JP7167404B2 - Mold monitoring system - Google Patents

Description

Disclosed herein is a mold monitoring system for monitoring molds for insert molding.

2. Description of the Related Art Conventionally, monitoring systems for monitoring molds used for resin molding have been known. For example, in Patent Literature 1, the presence or absence of abnormality is determined by capturing an image of a mold with a camera. Regarding this abnormality determination, for example, in Patent Document 2, a reference image of a mold captured in advance is compared with a captured image to determine whether there is an abnormality.

Japanese Patent No. 6584472 JP 2012-91354 A

By the way, in mold monitoring, there is a case where the machined surface of the mold for insert molding is monitored. The machined surface refers to the surface of the mold on which at least one of the cavity and the core and the mounting portion for mounting the insert component are formed. When the computer monitors the machined surface, the imaged image of the machined surface is compared with a reference image, which is the normal image.

For example, items to be monitored for the machined surface include whether or not an insert part is attached and whether or not there is a foreign object. In the former case, for example, a missing part in which the insert part is not attached to the attachment part, and insufficient insertion (floating) of the insert part into the attachment part are examples of improper attachment. Regarding the floating of the insert part, it is necessary to recognize the relative positional deviation between the mounting portion and the insert part, and the required determination accuracy is relatively fine.

On the other hand, in determining the presence or absence of foreign matter, it is determined whether or not a foreign matter such as an insert part has fallen on the machined surface. In this judgment, in short, it is sufficient that the foreign matter can be recognized, and a rougher judgment accuracy than that for the mounting quality judgment of the insert part is sufficient.

In addition, so-called gas burning occurs on the working surface of the mold. Gas burn occurs when the gas inside the mold or the gas generated from the resin material is compressed when the mold is clamped and burns at a high temperature. Refers to the phenomenon of blackening. In addition, if the effect of gas burn is minor, such as a slight change in the color of the processed surface due to blackening or a minor blackened area, the molding process will be immediately stopped because it does not affect the quality of the molded product. No need.

However, if the presence/absence of foreign matter is determined with the accuracy required to determine whether or not the insert part is mounted properly, slight gas burns on the machined surface may be detected, and based on this, an erroneous determination of the presence of foreign matter may be made. .

Therefore, in this specification, a mold monitoring system capable of making determinations with appropriate accuracy when determining whether an insert part is mounted properly and whether or not there is a foreign object on a processed surface based on a captured image of the processed surface of the mold. is disclosed.

Disclosed herein is a mold monitoring system for monitoring molds for insert molding. The system includes an imager, an image extractor, an insert determiner, and a foreign object determiner. The imaging device captures an image of the working surface of the mold, in which at least one of the cavity and the core is formed and a mounting portion to which the insert component is mounted is formed. The image extracting section cuts out the mounting section area image from the processing surface image captured by the imaging device. The insert determination unit determines whether the insert component is attached in the attachment area image. The foreign matter determining unit determines whether or not there is a foreign matter in the processed surface residual image from which the mounting portion area image is cut. The insert determination unit includes an insert determination neural network in which an input layer receives an image of the attachment area and an output layer outputs whether or not the insert component is attached. The foreign matter determination unit includes a foreign matter determination neural network in which a machined surface residual image is input to an input layer and the presence or absence of foreign matter is output from an output layer. In the above configuration, the number of hidden layers in the foreign matter determination neural network is smaller than that in the insert determination neural network.

Generally, it is known that in a neural network, the more the number of hidden layers (intermediate layers) increases, the higher the determination accuracy becomes. According to the above configuration, by making the number of hidden layers of the foreign matter determination neural network smaller than the number of hidden layers of the insert determination neural network, it is possible to make a relatively rough determination, and the Detecting slight gas burning and erroneously determining it as a foreign object is suppressed.

Further, in the above configuration, the machined surface residual image may be divided into a plurality of images according to the illuminance, and a foreign matter determination neural network may be provided for each divided image.

In the captured image of the processing surface, in the image area with relatively high illumination (bright), the change of the processing surface due to gas burn appears relatively large (clearly), and in the image area with relatively low illumination (dark) , the change in the machined surface due to gas burning appears relatively small. By dividing the machined surface residual image according to the illuminance, the low-illuminance area, which is an image area with low illuminance, is separated from the high-illuminance area, which is an image area with high illuminance. As a result, the low-illuminance area is not affected by changes in the processing surface due to gas burn in the high-illuminance image area.

