KR102632508B1 - Artificial intelligence-based tampon outer high-speed quality inspection system - Google Patents

Abstract

The present invention relates to an artificial intelligence-based high-speed quality inspection system for injection-molded products. The artificial intelligence-based high-speed quality inspection system for injection-molded products includes a supply unit for supplying injection-molded products; an alignment transfer unit that aligns and transfers the injection molded product supplied from the supply unit; an image acquisition unit configured to receive the injection-molded product transferred from the alignment transfer unit and transfer it to one side, while acquiring an image by imaging the transferred injection-molded product; Defective products are classified by analyzing them based on an artificial intelligence module using the image of the injection product obtained from the image acquisition unit, and a control analysis unit that controls each part is provided in the transport direction of the injection product transported by the image acquisition unit. It may be configured to include a defective product detection unit that detects defective products.

Description

Artificial intelligence-based tampon outer high-speed quality inspection system}

The present invention relates to an injection molded product quality inspection system, and in particular, to an artificial intelligence-based injection molded product high-speed quality inspection system that can perform quality inspection accurately and at high speed by performing machine vision based on an artificial intelligence module.

An injection product is an object made by melting plastic resin in a heated cylinder and then shooting the melted resin into a mold with a piston. An example of an injection product is a tampon outer.

Tampon outerwear is used in tampons, which are disposable absorbent products inserted into a woman's vaginal cavity to absorb body fluids discharged during a woman's menstrual period. This tampon outerwear is disclosed in Korean Patent No. 10-1609006, 'Tampon Applicator'. As shown, it is assembled into a tampon, which absorbs menstrual blood, and is used to insert the tampon into the body.

Here, the tampon outer is a product that is inserted into the body and must be hygienic, so quality inspection is essential and very important during and after manufacturing the tampon outer, but the development of a quality inspection system for the tampon outer is minimal.

To overcome this, conventional inspection systems incorporating a vision system are being used to inspect tampon outerwear, but due to their structure, the entire tampon outerwear cannot be inspected due to its cylindrical base, and detailed inspection cannot be performed. The false positive rate is high and the inspection speed is also slow, which has the disadvantage of slowing down the entire product production process, so a solution to this problem is required.

The present invention is proposed to solve the above-mentioned problems. It performs machine vision based on an artificial intelligence module, and further allows inspection of the entire surface of injection-molded products such as cylindrical-based tampon outer, while being very accurate. The purpose is to provide an artificial intelligence-based high-speed quality inspection system for injection molded products that can perform quality inspection at high speed.

An artificial intelligence-based high-speed quality inspection system for injection-molded products according to an embodiment of the present invention to solve the above problems includes a supply unit for supplying injection-molded products; an alignment transfer unit that aligns and transfers the injection molded product supplied from the supply unit; an image acquisition unit configured to receive the injection-molded product transferred from the alignment transfer unit and transfer it to one side, while acquiring an image by imaging the transferred injection-molded product; A control analysis unit that analyzes the image of the injection-molded product obtained from the image acquisition unit and analyzes it based on an artificial intelligence module and a vision module to classify defective products and controls each part, and the transfer direction of the injection-molded product transported by the image acquisition unit. It may include a defective product detection unit that is provided in and searches for classified defective products.

Here, the alignment transfer unit provides a receiving space therein to receive the injection molded product supplied from the supply unit, and forms a guide plate extending spirally from the lower end of the receiving space to the upper end, and rotates by power to guide the received injection molded product through the guide plate. It may include a rotary feeder that feeds while aligning in a line, and a straight feeder that forms a straight length from the end of the guide plate of the rotary feeder to the image acquisition unit and transfers the injection molded product transferred from the rotary feeder to the image acquisition unit.

In addition, the linear feeder is a constant-speed linear feeder that receives the injection material from the rotary feeder and transfers it at a constant speed, and receives the injection material transferred from the constant-speed linear feeder and transfers it to the image acquisition unit at a faster speed than the constant-speed linear feeder. It may include a high-speed linear feeder.

In addition, the linear feeder may further include an escape roller that separates the injection molded product transferred from the constant-speed linear feeder to the high-speed linear feeder.

In addition, the image acquisition unit includes a transfer conveyor in which a plurality of separation membranes are installed at regular intervals sufficient to accommodate one injection product, receives and stacks the injection product from the alignment transfer unit, and transfers it in one direction, and is installed on the upper side of the transfer conveyor. It may include a machine vision camera that acquires an image by capturing an installed and transported injection molding product, and transmits the obtained image to the control analysis unit.

