JP2021023982A - Soldering condition learning device, soldering condition decision device and soldering device - Google Patents

Abstract

To obtain a soldering condition learning device for learning, at each processing object to be soldered, a moving speed of a nozzle which forms a solder injection flow, an interval dimension between the processing object and a nozzle tip, and a set value of a solder injection flow height, a soldering condition decision device, and a soldering device.SOLUTION: A soldering condition learning device for learning, at each processing object to be soldered, a moving speed of a nozzle which forms a solder injection flow, an interval dimension between the processing object and a nozzle tip, and a set value of a solder injection flow height comprises: a set value input part into which the set value correlated with the processing object at each device for performing soldering is inputted; a quality determination result input part into which a quality determination result of a good product or a bad product of the processing object which is processed at the set value inputted into the set value input part is inputted; and a learning part for learning the set value being the good product at each processing object on the basis of the set value and the quality determination result.SELECTED DRAWING: Figure 1

Description

The present invention is a soldering condition learning device that learns the moving speed of a nozzle forming a solder jet, the gap between the machining target and the tip of the nozzle, and the set values of the solder jet height for each soldering target. , The soldering condition determination device and the soldering device using this.

Conventionally, some soldering devices are provided with at least one molten solder tank, molten solder supply path, and solder jet nozzle (rectifying nozzle) (see, for example, Patent Documents 1, 2 and 3). A soldering device using one jet nozzle (rectifying nozzle) may be called a point dip type. Further, a soldering device using a plurality of jet nozzles (rectifying nozzles) may be referred to as a multi-dip type. In such a soldering apparatus, as disclosed in Patent Document 3, it is common to manually input a set value to the soldering apparatus.

On the other hand, there is one in which a learning model by machine learning using AI (Artificial Intelligence) or the like is applied to a soldering apparatus (see, for example, Patent Document 4). Patent Document 4 discloses that AI is used to extract data from a solder shape or the like obtained from an image. Further, Patent Document 3 discloses that information on a substrate or the like is learned after a soldering process of a soldering apparatus.

WO2012 / 1493967 Japanese Unexamined Patent Publication No. 2008-260028 Japanese Unexamined Patent Publication No. 2008-109034 Japanese Unexamined Patent Publication No. 9-26802

However, in the conventional soldering apparatus, it is necessary to input a set value for each processing target to be soldered, and there is a problem that the yield is affected by the skill of the person who inputs.

The present invention has been made to solve the above-mentioned problems, and for each processing target to be soldered, the moving speed of the nozzle forming a jet of solder, the gap size between the processing target and the tip of the nozzle, and It is an object of the present invention to obtain a soldering condition learning device for learning a set value of a solder jet height, a soldering condition determining device using the soldering condition learning device, and a soldering device.

In the soldering condition learning device according to the present invention, the moving speed of the nozzle forming the jet of solder, the gap size between the object to be processed and the tip of the nozzle, and the jet height of the solder are used for each processing target to be soldered. This is a soldering condition learning device for learning the set value of the nozzle, and for each device to be soldered, a set value input unit for inputting the set value associated with the machining target and the set value input unit. Based on the good / bad judgment result input unit in which the good / bad judgment result of the good or bad product of the processing target processed by the set value input to is input, and the set value and the good / bad judgment result. It is characterized by being provided with a learning unit for learning the set value, which is a non-defective product for each processing target.

The soldering condition determining device according to the present invention has, for each processing target to be soldered, the moving speed of the nozzle forming the solder jet, the gap size between the machining target and the tip of the nozzle, and the solder jet height. This is a soldering condition learning device for learning the set value, and for each device to be soldered, a set value input unit for inputting the set value associated with the machining target and the set value input unit. Based on the good / bad judgment result input unit in which the good / bad judgment result of the good product or the bad product of the processing target processed by the set value input to is input, and the set value and the good / bad judgment result. It is a soldering condition determination device using the learning result of the soldering condition learning device, which is provided with a learning unit for learning the set value which is a good product for each processing target, and information on the processing target. And the soldering condition determination unit that determines the set value that the processing target is a good product for which information is input to the processing target input unit using the learning result of the learning unit. It is characterized by being equipped with.