Further, in the above configuration, the die monitoring system may include a light source for illuminating the machining surface. In this case, the light source may be a red light source.

When the machined surface of the mold is irradiated with a light source, the irradiated light may be reflected and halation may occur in the machined surface image, making it difficult to grasp the machined surface. Here, by utilizing the characteristic that the total reflectance of the red light source with respect to the metal mirror surface is lower than that of other color light sources, the occurrence of halation in the processed surface image can be suppressed by using the red light source.

Further, in the above configuration, the learning teacher data of the foreign matter determination neural network may be updated. In this case, the machined surface image in the most recent predetermined period may be used as the teacher data.

By re-learning the foreign matter determination neural network using training data that uses the processed surface image in the most recent predetermined period, it is possible to suppress erroneous determinations, especially in high-illuminance areas where changes in gas burns on the processed surface appear significantly. .

According to the work management system disclosed in the present specification, when determining whether the insert component is mounted properly and whether or not there is a foreign object on the processed surface based on the captured image of the processed surface of the mold, each determination can be made with appropriate accuracy. becomes possible.

It is a figure which illustrates the hardware constitutions of the metal mold monitoring system which concerns on this embodiment. It is a figure which illustrates the functional block of the die monitoring system concerning this embodiment. It is a figure which illustrates the processing surface image of a metal mold|die. FIG. 11 is a diagram illustrating an image of the machined surface of the mold when the insert component is correctly mounted; FIG. 10 is a diagram illustrating an image of the working surface of the mold when the insert component is incorrectly installed; It is a figure which illustrates the processing surface image of a metal mold|die when a foreign material falls on a processing surface. FIG. 10 is a diagram for explaining cutting of an applied portion area image; FIG. 10 is a diagram illustrating a process of dividing a machined surface residual image into a plurality of images; FIG. 4 is a diagram illustrating a determination unit of an insert determination section; FIG. 4 is a diagram illustrating a neural network (insertion determination neural network) in a determination unit of an insert determination section; FIG. 4 is a diagram illustrating a determination unit of a foreign matter determination section; FIG. 4 is a diagram illustrating a neural network (neural network for foreign matter determination) in a determination unit of a foreign matter determination section; It is a figure which illustrates the die monitoring flow in the die monitoring system which concerns on this embodiment. FIG. 9 is a diagram showing another example of the die monitoring flow in the die monitoring system according to the present embodiment;

<Overall composition>
FIG. 1 illustrates a die monitoring system according to this embodiment. As described below, the system monitors molds for insert molding. During insert molding, the system determines whether or not the insert component is correctly attached to the mold and whether or not foreign matter has fallen on the processing surface of the mold. This die monitoring system includes a die monitoring device 10 , an imaging device 40 and a light source 42 .

The die monitoring device 10 is composed of, for example, a computer. The mold monitoring apparatus 10 includes a CPU 11 as an arithmetic unit, and a system memory 12 and a storage 13 as storage means. The storage 13 may be a non-transitory storage device such as a hard disk drive (HDD) or solid state drive (SSD). The mold monitoring apparatus 10 also includes an input unit 14 such as a keyboard and a mouse, and an input/output controller 15 that manages input/output of information with external devices such as the imaging device 40 .

Furthermore, the mold monitoring apparatus 10 includes a GPU 16 (Graphics Processing Unit), a frame memory 17, a RAMDAC 18 (Random Access Memory Digital-to-Analog Converter), and a display control unit as means for processing the captured image captured by the imaging device 40. 19. In addition, the mold monitoring device 10 includes a display section 20 for displaying processed images. The mold monitoring device 10 may include a touch panel display in which the input section 14 and the display section 20 are integrated.

The GPU 16 is an arithmetic unit for image processing, and is mainly operated when insert determination and foreign matter determination, which will be described later, are performed. The frame memory 17 is a storage device that stores images captured by the imaging device 40 and processed by the GPU 16 . The RAMDAC 18 converts the image data stored in the frame memory 17 into analog signals for the display unit 20, which is an analog display.

The display control unit 19 causes the display unit 20 to display images processed through the GPU 16 , the frame memory 17 and the RAMDAC 18 . For example, the display control unit 19 causes the display unit 20 to superimpose an image captured by the imaging device 40 and a highlighting (described later) that emphasizes a location that is the basis for abnormality determination by insert determination or foreign matter determination.

