the upper surface of the container. The surface of the ring is a substantially horizontal surface. It is, for example, substantially horizontal when a defect is present (for example, a non-rendering defect or an unwanted inclination defect), but it may also be substantially horizontal without a defect being present when it is frustoconical (for example, inclined outwards). The ring surface may also be horizontal when the container is in its vertical position with the opening upwards, which is conventionally the conveying position of containers that have just been formed and that are conveyed in the hot sector.

[0013] Document US5617204 describes a detection of defects called burrs, which proposes rotating bottles on themselves during an inspection of their ring. The person skilled in the art knows that it is not conceivable to rotate bottles on themselves in the hot sector, where this solution cannot be used.

[0014] Other solutions are also known for inspecting container rings in particular, but these solutions have the disadvantage of requiring the implementation of precise relative centering of the device above the ring to enable their inspection. This is particularly the case in documents EP1681559, JP2010151473, US20040150815, and FR2884611. The person skilled in the art knows that the implementation of this precise centering is not possible in a hot sector, these solutions are therefore not satisfactory.

[0015] Still other known documents require precise centering above the rings. We can also cite documents EP1817574, EP1676125, EP3482193, EP3207361, and EP3735579. It can be noted that document EP3207361 makes it possible to detect flatness defects in the rings (typically saddle-shaped hollows, or more localized ones). Document EP3735579 also makes it possible to detect these defects.

[0016] Finally, document JP2003083718 is known, which describes the inspection of containers and proposes using a camera that observes the upper surface of the containers to detect anomalies. Document JP05157523 is also known, which describes the use of cameras installed above a bottle conveyor for their inspection, and which describes the detection of deformations.

[0017] The solutions presented above have drawbacks. Some of the solutions are not suitable for hot areas. Also, some defects are not detected by previous solutions. Statement of the invention

[0018] The present invention aims to remedy all or part of the drawbacks of the prior art.

[0019] To this end, a method is proposed for controlling an installation for manufacturing glass containers having a ring, implemented by a control system, the containers having been manufactured by a manufacturing system in which for each container, a parison (of glass here) is loaded into a blank mold to be transformed into a blank, said container blank being placed in a finishing mold, and in which air is blown through the ring of the container blank to obtain the container (the method therefore applies to the BB, NNPB, and PB manufacturing methods).

[0020] The method comprises, downstream of the container manufacturing system and upstream of the passage of the containers through an annealing arch of the installation (thus, the control system is placed at least in part, or even in full, upstream of the annealing arch, i.e. within the hot sector), for at least one container moving on a conveyor, - obtaining at least one image, said at least one image comprising at least the surface of the entire ring of the container (i.e. the entire surface of the ring of the container, the upper and horizontal or substantially horizontal part of the ring, said image here being a two-dimensional image acquired by at least one matrix camera), the image resulting from the light emerging from the entire surface of the ring, - a detection, by processing said at least one image, of at least one container ring defect included in a list of container ring defects including at least the burr defect and the non-returned ring defect.

[0021] The inventors of the present invention have observed that it is possible, in the hot sector, to obtain images which include at least the surface of the entire ring of the containers, to detect by image processing defects of burr and unrendered ring.

[0022] An image comprising the entire ring of the container is an image in which this surface is visible as a ring or an ellipse. The light emerging from the entire surface of the ring can be light reflected by the ring, light transmitted by the surface of the ring (after passing through the wall of the ring), or even light emitted by the surface of the ring (for example by infrared radiation due to the temperature of the container in the hot sector).

[0023] The invention therefore overcomes the difficulties linked to the late detection of defects in the cold sector.

[0024] It may be noted that the image comprising at least the surface of the ring of a container may be acquired by means of a camera operating in the visible range, and / or in the infrared range, and / or in the UV range.

[0025] According to a particular embodiment, the method further comprises an identification of at least one parameter of the manufacturing system linked to said at least one detected defect.

[0026] This identification is implemented automatically, here by the control system. Also, the identification of the parameter is subsequent to the detection of the fault.

[0027] It has been observed that the defects in the list may be linked to parameters of the manufacturing system which are such that they cause the appearance of the defects. In this particular embodiment, at least one of these parameters is identified, for example to enable the problem to be resolved within the manufacturing system. Without this being limiting, the identification may comprise a recording in a memory of the system of the identified parameter, possibly with an indication of said detected defect and possibly with said image. Without this being limiting, the identification may comprise a restitution of the identified parameter to a user, the restitution being implemented for example by means of a display (fixed or remote, or even portable such as for example a telephone or tablet screen, etc.), a sound device, etc.

[0028] Preferably, the parameter is a controllable parameter of the manufacturing system, i.e., a parameter whose value can be controlled by an operator or automatically.

[0029] According to a particular mode of implementation, the detection of said fault and / or the identification of said at least one identified parameter triggers the display of said detected fault and / or of said at least one identified parameter on a display device of the installation.

[0030] In this particular implementation mode, if no parameter has been identified, the said fault is displayed, and if a parameter has been identified, the said fault and / or the said parameter can be displayed.

[0031] Also, in this particular mode of implementation, for example when a parameter is identified (but similarly, in the following, identified parameter can be replaced by detected fault), a display device of the installation is used to display the identified parameter. The display device may be a display device arranged at any location of the installation which may be a container manufacturing plant (for example on the manufacturing system, or in an operator's cabin located inside the installation, or even on a portable system such as a telephone or tablet screen, etc.). Without this being limiting, the display of the parameter can be implemented at the same time as a display of the detected fault and / or the image obtained.

[0032] According to a particular implementation mode, the identification of said at least one identified parameter triggers the transmission of a command to modify said at least one parameter, the command being intended for the manufacturing system.

[0033] In this particular implementation mode, the control system develops a command that can be executed by the manufacturing system being modified by acting on the identified parameter. The command can be transmitted by any suitable means of communication, for example wired or wireless.

[0034] According to a particular mode of implementation, obtaining said at least one image is implemented using at least one light source and at least one image sensor.

[0035] In this particular embodiment, a light source is used, for example to illuminate the surface of the container ring. The light source may emit electromagnetic radiation in the visible and / or non-visible spectrum range such as infrared and / or UV radiation (it should be noted that non-visible radiation is less disturbing for operators). The infrared radiation may be either the radiation from the containers or the infrared light emitted by a controlled light source.

[0036] The image sensor may be a matrix image sensor adapted to capture light radiation corresponding at least in part to the light source.

[0037] According to a particular embodiment, the method comprises obtaining a first image by said image sensor using said light source, and obtaining a second image by another image sensor using another light source, wherein when obtaining the first image, an observation axis of said image sensor is aligned along a first azimuth relative to a direction of movement of the containers on the conveyor, and when obtaining the second image, an observation axis of the other image sensor is aligned along a second azimuth.

[0038] In this particular implementation, two separate image sensors are used, each configured so that, when obtaining their image, the observation axes are aligned according to different azimuths. Since these conditions are verified when obtaining the images, it is understood that the containers are arranged when obtaining the images on the conveyor in a location which allows obtaining images on which the surface of the ring is visible (this location may be the same for both sensors, or not).

[0039] It can be noted that obtaining the first image and obtaining the second image can be simultaneous: in this case, the two observation axes can be configured to intersect at the location of the surface of the container ring (when obtaining the images, i.e. for a position at a time of the container on the conveyor track). In such a way alternatively, obtaining the first image and obtaining the second image can be offset in time: in this case, the two observation axes can be configured to intersect the surface of the container ring for different positions of the conveyed container.

[0040] Light sources can be portions of a single overall source. A global source can be a light panel.