In addition, the transfer conveyor is formed to have a width smaller than the length of the loaded injection-molded product so that at least one end of both ends of the injection-molded product protrudes outward, and has a device for rotating the injection-molded product on both sides or in the center between both sides in the transport direction of the injection-molded product. By providing a rotation generating means, the injection molded product transported by the transfer conveyor rotates and moves so that the machine vision camera can acquire an entire image of the injection molded product.

In addition, the image acquisition unit may further include a multi-stage stacking prevention means provided on the upper side of the transfer conveyor in front of the position of the machine vision camera to prevent multi-stage stacking of the transported injection molded product.

In addition, the image acquisition unit may further include a lighting device installed on both sides of the transfer conveyor where the machine vision camera is located to emit light to further increase the clarity of the image acquired by the machine vision camera.

In addition, the artificial intelligence module includes an image collection step of collecting images captured from the machine vision camera; If the image does not meet the learning and verification requirements in the image collection step, a control step of controlling the machine vision camera and lighting device to meet the learning and verification requirements; An image pre-processing step of pre-processing the images collected after the control step to remove light reflection using a vision module; A learning step in which defective positions are learned by comparing the image preprocessed in the image preprocessing step with a normal product image, and a defective product discrimination step in which defective positions learned through the learning step are determined, labeled, and defective products are distinguished, can be used to distinguish defective products. You can.

In addition, the defective product detection unit is provided on one side of the transport direction of the injection molded product transported by the image acquisition unit, and a punching means that receives a control signal from the control analysis unit and pushes the defective product from the provided position to the other side, and the punching means; It may include a recovery device provided at an opposing position to recover defective products pushed out by the punching means.

In addition, the artificial intelligence-based high-speed injection quality inspection system may further include a cloud server that receives classified defective product information from the control analysis unit, performs reanalysis of the defective product, and transmits the performed defective product reanalysis information to the control analysis unit. there is.

In addition, the artificial intelligence-based high-speed quality inspection system for injection-molded products may further include a non-defective product counter unit that is provided in the transfer direction of the injection-molded product transported by the image acquisition unit behind the defective product detection unit to count non-defective injection-molded products.

The artificial intelligence-based high-speed quality inspection system for injection molded products according to an embodiment of the present invention performs inspection using a machine vision module based on an artificial intelligence module, and further inspects using images captured while rotating a cylindrical-shaped injection molded product by 360°. By performing this, quality inspection can be performed very accurately and at high speed.

In addition, in the process of transporting the injection molded products, escape rollers or multi-stage stacking prevention means are used to align the injection molded products in a line so that they do not clump, thereby improving the accuracy of inspection and preventing malfunction of the device.

In addition, the false positive rate can be significantly lowered by performing a re-examination by the cloud server and reflecting it in the inspection results.

In addition, the effects according to the embodiments of the present invention mentioned above are not limited to the contents described, and may further include all effects predictable from the specification and drawings.

Figure 1 is a perspective view of an artificial intelligence-based high-speed quality inspection system for injection-molded products according to an embodiment of the present invention.
FIG. 2 is a diagram illustrating an example of a supply unit, which is a component of the artificial intelligence-based high-speed quality inspection system for injection molded products in FIG. 1.
Figure 3 is a detailed view of the alignment transfer unit, which is a component of the artificial intelligence-based high-speed quality inspection system for injection molded products in Figure 1.
FIG. 4 is a diagram showing the vicinity of the image acquisition unit and the defective product detection unit, which are components of the artificial intelligence-based high-speed quality inspection system for injection molded products in FIG. 1.
FIG. 5 is a detailed internal view viewed by cutting the top of the drawing of FIG. 4.
FIG. 6 is a view of the drawing of FIG. 5 viewed from a planar direction.
Figure 7 is an enlarged view of the image acquisition unit of Figure 4.
FIG. 8 is a view of the drawing of FIG. 7 viewed from a planar direction.
Figure 9 is a diagram showing the process in which an injection-molded product is transferred while rotating within a transfer conveyor according to an embodiment of the present invention.
Figure 10 is a diagram showing in detail an example of a multi-stage stacking prevention means according to an embodiment of the present invention.
FIG. 11 is a diagram showing a series of processes in which multi-stage stacking of an injection-molded product is prevented by the multi-stage stacking prevention means of FIG. 10.
Figure 12 is a diagram showing an example of a non-defective product counter unit configured according to an embodiment of the present invention.
Figure 13 is a block diagram showing the connections between each component of the artificial intelligence-based high-speed quality inspection system for injection molded products according to an embodiment of the present invention in Figure 1.
FIG. 14 is a diagram showing the detailed configuration of the control analysis unit shown in FIG. 13 and the connection status of the components.
Figure 15 is a detailed configuration diagram of the PC control panel shown in Figure 14.
FIG. 16 is a flowchart schematically showing the process of operating the artificial intelligence module shown in FIG. 15.