As will be described later, when the insertion determination unit 32 (see FIG. 2) determines that the insertion is defective or the foreign object determination unit 33 determines that there is a foreign object, the display control unit 19 displays Cause the unit 20 to display a warning message. Further, when the insertion determination section 32 determines that the insertion is normal and the foreign matter determination section 33 determines that there is no foreign matter, the display control section 19 causes the display section 20 to display an acceptance message.

In FIG. 1, the mold monitoring apparatus 10 includes the display unit 20. However, the mold monitoring apparatus 10 (computer) including components other than the input unit 14 and the display unit 20 may be separated from the assembly site. It may be installed in a separate server room. Also, the display unit 20 may be installed at the work site so as to be viewable by an insert molding operator.

Also, a touch panel display in which the display unit 20 and the input unit 14 are integrated may be installed at the work site. As will be described later, when at least one of mounting failure and foreign matter detection occurs, the molding operation is temporarily stopped. A confirmation button is displayed on the touch panel display, and the operator can operate the button so that the molding operation can be resumed after correcting the mounting failure of the insert part and removing the foreign matter.

FIG. 2 exemplifies functional blocks of the mold monitoring device 10 mixed with the hardware blocks shown in FIG. This functional block diagram is configured by the CPU 11 executing a program stored in, for example, the storage 13 or a non-transitory computer-readable storage medium such as a DVD.

The mold monitoring apparatus 10 includes an image extraction unit 31, an insert determination unit 32, a foreign matter determination unit 33, and a molding apparatus control unit 34 as functional blocks. As will be described later, the image extractor 31 cuts out the mounting portion area images 71, 72, and 73 (see FIG. 7) from the processing surface image 50 (see FIG. 3) captured by the imaging device 40. FIG. The insert determination unit 32 determines whether the insert components (the first insert screw 61, the second insert screw 62, and the cap 63 in FIG. 4) in the mounting portion area images 71, 72, and 73 are properly mounted. The foreign matter determination unit 33 determines the presence or absence of a foreign matter in a machined surface residual image 75 (see FIG. 8) from which the mounting portion area images 71, 72, and 73 are cut.

The molding device control unit 34 can output control commands (stop command and restart command) to the insert molding device. As will be described later, when the insertion of the insert part into the mold of the insert molding apparatus is completed, the worker presses a mold clamping button to match (clamp) the fixed mold and the movable mold. At this time, if the mold monitoring device 10 detects a mounting failure of the insert component or foreign matter on the machined surface, the molding device control unit 34 sends a stop command to the insert molding device.

As will be described later, when the operator corrects the mounting failure of the insert part and removes the foreign matter from the machined surface, the operator sends the input section 14 (confirmation button) to the molding apparatus control section 34 to indicate that the abnormality has been resolved. A confirmation signal is sent. In response to this, the molding device control unit 34 transmits a restart command for restarting the stopped process to the insert molding device.

The imaging device 40 is, for example, a camera device installed at an insert molding work site, and is capable of capturing still images and moving images of the insert molding work. The imaging device 40 includes an imaging device such as CMOS or CCD, for example. In addition, since the imaging device 40 performs fixed imaging, the imaging conditions such as the position and orientation, the angle of view, and the magnification are fixed.

In particular, the imaging device 40 images the processing surface of the mold for the same molding provided in the insert molding apparatus. The processing surface of the mold refers to the surface of the mold on which at least one of a cavity (female mold) and a core (male mold) is formed, and a mounting portion to which an insert part is mounted is formed. . For example, when closing the mold of the movable platen with the mold of the stationary platen, the facing surfaces (mating surfaces) of the molds serve as working surfaces.

For example, the imaging device 40 takes an image at a predetermined timing by an operator's operation. For example, after an operator mounts an insert part on a mounting part provided on the processing surface of the mold in the open state, the mold clamping button is pressed by the operator in order to proceed to the next process, for example, the mold closing process. be When the mold clamping button is pressed, the imaging device 40 images the machined surface of the mold.

Referring to FIG. 2 , a light source 42 irradiates light onto the processing surface of the mold, which is an imaging target of the imaging device 40 . The light source 42 may be, for example, a red light source, and the working surface of the mold is illuminated red by the light source. It is generally known that the total reflectance of a red light source on a metallic mirror surface is lower than that of other color light sources. In the die monitoring system according to the present embodiment, by utilizing this characteristic and using a red light source as the light source 42, the occurrence of halation in the processed surface image is suppressed.

<Example of captured image of mold>
FIG. 3 illustrates a processed surface image 50 that is an image of the processed surface captured by the imaging device 40 . For example, this working surface may be the working surface of a mold attached to a stationary platen of an insert molding apparatus.