In the soldering apparatus according to the present invention, for each processing target to be soldered, the moving speed of the nozzle forming the solder jet, the gap size between the machining target and the tip of the nozzle, and the solder jet height. A soldering condition learning device for learning set values, and for each device to be soldered, input to the set value input unit and the set value input unit associated with the machining target to input the set value. Based on the good / bad judgment result input unit in which the good / bad judgment result of the good or bad product of the processing target processed according to the set value is input, and the set value and the good / bad judgment result, the processing is performed. It is a soldering condition determining device using the learning result of the soldering condition learning device, which is provided with a learning unit for learning the set value which is a good product for each object, and information on the processing target is input. The processing target input unit to be processed and the soldering condition determination unit for determining the set value at which the processing target is a good product for which information is input to the processing target input unit are provided by using the learning result of the learning unit. A soldering device using the soldering condition determining device, wherein the soldering section of the device, the condition setting section into which the set value determined by the soldering condition determining section is input, and the soldering section. It is characterized by including a soldering control unit that controls the soldering unit so that soldering is performed under the conditions input to the condition setting unit.

According to the present invention, it is possible to obtain a soldering condition learning device capable of learning a set value which is a non-defective product for each processing target, a soldering condition determining device using the soldering condition learning device, and a soldering device.

It is a functional block diagram of the soldering condition learning apparatus which concerns on Embodiment 1 of this invention. It is a flowchart explaining operation (soldering condition learning method) of the soldering condition learning apparatus which concerns on Embodiment 1 of this invention. It is a functional block diagram of the soldering condition learning apparatus, the soldering condition determination apparatus, and the soldering apparatus which concerns on Embodiment 1 of this invention. It is a flowchart explaining operation (soldering condition determination method) of the soldering condition determination apparatus which concerns on Embodiment 1 of this invention. It is a flowchart explaining operation (soldering method) of the soldering apparatus which concerns on Embodiment 1 of this invention. It is a flowchart explaining operation (soldering method) of the soldering apparatus which concerns on Embodiment 1 of this invention. It is a functional block diagram of the soldering condition learning apparatus which concerns on Embodiment 2 of this invention. It is a flowchart explaining operation (soldering condition learning method) of the soldering condition learning apparatus which concerns on Embodiment 2 of this invention. It is a functional block diagram of the soldering condition learning apparatus which concerns on Embodiment 3 of this invention. It is a flowchart explaining operation (good and bad method) of the soldering condition learning apparatus which concerns on Embodiment 3 of this invention. It is a functional block diagram of the soldering condition learning apparatus, the soldering condition determination apparatus, and the soldering apparatus which concerns on Embodiment 3 of this invention.

Embodiment 1.
Hereinafter, the soldering condition learning apparatus according to the first embodiment of the present invention, the soldering condition determining apparatus using the soldering condition learning apparatus, and the soldering apparatus will be described with reference to FIGS. 1 to 6. In the figure, the same reference numerals indicate the same or corresponding parts, and detailed description thereof will be omitted. In FIG. 3, the processing target 1 is, for example, a substrate 1 such as a printed circuit board, and includes a single-layer substrate and a multilayer substrate. A printed wiring board, a printed wiring board, a printed circuit board, or a printed circuit board may be obtained by soldering wiring or parts (target parts) to the printed circuit board. A printed circuit board with wiring may be simply called a printed wiring board or a printed wiring board. In this case, it can be said that the processing target 1 (board 1) includes the printed wiring board and the printed wiring board. Further, it can be said that the processing target 1 (board 1) is composed of a target component for soldering and a member for which the target component is fixed by soldering. The target parts include electronic parts and analog parts, and the leads and terminals are fixed by soldering. The members to which the target component is fixed are the above-mentioned printed circuit board, printed wiring board, and printed wiring board. Leads and terminals of the target component include those that are inserted into through holes formed in the substrate 1 and fixed by soldering.