Hereinafter, the description of the present invention with reference to the drawings is not limited to specific embodiments, and various changes may be made and various embodiments may be possible. In addition, the content described below should be understood to include all conversions, equivalents, and substitutes included in the spirit and technical scope of the present invention.

In the following description, the terms first, second, etc. are terms used to describe various components, and their meaning is not limited, and is used only for the purpose of distinguishing one component from other components.

Like reference numerals used throughout this specification refer to like elements.

As used herein, singular expressions include plural expressions, unless the context clearly dictates otherwise. In addition, terms such as “comprise,” “provide,” or “have” used below are intended to designate the presence of features, numbers, steps, operations, components, parts, or a combination thereof described in the specification. It should be construed and understood as not excluding in advance the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts or combinations thereof.

In addition, terms such as “…unit”, “…unit”, and “…module” used in the specification refer to a unit that processes at least one function or operation, which may be implemented through hardware or software or a combination of hardware and software. You can.

Hereinafter, an artificial intelligence-based high-speed quality inspection system for injection molded products according to an embodiment of the present invention will be examined in detail with reference to the attached drawings.

The injection product described in the present invention is not limited to any specific shape, but is preferably a circular or cylindrical injection product that can be rotated by rotation generating means 317, which will be described later, and can be assembled into a tampon, for example. It could be a tampon outerwear. However, as described above, the injection molded product is not necessarily limited to any specific shape.

Referring to the attached drawings FIGS. 1 to 16, the artificial intelligence-based high-speed quality inspection system for injection molded products according to an embodiment of the present invention includes a supply unit 100, an alignment transfer unit 200, an image acquisition unit 300, and a control analysis unit. It may be configured to include 400 and a defective product detection unit 500.

Specifically, the supply unit 100 is configured to supply the injection-molded product, and the form of the supply unit 100 is not limited to any specific type. For example, as shown in the drawing, there is an input hopper 110 into which the injection-molded product is input, and an input hopper ( It may be configured to include a supply conveyor 120 forming a length from 110) to the alignment transfer unit 200.

At this time, one end of the supply conveyor 120 may be connected to the injection product discharge port (not shown) of the input hopper 110, and the other end may be connected to the receiving space 212 of the rotary feeder 210 of the alignment transfer unit 200, which will be described later. You can.

More preferably, the upper end of the input hopper 110 is open to form the inlet 112, and the lower end is open to form an outlet, so that the injection product input into the inlet 112 falls and is discharged through the outlet at the bottom. Here, one end of the supply conveyor 120 is connected to the input hopper 110 so as to be located at the bottom of the discharge port, so that the injection molded product discharged by dropping from the input hopper 110 can be transported along the longitudinal direction.

In addition, the rotation feeder 210 of the alignment transfer unit 200 is formed with an open top and can accommodate the injection molded product by providing a receiving space 212 inside, where the other end of the supply conveyor 120 rotates. It is connected to the rotary feeder 210 so that it is located at the top of the feeder 210, so that the transported injection molded product can be input into the receiving space 212.

At this time, due to the height difference between the discharge port of the input hopper 110 and the top of the rotary feeder 210, the supply conveyor 120 may be lengthened in an inclined direction as shown in the drawing, but this is an example and is not necessarily limited. When the discharge port of the input hopper 110 and the top of the rotary feeder 210 are at the same height, the supply conveyor 120 may be lengthened in a straight line.

On the other hand, the supply conveyor 120 may be formed in a straight shape as shown in FIG. 1, but is preferably formed with a length in an inclined direction to prevent the injection molded product from rolling in the reverse direction of the transport direction, as shown in FIG. 2. As shown, a movement prevention film 125 that protrudes in a vertical or oblique direction with respect to the supply conveyor 120 at regular intervals along the length may be formed.