[0041] According to a particular embodiment, for a container, when obtaining said at least one image, the light source and said image sensor are arranged at least above the ring and on either side of the ring, so that at any point on the surface of the ring, there is a ray coming from the light source which reaches the point with a non-zero and non-right incident angle relative to the normal to the surface (and therefore for example measured from a horizontal surface), and which is reflected in a region of a pupil of the image sensor (the pupil being the surface, for example delimited by a diaphragm, through which a light beam can arrive on the sensor).

[0042] This can be achieved as an advantageous example when the light source comprises at least one light-emitting surface, for example continuous and extended, such as a light panel.

[0043] In this particular embodiment, a reflection image of the ring surface is obtained, which is illuminated at an angle relative to the central and vertical axis of the container. The ring surface is also observed at an angle relative to the central and vertical axis of the container.

[0044] It has been observed by the inventors of the present invention that the images obtained in this manner easily allow the detection of the defects in said list. In this particular embodiment, the optical elements can be placed far from the container (typically several tens of centimeters), which limits the exposure of the optical elements (source and camera) to the radiation of the hot containers. The field of view of the camera as well as the size of the source can be enlarged (by a selection of appropriate camera and source) to allow detection with a large tolerance (several tens of millimeters) on the centering of the article on the conveyor.

[0045] Thus, according to one example, the light source and / or said image sensor are placed so as to each be several tens of centimeters from the ring of the container.

[0046] Also, according to another example, the image sensor obtains a single image at an exposure time limiting the motion blur of the surface of the container ring. The person skilled in the art will know how to choose this exposure time to consider that the motion blur is absent from the image.

[0047] This particular implementation mode is called "plunge reflection". In fact, we use here the notion of "plunge" which applies here as much to the light source as to the image sensor.

[0048] According to a particular embodiment (which is here an alternative to the embodiment called diving reflection), for a container, when obtaining said at least one image, said image sensor is arranged at least above the ring, the image sensor and the light source being arranged on either side of the ring, so that at any point on the surface of the ring, there is a ray coming from the light source which reaches the point with a non-zero and non-right incident angle after having passed through the container, and which is transmitted (here, it emerges from the surface of the ring by transmission) into a region of a pupil of the image sensor.

[0049] In this particular embodiment, a transmission image of the ring surface is obtained which is illuminated at an angle relative to the central and vertical axis of the container. The ring surface is also observed at an angle relative to the central and vertical axis of the container, but with a negative elevation relative to the surface of the container ring.

[0050] It has been observed by the inventors of the present invention that the images obtained in this manner easily allow the detection of defects in said list.

[0051] In this particular implementation mode, the optical elements can be placed far from the container (typically several tens of centimeters), which limits the exposure of the optical elements (source and camera) to radiation. of hot containers. The camera's field of view as well as the source size can be enlarged (by selecting the appropriate camera and source) to allow detection with a high tolerance (several tens of millimeters) on the centering of the item on the conveyor.

[0052] Thus, according to one example, the light source and / or said image sensor are placed so as to each be several tens of centimeters from the ring of the container.

[0053] Also, according to another example, the image sensor obtains a single image at an exposure time limiting the motion blur of the surface of the container ring. The person skilled in the art will know how to choose this exposure time to consider that the motion blur is absent from the image.

[0054] This particular implementation mode is called "plunge transmission". Here, the concept of plunging applies to the image sensor, the light source being at a lower level than the surface of the ring.

[0055] According to a particular mode of implementation, the light source and / or the other light source are sources included in sources adapted to be able to illuminate a container over at least its entire height.

[0056] Advantageously, a light source can also be used to inspect the container over its entire height, as described in document FR 3056297 where a light-emitting surface capable of illuminating a container over its entire height is used. This particular embodiment therefore makes it possible to reuse a light source. Preferably, the light source is in an upper part of the source capable of illuminating the container over its entire height, and the light source can be controlled independently of the rest of the source capable of illuminating the container over its entire height.

[0057] According to a particular implementation mode (which is here an alternative to the implementation mode called diving reflection and to that called "diving transmission"), for a container, when obtaining said at least one image, the light source and said image sensor are arranged above the ring, the image sensor facing the ring and the light source surrounding a principal axis of the sensor, such that at any point on the ring surface, there is a ray from the light source that reaches the point, and is reflected by the ring surface into a region of a pupil of the image sensor.

[0058] For example, in this embodiment, the light source may be in the shape of a ring, a truncated cone, etc.

[0059] This particular mode of implementation is called “coaxial reflection”.

[0060] In this particular implementation mode, the optical elements can be placed far from the container (typically several tens of centimeters), which limits the exposure of the optical elements (source and camera) to the radiation of the hot containers. The field of view of the image sensor as well as the size of the source can be enlarged (by a selection of appropriate camera and source) to allow detection with a large tolerance (several tens of millimeters) on the centering of the article on the conveyor.

[0061] Thus, according to one example, the light source and / or said image sensor are placed so as to each be several tens of centimeters from the ring of the container.

[0062] Also, according to another example, the image sensor obtains a single image at an exposure time limiting the motion blur of the surface of the container ring. The person skilled in the art will know how to choose this exposure time to consider that the motion blur is absent from the image. Advantageously, to implement this mode of implementation, a telecentric lens can be used with the image sensor. For information, the telecentric lens marketed under the name "TC10M series" by the Italian company Opto Engineering could be used.

[0063] According to a particular implementation method, for each container of a plurality of containers, the following are implemented: - obtaining at least one image, said at least one image comprising at least the surface of the entire ring of the container, the image resulting from the light emerging from the entire surface of the ring, - detection, by processing said at least one image, of at least one ring defect of the container included in the list of ring defects, - an identification of at least one parameter of the manufacturing system linked to said at least one defect detected for a given number of containers of the plurality of containers.

[0064] The identification of certain parameters may be linked to the appearance of the same defect on several containers. For example, several containers may have been manufactured within the same molding section.

[0065] According to a particular implementation method, for a container, a section number and / or forming cavity number in which the container was formed is obtained.

[0066] This particular implementation method makes it possible to make the identification of said parameter even more precise. Indeed, the parameter can be linked to a molding part used in the forming section or cavity (typically the rough mold or the finishing mold or a part of one of these molds).

[0067] According to a particular mode of implementation, the parameter of the manufacturing system linked to the burr defect is a parameter of the glass parison loaded into the blank mold to form the container and / or a parameter of positioning of a punch, and / or a parameter of drilling or pressing of said blank implemented prior to blowing.

[0068] It has been observed by the inventors that not only can the burr defect be detected in the hot sector on the basis of an image of the surface of the ring, but also that this defect can be linked to parameters of the manufacturing system which are controllable, for a correction for example. Thus, by displaying the defect and / or the parameter, an operator can consult the display and for example determine a parameter to modify and act on a parameter of the parison (weight, shape, volume, etc.), and / or act on a parameter for positioning a punch (typically in a PB or NNPB manufacturing context), and / or act on a parameter for drilling or pressing said blank implemented prior to blowing (typically in a BB manufacturing context). These parameters can also, in a particular implementation mode, be displayed with the default. These parameters can also, in another particular implementation mode, be modified directly by the manufacturing system which receives an appropriate command for this modification.

[0069] According to a particular embodiment, the plurality of defects comprises the non-rendered ring defect, and in which the parameter of the manufacturing system linked to the non-rendered ring defect is a parameter of the glass parison loaded into the blank mold to form the container, or a parameter for forming the parison into a blank.

[0070] It has been observed by the inventors that not only can the non-rendered ring defect in the hot sector be detected on the basis of an image of the ring surface, but also that this defect can be linked to parameters of the manufacturing system which are controllable, for correction for example. Thus, by displaying the defect and possibly also the parameter, an operator can consult the display and for example act on a parameter of the parison (weight, shape, volume, etc.), and / or act on a loading parameter of this parison. These parameters can also, in a particular mode of implementation, be modified directly by the manufacturing system which receives an appropriate command for this modification.