In FIGS. 1 and 3, the soldering condition learning device 2 has a setting condition input unit 3, a good / bad determination result input unit 4, and a learning unit 5. In the soldering condition learning device 2, the moving speed of the nozzle 6 (jet nozzle 6, rectifying nozzle 6) forming a jet of solder and the nozzle 6 with the machining target 1 (board 1) are set for each machining target 1 to be soldered. The set values of the gap dimension with the tip of the solder and the jet height of the solder are learned, and a learning model is constructed. The jet nozzle 6 forms a jet of solder. Further, the gap dimension between the processing target 1 (board 1) and the tip of the nozzle 6 may be information related to or included in the jet height of the solder. For example, the jet height of the solder may be substituted by the information of the gap dimension between the processing target 1 (the substrate 1) and the tip of the nozzle 6. Further, since the rectifying nozzle 6 forms a jet on the surface of the molten solder, the rectifying nozzle 6 can also be said to be a nozzle 6 (jet nozzle 6) that forms a jet of solder. The set value input unit 3 has nozzles 6 (jet nozzle 6, rectifying nozzle 6), and is associated with the processing target 1 for each soldering device (detailed, soldering device 10 described later). Values (setting values for the moving speed of the nozzle 6 that forms a jet of solder for each processing target 1 to be soldered, the gap between the processing target 1 (board 1) and the tip of the nozzle 6, and the jet height of solder. ) Is input.

Subsequently, in FIGS. 1 and 3, the good / bad judgment result input unit 4 inputs the good / bad judgment result of the non-defective product or the defective product of the processing target 1 processed by the set value input to the set value input unit 3. It is a thing. The quality judgment result initially input in the quality / defect judgment result input unit 4 is, for example, an appearance inspection (inspection by a person or a machine) of a printed circuit board (printed circuit board) which is a processing target 1 after processing. It is a data of the result of inspection) including. That is, the result of the visual inspection (inspection including inspection by a person or a machine) may be converted into data and associated with the processed set value (set value input to the set value input unit 3) of the processing target 1. The learning unit 6 learns a set value that is a non-defective product for each processing target 1 based on the set value and the quality determination result. The learning unit 5 learns the setting value that is a good product for each processing target 1 based on the setting value input to the setting condition input unit 3 and the good / bad judgment result input to the good / bad judgment result input unit 4. Is what you do.

Machine learning such as AI may be applied to the learning unit 5 (soldering condition learning device 2). The learning unit 5 (soldering condition learning device 2) constructs and accumulates a learning model. Preferably, the learning unit 5 may learn by associating the same processing target 1 with different set values that are good products for each soldering device 10 as an error for each soldering device 10. Further, as described above, the set value input unit 3 is defined as a combination of a target component for soldering (electronic component, analog component, etc.) and a member (printed circuit board, etc.) to which the target component is fixed by soldering. The processing target 1 to be processed and the set value associated with the soldering target 1 are input.

In this case, the set value input unit 3 further has at least one of the thickness, material, number of layers, and hole diameter of the combination of the target component for soldering and the member to which the target component is fixed by soldering. The set value associated with the processing target 1 distinguished by is may be input. Similarly, the set value input unit 3 is associated with the machining target 1 in which the target part is distinguished by the outer shape among the combinations of the target part to be soldered and the member to which the target part is fixed by soldering. The set value may be input. That is, the former setting value input unit 3 is set when the target component is the same among the combinations of the target component to be soldered and the member to which the target component is fixed by soldering, and the conditions of the member are changed. Each value is entered. Further, the latter setting value input unit 3 is set when the member is the same among the combinations of the target component to be soldered and the member to which the target component is fixed by soldering, and the conditions of the target component are changed. Each value is entered.

Here, a case where the conditions of the members are changed and a case where the conditions of the target parts are changed will be described. The change in the conditions of the member means that the thickness, material, number of layers, inner layer conductor foil thickness, density, and pore diameter of the member are different as described above. The inner layer conductor foil has an inner layer copper foil thickness when the inner layer of the processing target 1 (board 1) is a copper foil. Of course, when the hole is a through hole, the depth of the hole is also a condition from the thickness of the member. It is not limited to whether or not it is a through hole, and the depth of the hole may be a condition. Specifically, when the member is the substrate 1, the thickness means the specifications of the plate thickness (conductor foil thickness, conductor foil area, number of laminated sheets). In the case of the substrate 1, the hole diameter means a through hole. Through-holes are used to form circuits, and leads and terminals of target parts are inserted. In the case of the substrate 1, the material means the base material of the substrate and the conductor type of the conductor foil. Further, changing the conditions of the target component means that the lead, the pitch of the terminal, and the diameter of the lead or the terminal are different. Since the shape and size of the target component can be known from the number of leads (terminals) and their pitches, the pitch of the leads and terminals of the target component and the diameters of the leads and terminals can be said to be information on the outer shape of the target component.