In addition, the input hopper 110 may be provided with a safety door 114 linked to an operator detection sensor (not shown) on the input port 112 side, as shown in FIG. 2, and the safety door 114 detects the operator. It is formed to open only when a worker is not detected near the input hopper 110 through a sensor, thereby preventing the worker from accidentally falling into the input hopper 110.

At this time, the safety door 114 is preferably provided at a lower position than the input port 112 of the input hopper 110, so that if it is not opened, there is a space for the injection product to be temporarily held between the input port 112 and the safety door 114. It can be formed, and when the operator is not detected in the surrounding area, it can be opened and only the injection molded products accommodated in the temporary holding space can be dropped and transferred to the supply conveyor 120.

In addition, the receiving space 212 provided inside the rotary feeder 210 is equipped with an injection volume detection sensor (not shown) capable of detecting the amount of injection material, and the injection volume detection sensor is linked to the safety door 114. can do. Through this, the safety door 114 can be opened only when the amount of the injected product in the receiving space 212 is maintained below an appropriate level, thereby preventing the injected product from overflowing from the receiving space 212.

The alignment transfer unit 200 is configured to transport the injection molded product supplied from the supply unit 100 while aligning it. For this purpose, the alignment transfer unit 200 may be configured to include a rotary feeder 210 and a straight feeder 220.

Specifically, the rotary feeder 210 may provide a receiving space 212 therein to receive and accommodate the injection molded product transferred from the supply unit 100. Additionally, the rotary feeder 210 may form a guide plate 214 that extends spirally from the lower end to the upper end of the receiving space 212. Additionally, the rotary feeder 210 may be connected to a power generating means (not shown) such as a motor and rotated by power.

Through this, the rotary feeder 210 can rotate by the power of the power generating means and transport the injection-molded products accommodated in the receiving space 212 while aligning them in a line along the guide plate 214.

Here, only one guide plate 214 may be provided, but a plurality of guide plates 214 may be provided within the range of not interfering with each other in order to align and transport more injection molded products, and a separation prevention film 214a is formed upright at least at the outer end. This prevents the injection molded product from deviating to the outside due to centrifugal force.

The straight feeder 220 is provided in a number corresponding to the number of guide plates 214 of the rotary feeder 210, and forms a straight length extending from the end of the guide plate 214 of the rotary feeder 210 to the image acquisition unit 300. Thus, the injection molded product transferred from the rotary feeder 210 may be configured to be delivered to the image acquisition unit 300.

At this time, the linear feeder 220 includes a constant-speed linear feeder 222 and a high-speed linear feeder 224, and can transport the injection molded product at different speeds.

Specifically, the constant speed linear feeder 222 is a conveyor that rotates by power generation means 222a such as a motor, and is located between the rotary feeder 210 and the image acquisition unit 300 and is adjacent to the rotary feeder 210. can be placed. In addition, guide films 222b are erected on both sides to form a transfer line, and can be formed to transfer the injection molded product delivered from the rotary feeder 210 in a row along the transfer line.

Here, the constant speed linear feeder 222 is set to rotate at a constant speed at the same or similar speed as the transfer speed of the injection molded product delivered from the rotary feeder 210, so that the constant speed linear feeder 222 is transferred from the rotary feeder 210. It is possible to ensure that the injection molded product is transported more stably during the delivery process.

The high-speed linear feeder 224 is a conveyor that rotates by a power generating means 224a such as a motor, and guide films 224b are erected on both sides to form a transport line for the injection molded product. In reality, it is a constant-speed linear feeder 222. It may have the same form as, but the installation location and rotation speed may be configured to be different from the constant speed linear feeder 222.

Specifically, the high-speed linear feeder 224 may be placed between the rotary feeder 210 and the image acquisition unit 300 and adjacent to the image acquisition unit 300. That is, the high-speed linear feeder 224 is located behind the constant-speed linear feeder 222 and can deliver the injection product delivered from the constant-speed linear feeder 222 to the image acquisition unit 300.

At this time, the rotation speed of the high-speed linear feeder 224 may be set so that the injection product can be instantaneously delivered to the image acquisition unit 300 faster than the constant-speed linear feeder 222.

This is because the gap between the supply unit 100 and the image acquisition unit 300 is wide, so the distance for the injection product to be transferred is long, or even if a large number of injection products are continuously transferred from the rotary feeder 210 to the constant speed linear feeder 222, the high-speed linear feeder By instantly and quickly transferring the image from 224 to the image acquisition unit 300, stagnation due to friction can be prevented and faster work can be performed.