[0071] According to a particular embodiment, the list of defects further includes the glaze defect, and in which the parameter of the manufacturing system linked to the glaze defect is a parameter of cooling or contact time with the glass of a molded part.

[0072] The inventors of the present invention have observed that it is possible to detect glaze defects on images of the ring surface obtained in the hot sector. Furthermore, these defects can be linked to controllable manufacturing system parameters such as cooling or contact time with the glass of a molded part.

[0073] It may be noted that detecting a defect, in the present description, means not only knowing its presence within the image, but also knowing the type of defect among the defects in the image. list. For example, when a defect of a given type is detected, this can result in the generation of a signal indicating that type. Furthermore, we can say that when a defect is detected, its type is identified.

[0074] This allows the type of fault to be displayed (for example, using this signal). It also allows a process parameter to be determined manually or automatically, and subsequently the parameter in question to be modified manually or automatically, since different types of faults can have different causes.

[0075] The invention also relates to a method for manufacturing glass containers comprising an implementation of the control method as defined above (in all its modes of implementation).

[0076] According to a particular mode of implementation, the detection of said defect and / or the identification of said at least one identified parameter triggers the display of said detected defect and / or of said at least one identified parameter on a display device of the installation, and in which the display device of the installation is included in the manufacturing system or in a control cabin of the installation.

[0077] According to a particular mode of implementation, the identification of said at least one identified parameter triggers the emission of a command to modify said at least one parameter, the command being intended for the manufacturing system, and in which the manufacturing system receives said command to modify said at least one parameter, and executes said command to modify said at least one parameter.

[0078] The command may be a command executable by a processor of the manufacturing system that controls an actuator of the manufacturing system. Alternatively, the command may be executable directly by a processor of an actuator of the manufacturing system.

[0079] The invention also provides a system for controlling a plant for manufacturing glass containers having a ring, the containers having been manufactured by a manufacturing system in which for each container, a parison is loaded into a blank mold to be transformed into a blank, said container blank being placed in a finishing mold, and in which air is blown through the ring of the container blank to obtain the container, the control system being arranged downstream of the container manufacturing system and upstream of the passage of the containers through an annealing arch of the installation, and being configured to implement: - obtaining at least one image, said at least one image comprising at least the surface of the entire ring of the container, the image resulting from the light emerging from the entire surface of the ring, - a detection, by processing said at least one image, of at least one container ring defect included in a list of container ring defects including at least the burr defect and the non-returned ring defect.

[0080] This system can be adapted for the implementation of all modes of implementation of the control process as defined above.

[0081] According to another aspect, there is provided a system for hot inspection of glass containers having a ring, comprising at least one light source and at least one image sensor, in which the light source and said at least one image sensor are configured so that when inspecting a container, said at least one light source and said at least one image sensor are arranged at least above the ring and on either side of the ring, so that at any point on the surface of the ring, there is a ray from the light source which reaches the point with a non-zero and non-right incident angle (for example with respect to the normal to the surface of the ring, for example with respect to the vertical), and which is reflected in a region of a pupil of the image sensor.

[0082] Thus, a control system is proposed which is suitable for control in the hot sector, i.e. upstream of an annealing arch.

[0083] In this particular mode of implementation, we obtain a reflection image of the surface of the ring which is illuminated at an angle relative to the axis central and vertical axis of the container. The surface of the ring is also observed at an angle to the central and vertical axis of the container.

[0084] It has been observed by the inventors of the present invention that the images obtained in this manner easily allow the detection and / or identification of the defects of said list.

[0085] In this particular implementation, the optical elements can be placed far from the container (typically several tens of centimeters), which limits the exposure of the optical elements (source and camera) to radiation from hot containers. The camera field of view as well as the source size can be enlarged (by selecting the appropriate camera and source) to allow detection with a large tolerance (several tens of millimeters) on the centering of the item on the conveyor.

[0086] Thus, according to one example, the light source and / or said image sensor are placed so as to each be several tens of centimeters from the ring of the container.

[0087] Also, according to another example, the image sensor obtains an image at an exposure time limiting the motion blur of the surface of the container ring. The person skilled in the art will know how to choose this exposure time to consider that the motion blur is absent from the image.

[0088] The control system of this other aspect can be configured to be used in any of the modes of implementation of the method described above, except the "coaxial reflection" mode of implementation and the "plunge transmission" mode of implementation. Brief description of the drawings

[0089] [Fig. IA] Figure IA shows an installation comprising a manufacturing system and a hot sector control system according to an example.

[0090] [Fig. IB] Figure IB is a top view of the installation of Figure IA.

[0091] [Fig. 2] Figure 2 is a schematic representation of a container ring.

[0092] [Fig. 3] Figure 3 is a schematic representation of a ring with a burr defect.

[0093] [Fig. 4] Figure 4 shows the orientation of an image sensor, according to an example.

[0094] [Fig. 5A] Figure 5A shows an arrangement of an image sensor and a light source, according to an example.

[0095] [Fig. 5B] Figure 5B shows an arrangement of two image sensors and a light source, according to another example.

[0096] [Fig. 5C] Figure 5C shows an arrangement of two image sensors and two light sources, according to yet another example.

[0097] [Fig. 6] Figure 6 shows different phases of operation of a multi-image sensor system.

[0098] [Fig. 7] Figure 7 shows different phases of operation of another multi-image sensor system.

[0099] [Fig. 8A] Figure 8A shows, in top view, an arrangement of an image sensor and a light source.

[0100] [Fig. 8B] Figure 8B shows, in top view, an arrangement of two image sensors and two light sources.

[0101] [Fig. 9A] Figure 9A shows, in top view, an arrangement of four image sensors and four light sources.

[0102] [Fig. 9B] Figure 9B shows, in top view, another arrangement of four image sensors and four light sources.

[0103] [Fig. 10] Figure 10 schematically shows a control system comprising an image sensor which observes a ring surface with a non-zero and non-right angle, in reflection.

[0104] [Fig. 11] Figure 11 shows the configuration of Figure 10 in top view.

[0105] [Fig. 12] Figure 12 schematically shows a control system comprising an image sensor which observes a ring surface with a non-zero and non-right angle, in transmission.

[0106] [Fig. 13] Figure 13 schematically shows a control system comprising an image sensor which observes a ring surface from above, in reflection.

[0107] [Fig. 14A] Figure 14A is an image of a ring surface exhibiting a burr defect.

[0108] [Fig. 14B] Figure 14B is an image of a ring surface exhibiting an unrendered ring defect.

[0109] [Fig. 14C] Figure 14C is an image of a ring surface with a glaze defect.

[0110] [Fig. 15] Figure 15 is a scrolled image of a ring surface on which an unrendered ring defect is visible. Description of the embodiments

[0111] We will now describe the control of a glass container manufacturing facility. In particular, we will describe the detection of defects in the hot sector and the control systems that can be used to detect these defects. We will also describe the determination of parameters of a manufacturing system that are linked to these defects, and finally the issuing and execution of commands affecting these parameters.

[0112] Figures 1A and 1B schematically show an INS installation for manufacturing transparent or translucent glass containers 2. Figure 1A shows the installation in side view and Figure 1B in top view. This installation is shown in part since only the hot sector is visible in these figures.

[0113] An SF manufacturing system is shown in the figure. This system manufactures generally transparent glass containers of all types known per se. At the output of the SF manufacturing system, containers 2 such as example glass bottles or flasks, have a high temperature typically between 300°C and 600°C.