Up to now, the case where one of the combinations of the target component to be soldered and the member to which the target component is fixed by soldering is fixed has been described, but all combinations may be covered. That is, the set value input unit 3 is used for each outer shape of the target component to be soldered, and the thickness, material, number of layers, inner layer copper foil thickness (inner layer conductor foil), density, and pore diameter of the member to which the target component is fixed by soldering. The set value (the set value of the moving speed of the nozzle 6 forming the solder jet and the solder jet height) associated with the machining target 1 defined in combination with at least one of the above is input. It can be said that it is a thing. This includes at least one of the lead and terminal pitch of the target component to be soldered, the diameter of the lead and terminal, and the thickness, material, number of layers, and inner layer copper foil thickness of the member to which the target component is fixed by soldering (inner layer conductor). The above-mentioned set value associated with the processing target 1 defined in combination with at least one of the foil), the density, and the pore diameter is input. This makes it possible to deal with cases where the set value changes depending on the pitch of the leads and terminals of the target component and the difference between the through-hole diameter of the member to which the target component is fixed and the leads and terminals. When such a set value changes, it means the outer shape of the target part (the outer shape of the lead or terminal of the target part), the shape of the member to which the target part is fixed (the shape of the through hole), and the member to which the target part is fixed. It can be said that the set value changes depending on the positional relationship of.

Next, the operation of the soldering condition learning device according to the first embodiment (the soldering condition learning method according to the first embodiment) will be described with reference to FIG. In FIG. 2, in step 1, the set value (soldering process) associated with the processing target 1 is performed for each device (detailed, the soldering device 10 described later) for soldering to the set value input unit 3. This is a processing step in which the moving speed of the nozzle 6 forming the solder jet and the set value of the solder jet height) are input for each object 1. Step 2 is a processing step in which the quality / defect determination result of the non-defective product or the defective product of the processing target 1 processed by the set value input to the set value input unit 3 is input to the quality / defect determination result input unit 4. In steps 1 and 2, the order of processing does not matter. It may be at the same time. In step 3, the learning unit 5 is set to be a good product for each processing target 1 based on the set value input to the setting condition input unit 3 and the good / bad judgment result input to the good / bad judgment result input unit 4. This is a processing step for learning the value. Preferably, step 3 incorporates a processing step in which the learning unit 5 learns by associating each soldering device 10 with different set values that are good products even for the same processing target 1 as an error for each soldering device 10. It may be.

In the operation of the soldering condition learning device according to the first embodiment (the soldering condition learning method according to the first embodiment), as in the soldering condition learning device according to the first embodiment, preferably, step 1 is performed. , In the step where the set value is input to the set value input unit 3, which is associated with the machining target 1 defined by the combination of the target part to be soldered and the member to which the target part is fixed by soldering. is there. The conditions of the target component, the conditions of the members, and their combinations are the same as those of the soldering condition learning device according to the first embodiment.

In FIG. 3, the soldering condition determining device 7 uses the learning result (learning model) of the soldering condition learning device 2 shown in FIGS. 1 and 3 (the soldering condition determining device according to the first embodiment). ). The soldering condition determining device 7 has a processing target input unit 8 and a soldering condition determining unit 9. In FIG. 3, the processing target input unit 8 inputs information on the processing target 1 that is the target to be newly processed (soldered). The soldering condition determination unit 9 uses the learning result (learning model) of the learning unit 5 to determine a set value at which the processing target 1 whose information is input to the processing target input unit 8 is a good product.

Next, the operation of the soldering condition determining device according to the first embodiment (the soldering condition determining method according to the first embodiment) will be described with reference to FIG. In FIG. 4, step 11 is a processing step in which information on the processing target 1 to be newly processed (soldered) is input to the processing target input unit 8. Step 12 is a processing step in which the soldering condition determination unit 9 uses the learning result (learning model) of the learning unit 5. In step 13, the soldering condition determination unit 9 uses the learning result (learning model) of the learning unit 5 to determine a set value at which the processing target 1 whose information is input to the processing target input unit 8 is a good product. Is.