Further, FIG. 4 shows an example when the insert component is correctly mounted on the mounting portion of the machined surface. In the example of FIG. 3, the mounting portion includes a first pin 56, a second pin 57, and a cap receiver 58. As shown in FIG. Also in the example of FIG. 4 , the insert parts include a first insert screw 61 , a second insert screw 62 and a cap 63 . However, in the mold monitoring system according to this embodiment, the mounting portion and the insert part are not limited to the above examples.

As exemplified in FIG. 3, a parting surface 55 that is a main surface and indicates a parting line is formed in the processing surface image. Furthermore, the cavity 51 is formed so as to be recessed from the parting surface 55 . The cavity 51 faces the core provided in the mold of the movable platen when the mold is closed, and the gap formed between them serves as a molding space for the molded product.

A cylindrical protrusion 52 is formed in the cavity 51 so as to protrude from the bottom surface thereof in the height direction. A cylindrical protrusion 53 is formed at the center opening of the cylindrical protrusion 52 . The cylindrical protrusion 53 is concentric with the cylindrical protrusion 52 and has an outer peripheral surface spaced from the inner peripheral surface of the cylindrical protrusion 52 .

A cap receiver 58 is formed on the upper end surface of the columnar protrusion 53 so as to slightly protrude from the upper end surface. The cap receiver 58 is a mounting portion to which a cap 63 (see FIG. 4), which is an insert part, is mounted (fitted). A groove 58A for alignment is formed in the central portion of the cap receiver 58, and a key (not shown) provided on the surface of the cap 63 facing the cap receiver 58 is inserted into the groove 58A. By doing so, the cap 63 is normally attached to the cap receiver 58 .

A rod-like protrusion 54 extends from the cylindrical protrusion 52 along the radial direction thereof. A first pin 56 and a second pin 57, which are mounting portions protruding from the upper end surface of the bar-shaped protrusion 54, are formed. Both the first pin 56 and the second pin 57 are cylindrical, and as illustrated in FIG. 4, a first insert screw 61 and a second insert screw 62, which are cylindrical insert parts, are inserted.

FIG. 5 shows an example of improper mounting of an insert component. In this example, the lower end of the first insert screw 61 does not reach the lower end of the first pin 56, which is the so-called "floating". In order to determine such an abnormality on the image, it is necessary to be able to recognize at least the relative positional deviation between the first insert screw 61 and the first pin 56 .

FIG. 6 shows an example of falling foreign matter. In the die monitoring system according to the present embodiment, it is mainly assumed that the foreign matter is dropped onto the machining surface of the insert part. Therefore, when determining the presence or absence of foreign matter, the insert part may be set as the smallest foreign matter to be determined. When the insert part is composed of a plurality of types of members with different sizes, the smallest of them may be regarded as the smallest foreign matter, and the presence or absence of foreign matter may be determined.

In the example of FIG. 6, the insert screw 65 has fallen into the cavity 51. As shown in FIG. In order to determine such an abnormality on the image, it is sufficient that the insert screw 65 is at least visible, and the required determination accuracy is low compared to the improper mounting shown in FIG.

Also, in insert molding, it is known that the resin, which is the molding material, burns on the mold, resulting in "gas burning" in which the machined surface turns black. If such gas scorch occurs excessively on the machined surface, there is a risk of defects in the molded product, but gas scorching to some extent does not affect the quality of the molded product. In other words, if the gas burning is minor, the molding operation may be continued. In this way, in foreign matter determination on the machined surface, a so-called coarse determination accuracy is an appropriate determination accuracy that does not detect such minor gas burning as an abnormality.

As will be described later, the mold monitoring system according to the present embodiment is provided with an insert determination neural network for determining whether an insert component is mounted properly and a foreign matter determination neural network for determining the presence or absence of foreign matter. In view of the above-described difference in judgment accuracy, in the mold monitoring system according to the present embodiment, each network is configured so that the number of hidden layers of the foreign matter judgment neural network is smaller than that of the insert judgment neural network. be done. Since the number of hidden layers is relatively small, it is possible to make a rough judgment as described above, and it is possible to suppress erroneous detection of minor gas burns as abnormalities.

<Image extraction part>
7 and 8 show an example of image extraction by the image extraction unit 31 (see FIG. 2). The image extraction unit 31 cuts out the mounting portion region image from the processing surface image 50 . In FIG. 7, a first pin image 71, a second pin image 72, and a cap receiving image 73 are cut out from the processing surface image 50 as the mounting portion area image.