[0114] In a known manner, the containers 2 which have just been formed by the manufacturing system SF are placed successively on an output conveyor 5 to form a row of containers. The containers 2 are transported in a row by the conveyor 5 in a direction of travel F in order to convey them successively to different treatment stations and in particular an annealing arch 6, upstream of which is placed a surface treatment hood 7 generally constituting the first of the treatment stations after forming.

[0115] The SF manufacturing system comprises several separate forming sections 12, each comprising at least one roughing mold 13 and at least one finishing mold 14. The SF system comprises a source 16 of malleable glass, therefore hot glass, and a distributor 17 of glass gobs which distributes, by gravity, gobs of malleable glass 18 to each roughing mold 13. In a known manner, the source 16 of malleable glass is a reservoir supplied with molten glass, at the bottom of which is a bowl comprising one to four circular openings. A rotating tube whose height is regulated controls the flow of glass above the bowl, and a system of one to four plungers animated by a back and forth movement, extrudes the glass through the one to four openings of the bowl in order to deliver by gravity, the malleable glass in the form of one to four parallel strings.The malleable glass strings are definitively separated into independent drops by a scissors system 19 arranged at the outlet of the hot glass source 16 and which is actuated at regular intervals to cut the malleable glass coming from the source 16 into sections (this scissors system can be controlled automatically, for example to modify a weight or a parison shape).

[0116] For systems with several (up to four) molding cavities per section, possibly several sections are delivered in parallel and simultaneously. In this description, a parison 18 is called a drop or extruded section of malleable glass as cut by the scissors system 19. In English, the parison is, at this stage of a forming process, called a "gob". The malleable glass, at the level of cutting by the scissors system 19, generally has a temperature above 900°C, for example between 1100 and 1300°C. This parison is generally a solid cylinder of malleable glass having a volume and a length defined by the adjustment of the source 16 cooperating with the cutting of the scissors system 19. Indeed, the diameter of the parisons is defined by that of the openings of the bowl. The flow rate is controlled both by the height of the tube which acts on the overall flow rate and by the movements of the one to four plunger(s), which makes it possible to vary the flow rate separately for each opening of the bowl. The time interval between two actuations of the scissors system 19 determines the length of the parison.To summarize, the length, weight and volume of each parison are determined by the parameters of the source 16 (the tube and the plungers) and the scissors system 19. The source 16 of malleable glass is arranged above the blank molds 13 to allow the distribution by gravity of the parisons which are loaded through openings 22 arranged in the upper faces of the blank molds 13.

[0117] The distributor 17 extends along several branches between the source 16 of hot glass and the roughing molds 13 of each of the forming sections. Generally, the source 16 of hot glass, via the scissors system 19, simultaneously delivers as many parisons as there are roughing molds (respectively finishing molds) in a forming section. It is therefore understood that the forming sections are supplied with parisons successively one after the other.

[0118] The distributor 17 therefore collects the parisons cut by the scissors system 19 and leads them to each of the roughing molds 13 of each of the forming sections 12 according to a corresponding loading path. The loading paths for the different roughing molds 13 comprise common portions and specific portions. A specific portion is a portion of the loading path corresponding to a mold roughing mold 13 which is followed only by the parisons which are directed by the distributor towards this roughing mold.

[0119] The distributor 17 therefore comprises switching means which is a type of pivoting chute or group of chutes, then guiding the parisons comprising chutes and deflectors at the end of travel, above the roughing molds. In particular, the position of the deflectors relative to the associated roughing molds partly determines the position and orientation of the loading of each parison into said roughing molds. In the distributor, the chutes, the deflectors and switches determine the loading trajectory of the parisons.

[0120] Glass container manufacturing systems use different processes combining successive filling, pressing and / or blowing steps. For clarity of description, the example is taken from the forming of containers using the known BB, PB, or NNPB processes.

[0121] In container manufacturing systems, each forming section 12 may comprise several molds, for example two molds, one of which is a roughing mold 13 and the other is a finishing mold 14. Each section 12 may comprise a set of roughing molds and a set of associated finishing molds. It is understood in this case that a given parison is guided by the distributor 17 towards a roughing mold, for example a roughing mold 13 of the forming section where the parison undergoes a first forming operation, called piercing, carried out by blowing compressed air or by penetration of a punch. A transfer system (not shown) is then capable of removing the parison having undergone the first forming operation, namely the roughing, from the roughing mold 13 to take it to a finishing mold 14 where the roughing can undergo at least a second forming operation, the last operation called finishing.Generally, each roughing or finishing mold of a forming section comprises two half-molds respectively (half-molds 13a and 13b are visible in Figure 1B) which are movable relative to each other in a direction perpendicular to a parting plane by which the two half-molds are in contact in one position. closed. In the example shown, the joint plane extends along the vertical direction Z and the transverse direction X.

[0122] A section 12 may comprise a single finishing mold 14 receiving a blank from a single blank mold 13. However, as mentioned above, each of the different forming sections 12 may comprise at least two separate finishing molds 14 and as many blank molds 13. The Figures illustrate the case of four forming sections 12 offset in a longitudinal direction Y perpendicular to the transverse direction X. According to this example, each forming section 12 comprises three blank molds 13 respectively front, central and rear, (or external, central and internal) each associated with a finishing mold 14 respectively front, central and rear, that is to say, each receiving the blank from a blank mold 13. In the example illustrated, the different blank molds 13 and respectively the finishing molds 14 of the same section are offset relative to each other in a transverse direction X.In the example illustrated, the finishing molds 14 of the same section are of identical shape, therefore generally intended to form identical containers, but different shapes and weights could be provided.

[0123] It should be noted that each finishing mold 14 is identified in the forming installation relative to the other finishing molds 14. Similarly, each roughing mold 13 is identified in the manufacturing system. It is thus possible to identify the forming section 12, the roughing mold 13 and the finishing mold 14 from which each container 2 comes.

[0124] In a glass container manufacturing system, each blank mold location 13 of each section carries, according to various possible conventions, an identifier, for example a number or a letter. These locations can be referred to as forming cavities, which are identified by a forming cavity number.

[0125] Furthermore, the finishing molds can carry an imprint in order to print in relief on the containers 2, the mold number for example between 1 and 99 or between 1 and 128, etc. A correspondence table between the forming cavity numbers and the mold numbers is permanently available for the operators or the installation's information system. In some installations, a laser marker is used as described in patent EP 2 114 840 B1 to print a code on each container, which is still hot, immediately after it has been formed, indicating the mold number or the forming cavity number.

[0126] Thus, the containers generally carry either in a coded manner (bar code, dot code, Datamatrix code) or in an alphanumeric manner, the indication of the mold number or the forming cavity number. To reread, for example in the cold sector, these mold or forming cavity numbers carried by the containers, there are various optical reading systems for the production lines such as described in EP 1 010 126 or EP 2 297 672 or EP 2 992 315.

[0127] Thus, in the present description, it is understood that identifying the finishing mold from which a sample container comes therefore amounts to knowing either the forming cavity number or the mold number. It is understood that the identification of the finishing mold makes it possible to directly identify the associated roughing mold supplying the roughing.

[0128] In forming installations, the control and synchronization of the operations of forming parisons, scissor cutting, movements of the molds, movements of the punches, blowing, transfers, etc. are carried out by means of a control device 200 in the general sense, making it possible to control the various mechanisms necessary for the operation of the installation for the implementation of the container forming process.

[0129] Here, in the hot sector, either before passing through the annealing arch 6 and even here before passing through the surface treatment hood 7, we use a container control system SC.

[0130] The container control system is configured to control containers that are moving along conveyor lane 5.

[0131] More specifically, it is at least configured to obtain at least one image comprising at least the surface of the ring of the container, which is being checked, and to detect, by processing said at least one image, at least at least one container ring defect included in a list of container ring defects including at least the flash defect and the unrendered ring defect (and possibly the glaze defect).