In FIG. 3, the soldering device 10 uses the soldering condition determining device 7 shown in FIG. 3 (the soldering device according to the first embodiment). The soldering device 10 includes a soldering unit 6, a condition setting unit 11, and a soldering control unit 12. In FIG. 3, the soldering portion 6 is formed with nozzles 6 (jet nozzle 6, rectifying nozzle 6). Although not shown, the soldering unit 6 includes a molten solder tank, a molten solder supply path, and the like in addition to the nozzle 6. The condition setting unit 11 is for inputting the set value determined by the soldering condition determination unit 7. The soldering control unit 12 controls the soldering unit 6 so as to perform soldering under the conditions input to the condition setting unit 11. The soldering control unit 12 can process (solder) a new processing target 1 based on the moving speed of the nozzle 6 forming the solder jet and the solder jet height determined by the set value. In addition, the soldering unit 6 is controlled.

Next, the operation of the soldering apparatus according to the first embodiment (the soldering method according to the first embodiment) will be described with reference to FIG. In FIG. 4, step 21 is a processing step in which the set value determined by the soldering condition determination unit 7 is input to the condition setting unit 11. Step 24 is a processing step of controlling the soldering unit 6 so that the soldering control unit 12 solders under the conditions input to the condition setting unit 11. That is, in step 24, the soldering portion 6 is made to process a new processing target 1 (soldering) based on the moving speed of the nozzle 6 forming the solder jet and the height of the solder jet, which are determined by the set values. This is a processing step.

Soldering condition learning device according to the first embodiment, soldering condition determining device and soldering device using the same (soldering condition learning method according to the first embodiment, soldering condition determining method using the same, and soldering). Method), as mentioned in the explanation of the soldering condition learning device (soldering condition learning method), the learning unit 5 (in step 3) considers the same processing target 1 to be a good product for each soldering device 10. You may learn by associating those having different set values as an error for each soldering device 10. The error for each soldering device 10 can be used in the soldering condition determining device and the soldering device (the soldering condition determining method and the soldering method according to the first embodiment) as follows.

First, in the soldering condition determination device (step 12), the learning result of the learning unit 5 is used to determine a set value at which the processing target 1 whose information is input to the processing target input unit 8 is a good product, and the soldering condition is determined. The determination unit 9 uses the learning result of the learning unit 5 to identify the presence or absence of an error associated with the determined set value. Then, as shown in FIG. 6, in the soldering apparatus, the soldering control unit 12 has an error associated with the condition (process of step 21) input to the condition setting unit 11, and the condition setting unit 11 corrects the error. When completed (process in step 22 is YES), the soldering unit 6 is controlled to perform soldering (process in step 24) using a set value that serves as a reference value after correcting the error (process in step 23). ). On the other hand, the soldering control unit 12 is input to the condition setting unit 11 even if there is an error associated with the condition (process of step 21) input to the condition setting unit 11 (process of step 22 is NO). The soldering unit 6 is controlled so as to perform soldering under the conditions (process in step 24).

When going directly from step 22 to step 24, if there is no error associated with the condition (process of step 21) input to the condition setting unit 11, the process of the flowchart shown in FIG. 5 is performed. That is, when the error is not corrected by the condition setting unit 11, it is directly from step 22 (step 21) regardless of the presence or absence of the error associated with the condition (process of step 21) input to the condition setting unit 11. You will go to step 24.

Further, the set value serving as the reference value corrected for the error may be individually prepared, or the learning result (learning model) of the learning unit 5 may be used. That is, the learning unit 5 of the soldering condition learning device learns by associating each soldering device 10 with different set values that are good products even for the same processing target 1 as an error for each device, and corrects the error. Determine the setting value to be the value. Then, the soldering control unit 12 of the soldering device 10 has an error associated with the condition input to the condition setting unit 11, and when the condition setting unit 11 has corrected the error, the reference value determined by the learning unit 5 is used. The soldering unit 6 is controlled so as to perform soldering by using the set value (the set value as the reference value may be acquired via the soldering condition determining device 7).