Since the applied part area image is used to determine whether the insert part is properly attached to the applied part, the applied part area image includes each applied part and its surrounding area. The region of the mounting portion region image in the processing surface image 50 is set in advance by, for example, the administrator of the mold monitoring device 10 or the like.

Referring to FIG. 8, a processed surface residual image 75, which is a residual image obtained by cutting out the applied portion area image, is divided into a plurality of images. This division is performed according to the illuminance (brightness) of each pixel in the machined surface residual image 75 .

The processed surface of the mold has three-dimensional unevenness, and a dark area and a bright area appear on the captured image. For example, a location including shadows by the cylindrical protrusion 52, the columnar protrusion 53, and the rod-shaped protrusion 54, and a location away from the light source 42 are low-illuminance areas with relatively low illumination. On the other hand, the upper end surfaces of the cylindrical projection 52, the columnar projection 53, and the rod-shaped projection 54, and the portion of the parting surface 55 near the light source 42 are high-illuminance areas with relatively high illuminance.

As described above, the surface change (blackening) of the machined surface due to gas burning appears more clearly in the high-illuminance area than in the low-illuminance area. For example, the blackening of the machined surface due to gas burn is small in the originally dark low-illuminance area, but in the high-illuminance area brightly illuminated by the light source 42, the change (blackening) of the machined surface due to gas burn becomes clear. appear. The machined surface residual image 75 is divided into a plurality of images based on the difference in the degree of change in gas burn.

For example, as illustrated in FIG. 8, the machined surface residual image 75 is divided into an A area 75A, a B area 75B, a C area 75C, a D area 75D, and an E area 75E with thick dashed lines as boundaries. As will be described later, in foreign matter determination, a neural network for foreign matter determination is constructed for each of these divided area images, and the presence or absence of a foreign matter is determined for each area.

By dividing the image by illuminance in this way and constructing a neural network for detecting foreign matter for each image, for example, in a low-illuminance area, it is affected by the change (blackening) of the processed surface in a high-illuminance area. foreign object determination can be performed without In addition, in high-illuminance areas, frequent re-learning is set as described later, that is, by setting a short period from the previous learning to the next re-learning, it is possible to perform foreign matter determination that follows changes in the machined surface. .

<Insert determination part>
With reference to FIG. 2, the attachment area images extracted by the image extraction unit 31 (the first pin image 71, the second pin image 72, and the cap receiving image 73 in the example of FIG. 7) are sent to the insert determination unit 32. be done. FIG. 9 illustrates functional blocks of the insert determination unit 32 . The insert determination unit 32 includes determination units for each of the plurality of types of attachment area images transmitted from the image extraction unit 31 . For example, in the example of FIG. 9, a first pin determination unit 32A, a second pin determination unit 32B, and a cap reception determination unit 32C are provided in the insert determination section 32 according to the attachment area image extracted in FIG.

These determination units each have an insert determination neural network, which is an independent neural network. For example, these neural networks are configured from a convolutional neural network (CNN) from the viewpoint of image recognition.

These neural networks are preliminarily trained using teacher data. The teacher data includes an applied portion area image as input data, and an assembly quality class corresponding to the image as output data. The assembly quality class, in short, indicates the result of the determination of whether or not the mounting is necessary, and includes the normal class and the defective class. Defects are further subdivided into classes such as "floating", "missing item", and "wrong part". The selection of individual mounting portion region images and the assignment of classes corresponding to the respective images are performed by, for example, the administrator of the mold monitoring system.

For example, FIG. 10 illustrates the neural network of the first pin determination unit 32A. The training data of the first pin determination unit 32A includes the first pin image 71 as input data, and classes such as "normal", "floating", "missing item", and "wrong part" as output data corresponding to each image. .

The number of nodes in the input layer of the neural network of each determination unit provided in the insert determination section 32, including the first pin determination unit 32A, is the same as the number of pixels of the applied portion area image transmitted from the image extraction section 31. can be For example, the number of nodes in the input layer of the neural network of the first pin determination unit 32A may be the same number as the number of pixels of the first pin image 71 . Also, the number of nodes in the output layer of the neural network of each determination unit may be the same as the number of the above assembly quality classes.

Further, the number of hidden layers (intermediate layers) of the neural network of each determination unit of the insert determination section 32 is configured to be greater than the number of hidden layers of the neural network of the foreign matter determination section 33, which will be described later. For example, all of the neural networks of the determination units in the insert determination section 32 are set to have more hidden layers than the neural networks of any of the determination units in the foreign matter determination section 33 .