[0132] For this purpose, the control system SC may comprise elements similar to those of a computer system, and it may in particular comprise a processor 110 and a non-volatile memory 111 in which computer program instructions executable by the processor 110 are recorded.

[0133] The processor can in particular control an image sensor 101 of the control system, and a light source 102 of the control system.

[0134] The control system here comprises a communication module 112, which is in communication via a communication interface INF with the manufacturing system SF. The interface INF may be a wired or wireless interface, and it may in particular be accessible from a control cabin of the installation where a human-machine interface may make it possible to consult the information communicated via the interface INF.

[0135] Within the manufacturing system, a control device 200 is implemented. The control system has a structure analogous to that of a computer and here it comprises a processor 201 and a non-volatile memory 202 (in any form, for example arranged within the same semiconductor chip, arranged within separate chips, etc.) in which computer program instructions executable by the processor 201 can be stored. The control device 200 further comprises a communication module 203, which also uses the communication interface INF.

[0136] Preferably, the control system may be configured to identify a parameter of the manufacturing system linked to a defect that has been detected by the control system.

[0137] For this purpose, a correspondence table between faults and parameters can be used.

[0138] Alternatively, closed-loop control of the manufacturing system can be used.

[0139] Alternatively, one can use an expert system configured to output parameters based on identified faults.

[0140] Typically, when a fault is detected and a parameter is identified by means of a correspondence table, different actions can be implemented. For example, information relating to the fault and the parameter can be stored in the memory 111. Also, commands can be issued via the communication interface.

[0141] A first possible command may be a display command on a display device of the installation. Here, it is the manufacturing system SF which comprises a display device 300 visible in FIG. 1A, which receives the command here via the communication module 203. Thus, an operator can see on the display device that a parameter must be modified and / or that a fault has been detected (detection of the fault can be implemented even without identification of a parameter). Alternatively, the display device can be arranged within a control cabin of an operator of the installation.

[0142] A second possible command may be a command to modify the identified parameter, this command being intended for the manufacturing system. The manufacturing system receives this command by means of the communication module 203, then it can execute it using its processor 201 and possibly other computer program instructions stored in the non-volatile memory 202.

[0143] Here, the SC control system acts on the manufacturing system through a feedback loop.

[0144] In fact, on the one hand it was observed by the inventors that burr and unrendered ring defects are detectable in the hot sector, and on the other hand it was observed that these defects are linked to parameters of the manufacturing system.

[0145] In the following, we explain how a correspondence table can be developed (but in the following, the observations also apply to other solutions for delivering parameters based on defects).

[0146] For a burr defect (for example a burr of the internal ring, for example a sharp edge or "wire-edge" in English or strongly marked or "overpress" in English), the parameter of the manufacturing system can be a parameter of the glass parison loaded into the blank mold to form the container and / or a positioning parameter of a punch, and / or a parameter of piercing or pressing of said blank implemented prior to blowing. For example, a parameter of the glass parison can be a weight. A positioning parameter of a punch can be a position at a given time, for example at the start or end of rise, of the punch relative to the blank mold. A piercing or pressing parameter can be a time parameter of operation of the punch used for piercing or pressing (for example at what time it moves in a direction, for what duration it moves, etc.).

[0147] A modification order for a burr defect might be an order to lower the weight of the glass parison.

[0148] It may be noted that the detection of the defect and / or the development of the order may take into account a forming cavity or mold number. For example, if a flash defect occurs repeatedly for a given molding cavity, it may be appropriate to modify a punch parameter so that it rises earlier (for example by a command). On the other hand, if a flash defect occurs repeatedly for several molding cavities, it may be appropriate to reduce the weight of the glass parison by controlling a parison-producing device. It is therefore possible, automatically, by analyzing the statistical distribution of the defects on the cavities or sections, to trigger an automatic or manual order modification in the case where the system displays the defects.

[0149] For an unrendered ring defect, the manufacturing system parameter may be a parameter of the glass parison loaded into the blank mold to form the container, or a blank parison forming parameter. For example, a glass parison parameter may be a weight, or a glass parison temperature, or a parison shape parameter. The blank parison forming parameter may be a compression time, or a compression pressure (especially for a BB process).

[0150] A modification command for an unrendered ring defect may be a command to increase a parison weight or parison temperature (especially if the defect is detected for several different forming cavities), a command to increase a compression or pressing time and / or power.

[0151] The invention also relates to glaze defects that can be detected by obtaining an image. A parameter related to the glaze defect may be a cooling parameter or a contact time parameter with the glass of a molded part (rough mold or finishing mold). A cooling parameter is a parameter that controls the circulation of a fluid in the molds. Solenoid valves that control the circulation of a fluid can be acted upon, by modifying the start time of circulation of the fluid, or the end time of circulation of the fluid, and this in a differentiated manner to act on several circuits. It is also possible to act on the pressure of the fluid or its own cooling. Other measurements can be taken into account in this control, such as a measurement of the temperature or the temperature difference of the fluid entering or leaving the mold.Of course, displaying the defect can lead an operator to check mechanical defects in the section such as the general condition of the casting, and possibly proceed with a replacement or repair.

[0152] A modification command for a glaze defect may be a command to modify this cooling and / or contact time.

[0153] Figure 2 shows in side view a ring B of a container 2. The ring in this figure is a ring having a thread, and it should be noted that the invention is not limited to threaded rings but applies to other types rings, for example rings on which capsules are mounted by crimping.

[0154] The ring of Figure 2 has a ring surface S, which corresponds to the upper surface of the container. In fact, the surface S is a substantially horizontal surface when the container is in its vertical position with the opening upwards, which is conventionally the conveying position of containers which have just been formed and which are conveyed in the hot sector.

[0155] More precisely, the ring also has a counter ring CB and, due to the forming of the ring in a blank mold in several parts, we can observe both vertical mold joints JV and a horizontal mold joint JH at the base of the surface of the ring S.

[0156] As indicated above, the invention particularly aims at detecting defects in the ring surface S which is visible in Figure 2.

[0157] Figure 3 shows a sectional view of the wall of a ring of a container. Here, the ring is a ring intended to receive a cap by crimping. At the inner side of the wall and at the edge of the surface of the ring S, a defect of the burr type BV is shown. The defect of the burr type BV can be called in English, "over-press". In a manner known per se, the application by crimping of a cap on the container of Figure 3 could lead to breakage and shards of glass present inside the container.

[0158] Although not shown because it is well known to those skilled in the art, the defect of an unreturned ring leads to the appearance of a hollow on the ring surface.

[0159] In the present description, the detection of defects is done by obtaining an image of at least the surface of the ring of a container. Figure 4 shows the arrangement of an image sensor 101 for a control system SC according to an example. The image sensor is here a matrix camera capable of obtaining images by capturing radiation in the visible, and / or UV and / or infrared range. The general observation axis of the camera denoted D in the figure has a negative downward angle denoted a. The sensor The image sensor of Figure 4 is configured to obtain high-angle images. This configuration is well suited for detecting the defects mentioned above.

[0160] The image sensor 101 of Figure 4 may be used in combination with a light source configured to illuminate the ring surface of a container that is conveyed after its manufacture in the hot sector. Preferably, such a light source is extensive, having a light emitting surface width of at least 5 cm, preferably at least 8 cm, and more preferably at least 12 cm. Preferably, such a light source has a light emitting surface greater than 20 cm 2 , preferably greater than 30 cm 2 , and preferably even greater than 40 cm 2 . Also preferably, the light source is provided with a diffuser, for example arranged in front of a plurality of elementary sources of the light source such as light-emitting diodes. The light source is thus preferably diffuse, and is therefore not a laser source.