As described above, the soldering condition learning device according to the first embodiment, the soldering condition determining device and the soldering device using the soldering condition learning device (the soldering condition learning method according to the first embodiment, the soldering condition determining method using the same, and the soldering condition determining method). The soldering method) allows learning and use of set values that are good products for each processing target. The set value is the moving speed of the nozzle 6 forming the solder jet and the solder jet height for each processing target 1 to be soldered.

Embodiment 2.
Hereinafter, the soldering condition learning device according to the second embodiment of the present invention, the soldering condition determining device using the soldering condition learning device, and the soldering device will be described with reference to FIGS. 7 and 8. The description will be focused on the parts different from those of the first embodiment. In the figure, the same reference numerals indicate the same or corresponding parts, and detailed description thereof will be omitted. The major difference between the first embodiment and the second embodiment is that the learning unit 5 learns by associating the set value that is a good product for each processing target 1 with the processing target image data (photograph of the soldered portion, etc.). Is. There is no major difference in the soldering apparatus between the first embodiment and the second embodiment.

In FIG. 7, the good / bad determination result input unit 4 inputs the good / bad judgment result of the non-defective product or the defective product of the processing target 1 processed by the set value input to the set value input unit 3. Further, the quality / defect determination result input unit 4 inputs the processing target image data (photograph of the soldering portion, etc.) of the processing target associated with the quality / defect determination result. The learning unit 5 learns the setting value that is a good product for each processing target 1 based on the setting value input to the setting condition input unit 3 and the good / bad judgment result input to the good / bad judgment result input unit 4. Is what you do. Further, the learning unit 5 learns by associating the set value, which is a non-defective product, with the image data to be processed for each processing target 1.

As described above, since the soldering condition learning device according to the second embodiment associates the set value that is a good product for each processing target 1 with the image data to be processed, the determined processing target 1 is a good product. Since the image data to be processed corresponding to the set value can be easily specified, it is possible to easily obtain the image at the time of completion when the determined image to be processed is processed (soldered) with the set value at which the processing target 1 is a non-defective product. Can be done. The image data to be processed may be determined by determining the learning result (learning model) of the learning unit 5 when the soldering condition determining unit 9 determines the set value. The soldering condition determination device 7 may include a display unit that displays (determined) image data to be processed obtained from the learning result (learning model) of the learning unit 5, including the outside.

Next, the operation of the soldering condition learning device according to the second embodiment (the soldering condition learning method according to the second embodiment) will be described with reference to FIG. In FIG. 8, in step 1, the set value (soldering process) associated with the processing target 1 is performed for each device (more specifically, the soldering device 10 described later) for soldering to the set value input unit 3. In the processing step in which the moving speed of the nozzle 6 forming the solder jet, the gap size between the machining target 1 (board 1) and the tip of the nozzle 6 and the set value of the solder jet height) are input for each target 1. is there. In step 1 and the next step 2A, the order of processing does not matter. It may be at the same time.

In FIG. 8, step 2A is a process in which the good / bad judgment result of the good product or the defective product of the processing target 1 processed by the set value input to the set value input unit 3 is input to the good / bad judgment result input unit 4. A step and a processing step in which the image data of the processing target (photograph of the soldering portion, etc.) of the processing target associated with the quality determination result is input to the quality / defect determination result input unit 4. In step 3A, the learning unit 5 is set to be a good product for each processing target 1 based on the set value input to the setting condition input unit 3 and the good / bad judgment result input to the good / bad judgment result input unit 4. A processing step for learning the value, and a processing step for causing the learning unit 5 to learn by associating the set value that is a good product for each processing target 1 with the processing target image data. Similar to step 3 of the first embodiment, in step 3A, in the learning unit 5, each soldering device 10 has a different set value that is a good product even if the same processing target 1 is used, as an error for each soldering device 10. A processing step to be associated and learned may be incorporated.

As described above, the soldering condition learning device according to the second embodiment, the soldering condition determining device and the soldering device using the soldering condition learning device (the soldering condition learning method according to the first embodiment, the soldering condition determining method using the soldering condition learning device, and the soldering condition determining method using the same). The soldering method) allows learning and use of set values that are good products for each processing target. Furthermore, it is also possible to obtain image data to be processed that has actually been processed (soldered) in the past.