It is generally known that in neural networks, the more hidden layers there are, the higher the determination accuracy becomes. All of the neural networks of the determination units in the insert determination section 32 are set to have more hidden layers than the neural networks of any of the determination units in the foreign matter determination section 33. Higher determination accuracy can be obtained in the insert determination unit 32 than in the determination unit 33 .

Also, most of the image area of the mounting part area image is occupied by the insert part. The insert part is attached to the mold when the mold is open, and is imaged by the imaging device 40 before the mold is closed. Since gas burns occur on the working surface of the mold after the mold is closed, the surface of the insert component captured by the imaging device 40 before the mold is closed is not affected by gas burns. That is, since the mounting area image is basically narrowed down to areas that are less likely to be affected by gas burns, there is a low possibility of erroneous determination due to gas burns as in foreign matter determination. In this manner, even if the determination accuracy is higher than that of the foreign object determination unit 33, erroneous determination due to gas burning is less likely to occur in the attachment quality determination.

For example, as illustrated in FIG. 10, the first pin image is input to the input layer of the first pin determination unit 32A. In the example of this figure, an image of a state in which the first insert screw 61 is floating with respect to the first pin 56 is input. In response to this input, the output layer of the first pin determination unit 32A outputs whether or not the insert part needs to be mounted. Specifically, the degree of certainty of each class is obtained in the insert determination neural network, and an output is obtained that maximizes the degree of certainty of floating, which is one aspect of poor mounting. The first pin determination unit 32A outputs "float" with the highest certainty as the final determination result.

<Foreign matter detection unit>
2, the machined surface residual image 75 extracted by the image extracting unit 31 is divided by illuminance (in the example of FIG. sent.

FIG. 11 illustrates functional blocks of the foreign object determination unit 33. As shown in FIG. The foreign matter determination section 33 includes determination units for each of the divided areas 75A to 75E of the machined surface residual image 75 transmitted from the image extraction section 31 . For example, in the example of FIG. 11, the foreign matter determination section 33 is provided with an A area determination unit 33A, a B area determination unit 33B, a C area determination unit 33C, a D area determination unit 33D, and an E area determination unit 33E.

These determination units each have a neural network for foreign matter determination, which is an independent neural network. For example, these neural networks are configured from a convolutional neural network (CNN) like the determination unit of the insert determination section 32 . A machined surface residual image is input to the input layer of these foreign matter determination neural networks, and the presence or absence of foreign matter is output from the output layer.

The neural network of each determination unit of the foreign matter determination section 33 is trained in advance using teacher data. The teacher data includes each divided area image as input data, and includes a foreign matter presence/absence class corresponding to the image as output data. For example, for any of the determination units 33A to 33E, two classes of "foreign matter present" and "no foreign matter" are provided as output layers. Assignment of divided area images and classes corresponding thereto in the teacher data is performed by, for example, the administrator of the mold monitoring system.

The number of nodes in the input layers of the neural networks of the determination units 33A to 33E provided in the foreign matter determination section 33 may be the same as the number of pixels of the divided area images transmitted from the image extraction section 31, respectively. For example, the number of nodes in the input layer of the neural network of the C region determination unit 33C illustrated in FIG. Also, the number of nodes in the output layer of the neural network of each determination unit may be two, ie, "existing foreign matter" and "absence of foreign matter".

Further, the number of hidden layers (intermediate layers) of the neural networks of the determination units 33A to 33E of the foreign matter determination section 33 is configured to be smaller than the number of hidden layers of the neural network of the insert determination section 32. FIG. For example, any of the neural networks of the determination units 33A to 33E in the foreign matter determination section 33 is set to have fewer hidden layers than the neural networks of any of the determination units 32A to 32C in the insert determination section 32. be done.

For example, FIG. 12 illustrates the neural network of the C area determination unit 33C. The layer number n_SFC_C of the hidden layers of the C region determination unit 33C includes, for example, the layer number n_IST_PIN1 (see FIG. 10) of the hidden layers of the first pin determination unit 32A of the insert determination unit 32, the second pin determination unit 32B, and the cap receiver. A neural network is constructed so that the number of layers is less than any of the hidden layers of the decision unit 32C.