Meanwhile, the linear feeder 220 may further include an escape roller 230 that separates the injection molded product transferred from the constant-speed linear feeder 222 to the high-speed linear feeder 224.

The escape roller 230 may be provided above the clearance space created between the constant speed linear feeder 222 and the high-speed linear feeder 224, and is preferably located on the constant-speed linear feeder 222 and the high-speed linear feeder 224. A plurality of them may be provided so that at least one is positioned at a time. However, it is not limited, and only one may be provided between the constant speed linear feeder 222 and the high speed linear feeder 224.

The escape roller 230 is formed to pressurize the continuously delivered injection product to a certain degree and can separate the continuous injection product when it is transferred from the constant speed linear feeder 222 to the high-speed linear feeder 224, and the injection product can be transferred to the high-speed linear feeder ( By preventing it from floating before being quickly transferred in 224), it is possible to ensure that it is transferred with stability even at the high speed of the high-speed linear feeder (224).

At this time, the escape roller 230 can be connected to a link 230a, one end of which is connected to a rotating shaft and can rotate, to adjust the degree of pressurization of the injection molded product, and its height can be varied from the straight feeder 220 by adjusting the angle of the link. This allows you to adjust the degree of pressurization. In addition, the angle adjustment of the link 230a can be automated by connecting a power generating means such as a cylinder or a motor to the link 230a.

The image acquisition unit 300 may be configured to receive the injection molded product transferred from the alignment transfer unit 200 and transfer it in one direction, while acquiring an image by capturing the transferred injection product.

To this end, the image acquisition unit 300 may be configured to include a transfer conveyor 310 and a machine vision camera 320.

Specifically, the transfer conveyor 310 may be lengthened in a direction perpendicular to the high-speed linear feeder 224, and a plurality of separators 311 may be installed at regular intervals sufficient to accommodate one injection product. . Here, the longitudinal direction of the transfer conveyor 310 is only a preferred example and is not limited.

In addition, the transfer conveyor 310 is configured to rotate along the longitudinal direction, so that it can receive the injection molded product from the alignment transfer unit 200, load it, and transfer it in one direction.

The machine vision camera 320 is installed on the upper side of the transfer conveyor 310 for transporting the injection-molded product and can obtain an image by capturing the transported injection-molded product. Additionally, the acquired image may be configured to be transmitted to the control analysis unit 400.

Here, the control analysis unit 400 is configured to distinguish defective products by analyzing the image of the injection product transmitted from the image acquisition unit 300. A detailed description of the control analysis unit 400 will be described later.

Meanwhile, the image acquisition unit 300 is preferably configured to rotate and pass the injection product (TO) when the transfer conveyor 310 passes the imaging range of the machine vision camera 320, and captures the injection product at 360°. It may be configured to obtain an image of the front of the injection product.

For this purpose, as shown in FIGS. 7 to 9, the transfer conveyor 310 is formed to have a width smaller than the length of the loaded injection molded product TO, so that at least one end of both ends of the injection molded product protrudes out of the transfer conveyor 310. More preferably, the transfer conveyor 310 is formed with a width smaller than the length of the injection molded product, but guide films 315 can be erected on both sides of the transfer conveyor 310 so that both ends of the injection molded product (TO) protrude. .

Additionally, rotation generating means 317 for rotating the injection-molded product may be provided on both sides of the injection-molded product in the transport direction or in the center between both sides. That is, when the injection product is transported forward, rotation generating means 317 may be provided on the left and right sides, or rotation generating means 317 may be provided in the center between the left and right sides.

Here, the rotation generating means 317 may be a friction body that causes rolling friction or surface friction such as a roller or conveyor to rotate, and may be provided along the entire length of the transfer conveyor 310, but is preferably a machine vision camera. It can be formed only to a length corresponding to the imaging range of 320.

Through this, when the injection product (TO) transported by the transfer conveyor 310 passes through the imaging range of the machine vision camera 320, it rotates due to friction with the rotation generating means 317, causing the machine vision camera 320 to rotate. It is possible to acquire the entire image of the injection molded product (TO).

In addition, the image acquisition unit 300 may further include a multi-stage stacking prevention means 330 that prevents the injection-molded product (TO) being loaded and transported on the transfer conveyor 310 from being stacked in multiple stages.