Embodiment 3.
Hereinafter, the soldering condition learning device according to the third embodiment of the present invention, the soldering condition determining device using the soldering condition learning device, and the soldering device will be described with reference to FIGS. 9, 10, and 11. The description will be focused on the parts different from the first and second embodiments. In the figure, the same reference numerals indicate the same or corresponding parts, and detailed description thereof will be omitted. The major difference between the second embodiment and the third embodiment is that in the second embodiment, the learning unit 5 determines the set value and the image data to be processed (such as a photograph of the soldered portion) that are good products for each processing target 1. However, in the third embodiment, the new processing target image data input unit 13 is further provided. There are no major differences in the soldering apparatus between the first embodiment, the second embodiment, and the third embodiment. There are no major differences in the soldering condition determining device between the second embodiment and the third embodiment.

In FIGS. 9 and 10, the new processing target image data input unit 13 inputs the new processing target image data of the newly soldered processing target 1. The set value input unit 3 inputs a set value associated with the processing target 1 of the new processing target image data. Using the learning result (learning model), the learning unit 5 uses the learning result (learning model) to input similar processed image data (good / bad judgment result input unit 4) from the feature amount of the new processed image data input to the new processed image data input unit 13. The quality determination result of the processing target of the new processing target image data is determined from the quality determination result associated with the determined processed image data. The good / bad judgment result input unit 4 inputs the good / bad judgment result of the good or bad product of the newly soldered processing target 1 determined by the learning unit 5 using the learning result (learning model). is there.

The learning unit 5 can aim to improve the classification accuracy by extracting the feature amount from the new processing target image data and learning in association with the feature amount of the processing target image data. That is, although not shown, it can be said that the learning unit 5 has a feature amount extraction unit (feature amount extraction function) that extracts a feature amount from the new processing target image data or the processing target image data. This feature amount extraction unit (feature amount extraction function) may be formed in the setting condition input unit 3. The feature amount extraction unit (feature amount extraction function) may be formed in the quality determination result input unit 4. In this way, machine learning such as AI in the learning unit 5 (soldering condition learning device 2) captures features by learning using a large amount of sample data and its correct answer label, and predicts and determines new data. It can be said that it does.

Next, the operation of the soldering condition learning device according to the third embodiment (the soldering condition learning method according to the third embodiment) will be described with reference to FIG. In FIG. 10, step 31 is a processing step in which the new processing target image data of the newly soldered processing target 1 is input to the new processing target image data input unit 13. Further, the set value input unit 3 is input with the set value associated with the processing target 1 of the new processing target image data. This may also be collectively referred to as step 13.

In FIG. 10, in step 32, the learning result (learning model) is used in the learning unit 5, and similar processed image data is obtained from the feature amount of the new processing target image data input to the new processing target image data input unit 13. (The one input to the good / bad judgment result input unit 4) is determined, and the good / bad judgment result of the processing target of the new processing target image data is judged from the good / bad judgment result associated with the determined processed image data. It is a processing step. In step 33, the good / bad judgment result of the good or bad product of the processing target 1 newly soldered, which is determined by the learning unit 5 using the learning result (learning model), is input to the good / bad judgment result input unit 4. It is a processing step to be performed.

The soldering condition learning device according to the third embodiment (the soldering condition learning method according to the third embodiment) is the processing target image data of the processing target associated with the good / bad judgment result in the good / bad judgment result input unit 4. Is input, and as the learning progresses by associating the set value that is a good product for each processing target with the processing target image data in the learning unit 5, the "image data (implementation)" in which the quality judgment result is not associated is not associated. In the third form, the accuracy of determining the quality determination result is improved even for the "image data referred to as" new processing target image data ")".

FIG. 11 shows the good / bad judgment result determined by the learning unit 5 for the “image data (image data referred to as“ new processing target image data ”in the third embodiment)” to which the good / bad judgment result is not associated. The soldering condition determination device and the soldering device can determine a set value and perform soldering using the set value. Of course, it is also possible to determine the set value and solder using the set value, which are the operations described in the first and second embodiments.