[0161] Figure 5A shows an example in which the image sensor 101 observes from above (like the sensor described with reference to figure 4) a container 2 which circulates on a conveyor track 5 from left to right in the figure. A light source 102 is used (it is higher than the surface of the ring, so that rays which it emits are reflected towards the image sensor). Here, the light source 102 has a width L1 to facilitate the acquisition of an image on which the surface of the ring is well illuminated during its conveyance when obtaining an image of the surface of the ring by the image sensor 101. In fact, this image obtaining can be called obtaining in reflection, the light source 102 being configured to be at least partly above the surface of the ring and this is also the case for the image sensor 101 which observes the reflection of the light emitted by the source 102 on the surface of the ring.

[0162] Figure 5B shows an alternative configuration in which two image sensors 101A and 101B are used, spaced along the conveyor track but both oriented towards the same point. In fact, their observation axes respective are in two different azimuths, and their respective observation axes intersect at a point at the conveyor track 5. Thus when the container 2 is at the position in which it is shown in the figure, the image sensors 101A and 101B can be used simultaneously.

[0163] Here we use a 102' light source of width L2, with L2 here greater than L1. This configuration makes it easier to obtain images of the surface of the ring.

[0164] Figure 5C shows yet another configuration in which two image sensors 10 IA and 101 B are used which are not configured to be used simultaneously for the same container 2. In the figure, the image sensor, 101B is shown blackened, as is the light source 102B with which it cooperates. This blackening indicates an unused state of the image sensor 101B and the light source 102B, while the representation filled with white corresponds to a used and operating state (for example at the time the image sensor is obtaining the image). The image sensor 10 IA and the light source 102A are used and are operating in Figure 5C. To obtain a second image at a different time (not visible in Figure 5C) and very close to the time of operation of the camera 10 IA with the source 102A, the image sensor 101B and the source 102B are activated.The two light sources 102A and 102B are aligned and each have a width L3 corresponding for example to the width L1 described with reference to FIG. 5A. They can be part of the same lighting system (typically the same lighting panel) having an illuminating surface activated by independent regions or zones. Thus, the light sources 102A and 102B can overlap. In the example of FIG. 5C and in the example of FIG. 5B, two images of the surface of the ring are obtained.

[0165] As an example, when several images are obtained, defect detection can be implemented for each of the images. For example, a defect may only be detected on one of the several images, for example if this defect is easier depending on the observation axis of the image sensor used. Thus, the use of several images improves defect detection. This example is well suited for detecting burrs that may be located inside the ring and therefore more easily observable at a given azimuth.

[0166] Figure 6 shows, as a function of time (from left to right), an example in which two image sensors 101A and 101B are still used, and in which a light source 102' of the type illustrated in Figure 5B is used. The light source 102' is a portion (here the upper portion) of an overall light source, further comprising a lower portion 103 used in a conventional manner with the image sensors 104A and 104B. The image sensors 104A and 104B, with the light source 103, operate in the manner described in document FR 3056297, that is to say that they can take an image of the container over its entire height (they are oriented according to two different azimuths).

[0167] In this figure, we still use the convention that the blackened elements are not in operation and not used at a given time. In a first phase PHI, we use the image sensor 10 IA, and we obtain a first image of the surface of the ring using the light source 102' (the other image sensors and the other light source are not used).

[0168] In a second phase PH2, the image sensor 104A is used with the light source 103, possibly also with the light source 102' (the two possible configurations are shown in the figure).

[0169] In a third phase PH3, the image sensor 101B is used with the light source 102', the light source 103 is not used and is not in operation, to allow a good image of the surface of the ring to be obtained.

[0170] In a fourth phase PH4, the image sensor 104B is used with the light source 103, and possibly the light source 102'.

[0171] It can be noted that in figures 5 and 6, the elements referenced A are located upstream of the conveyance and the elements referenced B are located downstream of the conveyance.

[0172] Figure 7 shows for yet another configuration several phases of the use of a configuration corresponding to that shown in Figure 5C.

[0173] Here, an image sensor 101A, an image sensor 101B, and, as explained with reference to FIG. 6 for another configuration, image sensors 104A and 104B are used.

[0174] In a first phase PHI', the image sensor 101A and the light source 102A are used, the other elements are neither used nor in operation. In a second phase PH2', only the image sensor 101B and the light source 102B are used.

[0175] In a third phase PH3, only the image sensor 104A and the light source 103 are used to obtain an image of the container in its entire height.

[0176] For illustrative purposes, the light sources described with reference to Figures 5A, 5B, 5C, 6, and 7 may all be integrated into a single panel-type lighting system. This panel may include a light-emitting surface that may be activated or controlled by zones or regions, and these zones or regions may be separate or overlapping. In the example shown in Figure 6, region 103 and region 102' overlap completely (PH2 and PH4 at the bottom) or partially (PH2 and PH4 at the top).

[0177] Also, for information purposes, the light sources described with reference to Figures 5A, 5B, 5C, 6, and 7 may all be separate and distinct. The flat emitting surfaces of these sources (which may be individual light panels) are oriented parallel to the movement of the containers (for example along the X axis) as illustrated in Figure 9A described below. Alternatively, they may be oriented non-parallel to the movement of the containers, for example when their emitting surfaces are oriented orthogonally to the observation axes of the image sensors with which they operate as illustrated in Figure 9B described below. Alternatively, the light sources are non-planar, for example curved as described in patent FR3127574B1.

[0178] Figure 8A shows in top view the arrangement of an image sensor 101 (such as that of Figure 5A) with a container and with a light source 102, when obtaining an image of the surface S of the ring. The image sensor 101 is arranged so that its observation axis is aligned along an azimuth AZ relative to the movement of the containers when obtaining the image of the surface of the ring. Preferably, the observation axis of the image sensor 101 is aligned along an azimuth greater than 10°, relative to the movement direction F of the containers 2 on the conveyor 5, that is to say that the projection of the observation axis in a horizontal plane forms an angle of at least 10° relative to the movement direction F of the containers 2 on the conveyor 5, and more preferably still greater than 20°.Preferably, the image sensor 101 is placed on one side of a vertical plane containing the direction of travel F of the containers 2 on the conveyor 5, and the light source 102 illuminating the container imaged by the image sensor 101 is placed on the other side of this vertical plane. Thus, the light source 102 is arranged on the other side of the surface S of the container ring relative to its image sensor 101.

[0179] Figure 8B shows another configuration in which an image sensor 101A and an image sensor 101B are used cooperating respectively with a light source 102A and a light source 102B. The image sensors 101A and 101B have respective observation axes DA and DB aligned according to azimuths which are different and these axes meet the surface S of the ring of the container for the same position of the container 2 on the conveyor track 5. The light sources 102A and 102B are arranged on the other side of the surface S of the ring of the container with respect to their respective image sensors.

[0180] Figure 9A shows a configuration in which four image sensors referenced, 10 IA, 101 B, 10 IC, and 101 D are used cooperating respectively with light sources, 102A, 102B, 102C, and 102D.

[0181] The observation axes of the four image sensors 101A, 101B, 101C, and 101D, denoted DA, DB, DC, and DD respectively, are distributed around the container traveling on the conveyor track. More precisely, the axes DA and DB form a 90° angle between them and meet at the surface of the container ring of a container 2 traveling on the conveyor track 5 in the direction of travel (from left to right in the figure). Also, the image sensors 101A and 101B are arranged on the same side of the conveyor track 5. On the other side of the conveyor track 5, the image sensors 101C and 101D are arranged so that their observation axes DC and DD form a 90° angle between them and meet at the surface of the container ring of a container 2 which circulates on the conveyor track, at a location which differs from that of the axes DA and DB (these two locations are spaced along the conveyor track 5).