Specifically, the transfer conveyor 310 forms the separator 311 to a height equal to the diameter of the injection product (TO) so that the injection product (TO), which is rapidly transferred from the high-speed linear feeder 224, is not loaded in more than two stages. , it is not possible to prevent stacking between the injection molded products on the transfer conveyor 310, but stacking can be prevented by constructing a multi-stage stacking prevention means 330.

Here, the multi-stage stacking prevention means 330 may be provided on the upper side of the transfer conveyor 310 in front of the position of the machine vision camera 320, and is not limited, but preferably, when the injection molded product (TO) is stacked in two stages. It can be provided with a fixing plate 331 or a fixing rod 332 located at a height lower than the height of.

In the case of the fixing plate 331, as shown in Figure 4 or 5, it is formed in an 'ㄱ' shape, for example, and can be fixedly installed on one side of the transfer conveyor 310, and in the case of the fixing rod 332, as shown in Figure 10 As shown, a mounting groove 332a is formed on the guide film 315 formed in front of the machine vision camera 320, and is formed to be mounted in the mounting groove 332a so that it can be easily separated. .

However, as mentioned above, the form of the multi-stage stacking prevention means 330 is not limited to these. It does not matter what form it is formed in as long as it can prevent stacking of the injection-molded product in more than two stages. In addition to the stationary type, it can be of an insertion type or a fastening type. It may be provided.

The multi-stage stacking prevention means 330, which consists of a fixing plate 331 or a fixing rod 332, is an injection molded product that is stacked in multiple stages at a fixed or mounted position and transported along the transfer conveyor 310, as shown in FIG. 11. can be pushed backward, and the injection molded product pushed by the multi-stage stacking prevention means 330 can be moved to the empty space between the separation membranes 311 of the transfer conveyor 310 and loaded.

In addition, the image acquisition unit 300 may further include a lighting device 340 installed on both sides of the transfer conveyor 310 where the machine vision camera 320 is located. The lighting device 340 emits light within the range where the machine vision camera 320 captures images of the injection product, thereby further increasing the clarity of the image acquired by the machine vision camera 320, and through this, the control analysis unit 400 performs analysis. When performing the process, defective products can be distinguished with clearer image quality, and the defective product classification rate can be further improved.

The control analysis unit 400 is linked with the image acquisition unit 300 and can receive the image of the injection molded product obtained from the image acquisition unit 300. In addition, the control analysis unit 400 is equipped with an artificial intelligence module and can distinguish defective products by analyzing the received injection product image based on the artificial intelligence module.

In addition, the control analysis unit 400 includes each part of the artificial intelligence-based high-speed quality inspection system for injection molded products of the present invention, namely, the supply unit 100, the alignment transfer unit 200, the image acquisition unit 300, the defective product detection unit 500, etc. Can be configured to control.

Through this, the control analysis unit 400 can classify defective products based on the artificial intelligence module and then control the defective product detection unit 500, which will be described later, to cause the defective product detection unit 500 to detect defective products.

Specifically, the control analysis unit 400 includes a PLC control panel 410 linked to an HMI (Human Machine Interface, 415) and a PC control panel 420 consisting of an artificial intelligence module 421 and a vision module 422. The PLC control panel 410 and the PC control panel 420 can be configured to be connected to each other through a wired or wireless network and can be operated in conjunction with each other.

At this time, the PC control panel 420 runs the vision module 422, which is linked to the machine vision camera 320, by interacting with the artificial intelligence module 421, and detects defective products through learning by the artificial intelligence module 421. This allows for more detailed, faster, and more precise classification of defective products than distinguishing defective products with a simple vision module.

More specifically, when acquiring an image of an injection product using the machine vision camera 320, the artificial intelligence module 421 adjusts the camera exposure time and illuminance through control of the machine vision camera 320 and the lighting device 340. It is possible to create various types of image data by adjusting them and collect them to collect learning and verification data sets with control values most suitable for the artificial intelligence module 421.

Afterwards, the collected data set can be preprocessed. During this process, light reflection due to the curved nature of the product can be processed with the vision module 422, and the artificial intelligence module 421 processes the preprocessed image data. By labeling defective locations such as black spots, browning, and scratches, you can distinguish between normal and defective products in a more detailed, faster, and more precise manner.