All of the neural networks of the determination units 33A to 33E in the foreign matter determination section 33 are set to have fewer hidden layers than the neural networks of any of the determination units 32A to 32C in the insert determination section 32. As a result, the determination accuracy of the foreign matter determination unit 33 becomes coarser than that of the insert determination unit 32 . By roughening the judgment accuracy of the foreign matter judging section 33, in other words, making it dull for minor changes on the machined surface, slight gas burns occurring on the machined surface can be ignored and erroneously detected as foreign matter. is suppressed.

In addition, all of the neural networks of the determination units 33A to 33E in the foreign matter determination unit 33 have fewer nodes in the hidden layer than the neural networks of any of the determination units 32A to 32C in the insert determination unit 32. may be set to Even with such a method, any of the neural networks of the determination units 33A to 33E in the foreign matter determination section 33 can lower the determination accuracy than any of the determination units 32A to 32C in the insert determination section 32. Become.

Note that the determination units 33A to 33E of the foreign matter determination section 33 may have a shorter updating period of the learning teacher data than the determination units 32A to 32C of the insert determination section 32. FIG. In accordance with this, each of the determination units 33A to 33E of the foreign matter determination section 33 has a shorter learning update period than each of the determination units 32A to 32C of the insert determination section 32. FIG. At this time, the machined surface image 50 in the most recent predetermined period may be used as the teacher data to be updated.

By re-learning the foreign matter determination neural network using training data using the machined surface image 50 in the most recent predetermined period (for example, the period from the time of re-learning to one week before), gas burns on the machined surface in particular can be prevented. It is possible to suppress erroneous determination in a high-illuminance region where the change in is large.

<Mold monitoring flow>
FIG. 13 illustrates a die monitoring flow in the die monitoring system according to this embodiment. This flow is started, for example, when the operator completes mounting the insert part to the mold and presses the mold closing button to proceed to the mold closing process. As the mold clamping button is pressed, the imaging device 40 images the machined surface of the mold (S10). The machined surface image 50 (see FIG. 4) thus obtained is sent to the image extractor 31 (see FIG. 2).

The image extraction unit 31 divides the processed surface image 50 into an applied portion region image and a processed surface residual image 75 (S12). At this time, the processing surface image is obtained by cutting out the mounting portion area image from the processing surface image 50 for each mounting portion. For example, referring to FIG. 7, the image extraction unit 31 extracts three types of applied part area images, a first pin image 71, a second pin image 72, and a cap receiving image 73, from the processed surface image 50 as applied part area images. cut out. The clipped attachment area image is sent to the insert determination section 32 .

The image extraction unit 31 also divides the processed surface residual image 75 into a plurality of regions according to the illuminance (S14). The divided image is sent to the foreign matter determination section 33 .

The insert determination unit 32 receives the determination result of each of the determination units 32A to 32C, and at least one of the determination units outputs the determination result of improper mounting as the final value, that is, as the class with the highest degree of certainty. It is determined whether or not there is (S16). That is, the insert determination unit 32 refers to FIG. 9 and checks whether or not there is any determination unit 32A to 32C outputting a result other than "normal".

When at least one of the determination units 32A to 32C of the insert determination section 32 outputs the determination result of improper mounting as the final value, the insert determination section 32 outputs the determination result of improper mounting. A signal to that effect (hereinafter referred to as an improper mounting signal as appropriate) is transmitted to the molding apparatus control unit 34 . Upon receiving this signal, the molding device control section 34 stops the operation of the molding device and stops the transition to the mold clamping process (S18).

In addition, the improper mounting signal is transmitted from the insert determination section 32 to the display control section 19 as well. The display control unit 19 causes the display unit 20 (for example, touch panel display) to display a warning message (S24). For example, the display control unit 19 causes the display unit 20 to display the characters "NG" beside the captured photograph of the mold. In addition, the display control unit 19 may perform image processing for emphasizing the image area (for example, the first pin image 71) for which the determination result of mounting failure is output, on the captured photograph of the mold. Furthermore, the display control unit 19 causes the display unit 20 to display a confirmation button in order to restart the mold clamping process after resolving the mounting failure.

On the other hand, in step S16, when it is determined that none of the determination units 32A to 32C output a mounting error, that is, all the mounting conditions are normal, the display control unit 19 displays the determination result to that effect. send to At the same time, presence/absence of a foreign object is performed by the foreign object determination unit 33 (S20).

The foreign matter determination unit 33 confirms the determination results of the A area determination unit 33A to E area determination unit 33E, that is, the class with the highest degree of certainty. When the determination results of the A area determination units 33A to E area determination units 33A to E area determination units 33A to 33E indicate that there is no foreign matter, the foreign matter determination section 33 sends the determination result to that effect to the display control section 19. FIG.