[0182] Figure 9B shows another configuration similar to that of Figure 9A but which differs therefrom in that said two locations are further apart along the conveyor track 5.

[0183] In the following and more specifically in the example illustrated in figures 10 and 11, the rays coming from the light source are reflected by the surface of the ring towards the image sensor. It should be noted that this reflection is based on the "specular reflection" of the light incident on the ring surface. In the absence of a defect, the surface of the glass is specular and therefore it reflects the light according to the Snell-Descartes laws. On the other hand, we have a reflection which can be diffuse in the presence of a defect.

[0184] Figures 10 and 11 show the path of incident light rays RI, R2, R3, and R4 initially originating from a light source 102, and which are reflected on the surface S of the ring, of a container 2 to respectively form the reflected rays RI', R2', R3', and R4' directed towards the pupil of an objective of the image sensor 101 oriented to observe ring surfaces of a container in the same manner as the image sensor 101 described with reference to Figure 4.

[0185] More precisely, in Figure 10, two incident rays RI and R2 are shown, which reach the ring surface S with incident angles noted respectively aRl and aR2, then which are reflected in two reflected rays RI' and R2', with reflected angles respectively aRl and aR2 equal to the incident angles. For reasons of simplicity, the ring surface is here assumed to have the shape of a plane ring, its normal being vertical and therefore parallel to a vertical axis Z passing through its center. The reflection plane of the rays RI and R2 is therefore assumed to contain this vertical axis Z. The height H of the source 102 (measured in the vertical direction in the figure, i.e. along Z) is such that there are reflected rays RI' and R2' which enter the pupil of the objective to form the image of the ring. In fact, the height H is such that the rays RI and R2 are emitted to obtain the reflected rays RI' and R2', and moreover the field C of the camera is represented here by its extent in the horizontal plane of the ring covers at least the surface S.In order to accept positioning tolerances of the articles, the field C of the camera and the height H of the source are chosen to be sufficiently large.

[0186] In Figure 11, two incident rays R3 and R4 are shown in top view, which reach the ring surface S with incident angles respectively noted oR3 and oR4 (but not visible in the figure in top view), then are reflected in two reflected rays R3' and R4', with reflected angles respectively oR3 and oR4 equal to said incident angles, not shown. For reasons of simplification, in the illustrated example, the ring surface is assumed to have a slightly frustoconical shape, its normal N being inclined relative to a vertical axis Z passing through its center. The width L of the source 102 is such that there are reflected rays R3' and R4' which enter the pupil of the objective to form the image of the ring, therefore in such a way as to emit the incident rays R3 and R4. Furthermore, the field C of the camera represented here by its limits in the horizontal plane of the ring (in the form of a trapezoid in the figure), covers at least the surface S.In order to accommodate item positioning tolerances, the camera field C and the source width L are chosen to be sufficiently large.

[0187] The image sensor is positioned and the light source is positioned and sized in width and height based on the reverse path of the rays, according to the following steps: - position in the inspection station, a ring surface of a given diameter, the ring diameter being chosen as being the largest of a range of containers to be inspected, for example a pot ring of diameter 90mm, - position the image sensor laterally at the workstation corresponding to the control system and therefore outside the conveyance of the containers, on one side opposite the source, at a distance from the ring surface chosen to allow free passage of the containers without risk of collision with the image sensor, - the focal length of an image sensor lens is chosen (taking into account the sensor) to cover a field of observation containing all the successive rings which will scroll successively, - by the reverse path of light, determine the reflected rays RI', R2', R3', R4', etc. leaving the pupil of the objective to reach the ring surface, - determine the direction of the incident rays RI, R2, R3, R4, etc. by the rules of specular reflection, - position a light source, for example a flat one, at a given distance from the container, a distance chosen to allow free passage of the containers without risk of collision with the source - size the light source, for example its illuminating surface, so that it emits light at least for the incident rays RI, R2, R3, R4, etc.

[0188] The system can be designed and sized by taking into account random deviations in the position of the containers from an average or theoretical rectilinear trajectory on the conveyor to determine the necessary field of observation, then by deduction position and size the source as explained above. Ideally, the source should be able to illuminate a portion of the horizontal plane covering the field of view of the camera.

[0189] Figure 12 shows yet another configuration in which the ring B and its ring surface S are observed in transmission. The image sensor 101 of this figure has a position similar to that of the sensor 101 of Figure 4, it observes the surface of the ring in a plunging view. The dimensioning of the source is again done so as to allow the inspection of a range of containers and with a sufficient observation field and an illuminated field to allow for variations in the positioning of the containers at the time of image capture.

[0190] As can be seen in Figure 12, the rays from the light source 102 pass through the wall of the ring B of the container at one or more locations. Also, for reasons of simplicity, the refraction of the three light rays shown in the figure when they pass through the glass walls of the container has not been shown. These three light rays, coming from the light source 102, reach the surface of the ring by transmission through the wall, with a non-zero and non-straight incident angle (with a vertical plane like the incident angles of the examples in Figures 10 and 11) and are transmitted to the image sensor 101.

[0191] Another configuration can be provided in which an image sensor 101 is used which observes the ring surface S of a ring B from above. For this purpose, a light source could be used which surrounds the observation axis of an image sensor, the light source having an annular shape, or other shapes such as for example a truncated cone or a dome. However, it should be noted that unlike the previous solutions, such a system known in the cold sector requires adaptations to operate in the hot sector. These systems are in fact not very tolerant of the centering defect of the containers, not aligned during scrolling. In a first variant, the containers can be aligned, but it is preferable not to touch the hot containers. It is therefore preferable to modify, in a variant, the optical system to make it tolerant of the lateral position deviations (along Y) of the containers on the conveyor.

[0192] According to a preferred variant for observation from above, the coaxial lighting system and telecentric observation optics can therefore be provided, shown in Figure 13. For this purpose, a light source 102 is used which is a diffuse surface. This light source is for example rectangular with a width noted L or circular with a diameter noted L). A beam splitter such as a semi-reflecting beam splitter LS is also used. The incident light is substantially parallel to the vertical optical axis of the objective O due to the use of a field lens LC. The incident light reflected by the plate is therefore in turn substantially vertically parallel. Of course, the semi-reflecting plate can be replaced by a beam splitter cube.

[0193] The rays reflected on at least a horizontal portion of the ring surface are also substantially vertically parallel, and the image is obtained with a telecentric observation optic composed of the objective O and the field lens LC centered on the optical axis of the objective O, the focus of which substantially coincides with the entrance pupil of the objective O of the image sensor 101.

[0194] Thus, the observation of the ring surface is not or very little modified in terms of lighting and apparent shape, when the ring surface is shifted laterally, since at any point in the field C, the incidence of the light coming from the source 102 and the direction of the rays reflected through the field lens LC then the objective O is the same for all the rays. It is noted that the optical paths of the incident rays from the source to the ring surface and the optical paths of the rays reflected to make the image, from the ring surface to the pupil of the objective O are of the same length and correspond to the focal length of the field lens LC. The distance between the focus of the field lens LC and the optical center can be adapted between plus or minus 1 cm, in order to give the observation a strictly telecentric or slightly pericentric character.

[0195] Furthermore, for such assemblies, the optical system can be equipped with temperature management means, which may include, for example, a protection filter FP against radiation from hot containers, and / or a containment enclosure E, cooled and / or ventilated (as shown in the figure or a ventilation system bears the reference VT) and / or insulated thermally. Finally, anti-fouling devices can be provided, taking into account the environment, such as a laminar flow FL of dry, clean air between the containers and the bottom of the enclosure.

[0196] Figure 14A shows an image obtained by a device such as that of Figure 10, in which a DEFI burr defect can be observed, corresponding to a reflection at the edge of the interior of the ring.