That is, the artificial intelligence module 421 collects images captured from the machine vision camera 320 in the image collection step (S10). If the image does not meet the learning and verification requirements in the image collection step (S10), the machine vision camera A control step (S20) that satisfies the learning and verification requirements by controlling (320) and the lighting device (340), and the images collected after the control step (S20) are used to remove light reflection, etc. using the vision module (422). An image pre-processing step (S30), which pre-processes the image pre-processing step (S30) to learn the location of defects such as black spots, browning, and scratches by comparing the image pre-processed in the image pre-processing step (S30) with a normal product image, and a learning step (S40). It is possible to distinguish defective products, including the defective product determination step (S50), which identifies and labels the learned defective location and distinguishes defective products.

Here, the artificial intelligence module 421 is not limited to any specific form, but preferably, for example, a deep learning model of PaDiM can be used, and the operation of the control analysis unit 400, such as classification of defective products or control, can be performed with the naked eye. The PC control panel 420 may be connected to the monitor 425 for confirmation.

The defect detection unit 500 may be provided in the transport direction of the injection-molded product transported by the image acquisition unit 300. Additionally, the defective product detection unit 500 may be controlled by the control analysis unit 400 and configured to detect classified defective products.

To this end, the defective product detection unit 500 may be configured to include a punching means 510 and a recovery device 520.

Specifically, the punching means 510 may be provided on one side of the injection direction of the injection molded product transported by the image acquisition unit 300. Additionally, the punching means 510 may be configured to receive a control signal from the control analysis unit 400 and push the defective product from a provided position to the other side.

The recovery device 520 may be provided in a position opposite to the punching means 510. That is, the punching means 510 and the recovery device 520 may be provided on both sides of the transfer conveyor 310 and formed at positions facing each other, with the transfer conveyor 310 in between.

Through this, the recovery device 520 can recover defective products pushed out by the punching means 510 on the other side of the punching means 510, and good products excluding the recovered defective products continue to be transported along the transfer conveyor 310. It can be.

Meanwhile, the artificial intelligence-based high-speed quality inspection system for injection molded products according to an embodiment of the present invention includes a cloud server 600 that receives classified defective product information from the control analysis unit 400 and performs reanalysis of defective products, as shown in FIG. 13. It may also include more.

The cloud server 600 can receive defective product information, perform reanalysis of the defective product, and store the reanalyzed results separately. At the same time, it can transmit the defective product reanalysis information to the control analysis unit 400.

The control analysis unit 400, which receives defective product reanalysis information from the cloud server 600, can perform artificial intelligence learning by comparing the defective product information analyzed by itself with the defective product reanalysis information transmitted from the cloud server 600. Defective products can be judged more accurately.

In addition, the artificial intelligence-based high-speed quality inspection system for injection-molded products according to an embodiment of the present invention is provided in the transfer direction of the injection-molded product transported by the image acquisition unit 300 behind the defective product detection unit 500, as shown in FIG. 12. It may further include a good product counter unit 700 that counts injection molded products of good quality.

The non-defective product counter unit 700 is equipped with a photo sensor, etc. to capture and count the transported injection molded product, and can determine the defect rate by comparing it with the amount supplied using the supply unit 100.

Although embodiments of the present invention have been described above with reference to the attached drawings, those skilled in the art will understand that the present invention can be implemented in other specific forms without changing the technical idea or essential features of the present invention. You will be able to understand it. Therefore, the embodiments described above are illustrative in all respects and are not restrictive.

100: supply department
110: Input hopper
112: Inlet
114: safety door
120: Supply conveyor
125: Movement prevention shield
200: Alignment transfer unit
210: Rotating feeder
212: Accommodation space
214: Information board
214a: Breakaway prevention film
220: straight feeder
222: Constant speed straight feeder
222a: Power generation means
222b: guide membrane
224: High-speed straight feeder
224a: Power generation means
224b: guide membrane
230: Escape roller
230a: link
300: Image acquisition unit
310: transfer conveyor
311: Separator
315: guide membrane
317: Rotation generating means
320: machine vision camera
330: Multi-stage stacking prevention means
331: fixing plate
332: fixed rod
332a: Holding groove
340: lighting device
400: Control analysis department
410: PLC control panel
415: HMI
420: PC control panel
421: Artificial intelligence module
422: Vision module
425: monitor
500: Defective product detection unit
510: Punching means
520: recovery device
600: Cloud server
700: Good product counter unit
TO: injection molded product

Claims (12)