In the insert determination section 32, any of the determination units 32A to 32C receives the determination result that the mounting is normal, and in the foreign matter determination section 33, all of the determination units 33A to 33E receive the determination result that there is no foreign matter, and the display is controlled. The unit 19 causes the display unit 20 to display an acceptance message (S22). For example, the display control unit 19 causes the display unit 20 to display the characters "OK" next to the captured photograph of the mold.

On the other hand, in step S20, when at least one of the A area determination unit 33A to E area determination unit 33E outputs the result of determination that there is a foreign object as the final value, the defective mounting signal is output to the molding apparatus control unit. 34, and the flow moves to step S18. In the subsequent step S24, the display control unit 19 may perform image processing for emphasizing the image area for which the determination result that the presence of foreign matter is output is performed on the captured photograph of the mold.

When the warning message is displayed on the display unit 20, the insert molding operator remounts the insert component determined to be improperly mounted, or removes the foreign matter that has fallen on the processing surface. The molding apparatus control unit 34 determines whether or not the confirmation button for restarting the molding process has been pressed by the operator (S26).

If the confirmation button has not been pressed by the operator, the process returns to step S26 and the flow enters a standby state. On the other hand, when the confirmation button is pushed by the operator, the molding apparatus control section 34 resumes the temporarily stopped molding apparatus, and shifts the molding process to the mold clamping process (S28).

<Another example of die monitoring flow>
FIG. 14 shows another example of the die monitoring flow by the die monitoring system according to this embodiment. In this example, steps S20 and S22 of FIG. 13 are deleted, and step S30 is provided instead of step S16.

In this step, when an abnormality is detected in at least one of the mounting portion area image and the processing surface residual image (S30), the molding device control section 34 transmits a stop command to the molding device. Such a flow may be used, for example, at the time of inspection for confirming whether or not abnormality detection by the mold monitoring system according to this embodiment is operating normally. For example, in this case, as the captured image, an image in which a foreign object is intentionally dropped on the processing surface, or an image in which an insert component is intentionally forgotten to be attached to the attachment portion is used. In such a flow as well, it is possible to determine whether the attachment is good or not and whether there is a foreign object.

10 mold monitoring device, 20 display unit, 31 image extraction unit, 32 insert determination unit, 32A first pin determination unit, 32B second pin determination unit, 32C cap reception determination unit, 33 foreign matter determination unit, 33A A region determination unit , 33B B area determination unit, 33C C area determination unit, 33D D area determination unit, 33E E area determination unit, 34 Molding apparatus control unit, 40 Imager, 42 Light source, 50 Processing surface image, 51 Cavity, 52 Cylindrical protrusion , 53 cylindrical projection, 54 rod-shaped projection, 55 parting surface, 56 first pin, 57 second pin, 58 cap receiver, 61 first insert screw, 62 second insert screw, 63 cap, 71 first pin image , 72 second pin image, 73 cap receiving image, 75 machined surface residual image, 75A to 75E machined surface residual image divided areas.

Claims (4)

A mold monitoring system for monitoring a mold for insert molding,
an imaging device that captures an image of the processing surface of the mold, in which at least one of a cavity and a core is formed and a mounting portion to which an insert component is mounted is formed;
an image extracting unit that cuts out an image of the applied part area from the processed surface image captured by the imaging device;
an insert determination unit that determines whether the insert component is mounted in the mounting portion area image;
a foreign matter determination unit that determines the presence or absence of a foreign matter in the processing surface residual image from which the mounting portion region image is cut;
with
The insert determination unit includes an insert determination neural network in which the applied portion region image is input to an input layer, and the mounting quality of the insert component is output from an output layer,
The foreign matter determination unit includes a foreign matter determination neural network in which the processing surface residual image is input to an input layer and the presence or absence of foreign matter is output from an output layer,
The number of hidden layers of the foreign matter determination neural network is smaller than that of the insert determination neural network,
Mold monitoring system. A mold monitoring system according to claim 1, wherein
The processed surface residual image is divided into a plurality of images according to the illuminance, and the foreign matter determination neural network is provided for each divided image.
Mold monitoring system. The mold monitoring system according to claim 1 or 2,
A light source for irradiating the processing surface,
wherein the light source is a red light source;
Mold monitoring system. A mold monitoring system according to any one of claims 1 to 3,
The neural network for foreign matter determination is updated with teacher data for learning,
The machined surface image in the most recent predetermined period is used as the training data,
Mold monitoring system.

Images