[0197] Figure 14B shows another ring surface exhibiting an unrendered ring defect DEF2, corresponding to a hollow in the ring surface, at one location.

[0198] Figure 14C shows a DEF3 glaze defect, corresponding to a blackened line in the figure.

[0199] In view of Figures 14A, 14B, and 14C, the person skilled in the art is able to implement image processing algorithms to detect these three defects.

[0200] Figure 15 shows how ring surface images such as those described in Figures 14A, 14B, and 14C can be processed, here by implementing an unrolling of the ring surface on which an unrendered ring defect is measured (which corresponds to an absence of an unrolled ring surface in the figure).

[0201] Different processing techniques can be used, including elliptical shape detection, unrolling of the ring surface as shown in figure 15, surface detection (possibly with adjustment, etc.).

[0202] Some non-limiting algorithm variants that can be used for defect detection are described below. They show that burr, unrendered ring and glaze defects can be detected using the methods described above, in particular the variant in Figures 10 and 11, which is well suited to the hot sector.

[0203] We note that in image 14A, the burr defect is manifested by a light signal inside the ring surface. The algorithm then consists of for example to determine the position of the surface of rings (in the image or its unfolding) and to look for the presence of arcs of luminous circles nearby.

[0204] We note that in image 14B, the unrendered ring defect manifests itself by a transverse constriction of the ring surface ring, or even its cut. The algorithm then consists, for example, of determining whether the radial thickness of the ring surface image is constant or varies slowly by traversing it circularly, or by traversing its unrolling.

[0205] The appearance of smear and non-rendering defects characterizes them in the image, which ensures that they can be identified.

[0206] Of course, the above algorithms are not limiting and the invention can be carried out by implementing in the processing unit any method of image analysis and recognition, including automated learning methods, in particular supervised methods of the Deep Learning type, with image segmentation or without image segmentation of the convolutional neural network type.

[0207] The invention is not limited to the examples described and shown because various modifications can be made thereto without departing from its scope.

Claims

Claims 1. Method for controlling an installation (INS) for manufacturing glass containers (2) having a ring, implemented by a control system, the containers having been manufactured by a manufacturing system in which for each container, a parison (18) is loaded into a blank mold (13) to be transformed into a blank, said container blank being placed in a finishing mold (14), and in which air is blown through the ring of the container blank to obtain the container, the method comprising, downstream of the container manufacturing system and upstream of the passage of the containers through an annealing arch (6) of the installation, for at least one container moving on a conveyor: - obtaining at least one two-dimensional image by acquisition by an image sensor (101) which is a matrix camera, using at least one light source, the image sensor (101) being arranged so that its observation axis is aligned along an azimuth relative to the movement of the containers on the conveyor, said at least one image comprising at least the surface of the entire ring of the container, the surface of the ring being visible in the image in the form of a ring or an ellipse, the image resulting from the light emerging from the entire surface of the ring, - a detection, by processing said at least one image, of at least one container ring defect, the detection comprising the identification of the type of defect from a list of container ring defects comprising at least the burr defect and the unrendered ring defect.

2. Method according to claim 1, further comprising an identification of at least one parameter of the manufacturing system linked to said at least one detected defect.

3. Method according to claim 1 or 2, wherein the detection of said fault and / or the identification of said at least one identified parameter triggers the display of said detected fault and / or of said at least one identified parameter on a display device of the installation.

4. Method according to claim 2 or 3, wherein the identification of said at least one identified parameter triggers the emission of a command to modify said at least one parameter, the command being intended for the manufacturing system.

5. A method according to any one of claims 1 to 4, comprising obtaining a first image by said image sensor using said light source, and obtaining a second image by another image sensor using another light source, wherein when obtaining the first image, an observation axis of said image sensor is aligned along a first azimuth relative to a direction of movement of the containers on the conveyor, and when obtaining the second image, an observation axis of the other image sensor is aligned along a second azimuth.

6. Method according to claim 4 or 5, wherein for a container, when obtaining said at least one image, the light source and said image sensor are arranged at least above the ring and on either side of the ring, so that at any point on the surface of the ring, there is a ray coming from the light source which reaches the point with a non-zero and non-right incident angle, and which is reflected in a region of a pupil of the image sensor.

7. Method according to claim 4 or 5, wherein for a container, when obtaining said at least one image, said image sensor is arranged at least above the ring, the image sensor and the light source being arranged on either side of the ring, so that at any point on the surface of the ring, there is a ray coming from the light source which reaches the point with a non-zero and non-right incident angle after having crossed the container, and which is transmitted into a region of a pupil of the image sensor.

8. Method according to one of claims 4 to 7, in which the light source and / or the other light source are sources included in sources adapted to be able to illuminate a container over at least its entire height.

9. Method according to any one of claims 1 to 8, in which for each container of a plurality of containers, the following are implemented: - obtaining at least one two-dimensional image by acquisition by an image sensor which is a matrix camera, using at least one light source, the image sensor (101) being arranged so that its observation axis is aligned along an azimuth relative to the movement of the containers on the conveyor, said at least one image comprising at least the surface of the entire ring of the container, the surface of the ring being visible in the image in the form of a ring or an ellipse, the image resulting from the light emerging from the entire surface of the ring, - a detection, by processing said at least one image, of at least one ring defect of the container, the detection comprising the identification of the type of defect from the list of ring defects, - an identification of at least one parameter of the manufacturing system linked to said at least one defect detected for a given number of containers of the plurality of containers.

10. Method according to any one of claims 1 to 9, in which, for a container, a section number and / or forming cavity in which the container has been formed is obtained.

11. The method of claim 2, wherein the parameter of the manufacturing system related to the flash defect is a parameter of the glass parison loaded into the blank mold to form the container and / or a parameter positioning of a punch, and / or a drilling or pressing parameter of said blank implemented prior to blowing.

12. A method according to any one of claims 2 or 11, wherein the plurality of defects includes the unrendered ring defect, and wherein the manufacturing system parameter related to the unrendered ring defect is a parameter of the glass parison loaded into the blank mold to form the container, or a forming parameter of the parison into a blank.

13. A method according to any one of claims 2, 11, or 12, wherein the list of defects further comprises the glaze defect, and wherein the manufacturing system parameter related to the glaze defect is a cooling or glass contact time parameter of a molded part.

14. Method of manufacturing glass containers comprising an implementation of the control method according to any one of claims 1 to 13.

15. Method according to claim 14, wherein the control method is according to at least claim 3, and wherein the display device of the installation is included in the manufacturing system or in a control cabin of the installation.

16. Method according to claim 14 or 15, wherein the control method is according to at least claim 4, wherein the manufacturing system receives said command to modify said at least one parameter, and executes said command to modify said at least one parameter.

17. Control system for a plant for manufacturing glass containers having a finish, the containers having been manufactured by a system manufacturing system in which for each container, a parison is loaded into a blank mold to be transformed into a blank, said container blank being placed in a finishing mold, and in which air is blown through the ring of the container blank to obtain the container, the control system being arranged downstream of the container manufacturing system and upstream of the passage of the containers through an annealing arch of the installation, and being configured to implement: - obtaining at least one two-dimensional image by acquisition by an image sensor which is a matrix camera, using at least one light source, the image sensor (101) being arranged so that its observation axis is aligned along an azimuth relative to the movement of the containers on the conveyor, said at least one image comprising at least the surface of the entire ring of the container, the surface of the ring being visible in the image in the form of a ring or an ellipse, the image resulting from the light emerging from the entire surface of the ring, - a detection, by processing said at least one image, of at least one container ring defect, the detection comprising the identification of the type of defect from a list of container ring defects comprising at least the burr defect and the unrendered ring defect.

Images