A supply unit that supplies injection molded products;
an alignment transfer unit that aligns and transfers the injection molded product supplied from the supply unit;
an image acquisition unit configured to receive the injection-molded product transferred from the alignment transfer unit and transfer it to one side, while acquiring an image by imaging the transferred injection-molded product;
A control analysis unit that analyzes the image of the injection product obtained from the image acquisition unit based on an artificial intelligence module and a vision module to classify defective products and controls each part;
A defective product detection unit provided in the transport direction of the injection molded product transported by the image acquisition unit to detect classified defective products,
The alignment transfer unit,
A receiving space is provided inside to accommodate the injection molded product supplied from the supply unit, and a guide plate is formed spirally extending from the lower end of the receiving space to the upper end, and the received injection molded product is aligned and transported along the guide plate while rotating by power. rotary feeder and
It includes a straight feeder that forms a straight length from the end of the guide plate of the rotary feeder to the image acquisition unit and delivers the injection product transferred from the rotary feeder to the image acquisition unit,
The straight feeder is,
A constant speed linear feeder that receives the injection material from the rotary feeder and transfers it at a constant speed, and
It includes a high-speed linear feeder that receives the injection molded material transferred from the constant-speed linear feeder and transfers it to the image acquisition unit at a faster speed than the constant-speed linear feeder,
The straight feeder is,
Further comprising an escape roller that separates the injection molded product delivered from the constant-speed linear feeder to the high-speed linear feeder,
Artificial intelligence-based high-speed quality inspection system for injection molded products.
delete delete delete According to claim 1,
The image acquisition unit,
A transfer conveyor in which a plurality of separation membranes are installed at regular intervals to accommodate one injection product, receives the injection product from the alignment transfer unit, loads it, and transfers it in one direction, and
Comprising a machine vision camera installed on the upper side of the transfer conveyor to acquire an image by imaging the transported injection molded product, and transmitting the obtained image to the control analysis unit,
Artificial intelligence-based high-speed quality inspection system for injection molded products.
According to claim 5,
The transfer conveyor,
It is formed to have a width smaller than the length of the loaded injection-molded product so that at least one end of both ends of the injection-molded product protrudes outward,
Rotation generating means for rotating the injection-molded product are provided on both sides of the transport direction of the injection-molded product or in the center between the two sides,
Characterized in that the machine vision camera can acquire the entire image of the injection molded product by allowing the injection molded product transported by the transfer conveyor to rotate and move.
Artificial intelligence-based high-speed quality inspection system for injection molded products.
According to claim 5,
The image acquisition unit,
Further comprising a multi-stage stacking prevention means provided on the upper side of the transfer conveyor in front of the position of the machine vision camera to prevent multi-stage stacking of the transported injection molded product,
Artificial intelligence-based high-speed quality inspection system for injection molded products.
According to claim 5,
The image acquisition unit,
Further comprising a lighting device installed on both sides of the transfer conveyor where the machine vision camera is located and emitting light to further increase the clarity of the image acquired by the machine vision camera,
Artificial intelligence-based high-speed quality inspection system for injection molded products.
According to claim 8,
The artificial intelligence module is,
An image collection step of collecting images captured from the machine vision camera;
If the image does not meet the learning and verification requirements in the image collection step, a control step of controlling the machine vision camera and lighting device to meet the learning and verification requirements;
An image pre-processing step of pre-processing the images collected after the control step to remove light reflection using a vision module;
A learning step of learning the defective location by comparing the image preprocessed in the image preprocessing step with the normal product image, and
Characterized by distinguishing defective products, including a defective product determination step of determining and labeling the defective location learned through the learning step and distinguishing defective products.
Artificial intelligence-based high-speed quality inspection system for injection molded products.
According to claim 1,
The defective product detection unit,
A punching means provided on one side of the transport direction of the injection molded product transported by the image acquisition unit and receiving a control signal from the control analysis unit to push the defective product from the provided position to the other side;
A recovery device provided at a position opposite the punching means to recover defective products pushed out by the punching means,
Artificial intelligence-based high-speed quality inspection system for injection molded products.
According to claim 1,
Further comprising a cloud server that receives classified defective product information from the control analysis unit, performs reanalysis of the defective product, and transmits the performed defective product reanalysis information to the control analysis unit,
Artificial intelligence-based high-speed quality inspection system for injection molded products.
According to claim 1,
Further comprising a good product counter provided in the transport direction of the injection-molded product transported by the image acquisition unit behind the defective product detection unit to count the injection-molded product of the non-defective product,
Artificial intelligence-based high-speed quality inspection system for injection molded products.