As described above, the soldering condition learning device according to the third embodiment, the soldering condition determining device using the soldering condition learning device, and the soldering device (the soldering condition learning method according to the first embodiment, the soldering condition determining method using the soldering condition learning device, and the soldering condition determining method using the same). The soldering method) allows learning and use of set values that are good products for each processing target. Further, it is possible to obtain the processing target image data actually processed (soldered) in the past, and it is possible to carry out a new quality determination using the processing target image data.

1 Machining target, 2 Soldering condition learning device, 3 Setting condition input unit,
4 Good / bad judgment result input unit, 5 learning unit, 6 nozzles (jet nozzle, rectifying nozzle),
7 Soldering condition determination device, 8 Processing target input unit, 9 Soldering condition determination unit,
10 Soldering device, 11 Condition setting unit, 12 Soldering control unit,
13 Newly processed image data input unit.

Claims (14)

Learning conditions for soldering to learn the moving speed of the nozzle that forms the jet of solder, the gap between the object to be machined and the tip of the nozzle, and the set values of the jet height of the solder for each processing target to be soldered. It ’s a device,
For each device to be soldered, a set value input unit for inputting the set value associated with the processing target and a non-defective product of the processing target processed by the set value input to the set value input unit. Alternatively, a good / bad judgment result input unit in which a good / bad judgment result of a defective product is input, and a learning unit that learns the set value which is a good product for each processing target based on the set value and the good / bad judgment result. A soldering condition learning device characterized by being equipped with. In the set value input unit, the set value associated with the machining target defined by the combination of the target part to be soldered and the member to which the target part is fixed by soldering is input. The soldering condition learning device according to claim 1. The solder according to claim 2, wherein the set value input unit inputs the set value in which the target part is associated with the machining target distinguished by the outer shape in the combination. Conditional learning device. The setting value input unit is associated with the processing target in which the member is distinguished by at least one of thickness, material, number of layers, inner layer conductor foil thickness, density, and pore diameter in the combination. The soldering condition learning device according to claim 2 or 3, wherein a value is input. The set value input unit is for each outer shape of the part to be soldered and at least one of the thickness, material, number of layers, inner layer conductor foil thickness, density, and hole diameter of the member to which the target part is fixed by soldering. The soldering condition learning apparatus according to claim 1, wherein the set value associated with the processing target defined in combination is input. Any of claims 1 to 5, wherein the learning unit learns by associating each device with different set values, which are good products even if they are the same processing target, as an error for each device. The soldering condition learning device according to item 1. Further, the processing target image data of the processing target associated with the quality determination result is input to the quality / defect determination result input unit.
The method according to any one of claims 1 to 6, wherein the learning unit further learns by associating the set value, which is a non-defective product for each processing target, with the processing target image data. Soldering condition learning device. Equipped with an image data input unit for new processing
The new processing target image data input unit is input with the newly soldered new processing target image data.
The set value input unit is input with the set value associated with the processing target of the new processing target image data.
The learning unit determines similar processed image data from the feature amount of the new processed image data input to the new processed image data input unit using the learning result, and determines the processed image. From the good / bad judgment result associated with the data, the good / bad judgment result of the processing target of the new processing target image data is determined.
The soldering according to claim 7, wherein the good / bad judgment result input unit inputs the good / bad judgment result of the good product or the defective product to be processed, which is determined by the learning unit using the learning result. Condition learning device. A soldering condition determining device using the learning result of the soldering condition learning device according to any one of claims 1 to 8.
Using the processing target input unit into which the processing target information is input and the learning result of the learning unit, solder that determines the set value in which the processing target for which information is input to the processing target input unit is a good product. A soldering condition determination device including a attachment condition determination unit. A soldering condition determining device using the learning result of the soldering condition learning device according to claim 6.
Using the processing target input unit into which the processing target information is input and the learning result of the learning unit, the setting value in which the processing target in which the information is input to the processing target input unit is a good product is determined. A soldering condition determining device including a soldering condition determining unit for specifying the presence or absence of the error associated with the determined set value. A soldering apparatus using the soldering condition determining apparatus according to claim 9 or 10.
The soldering unit of the device, the condition setting unit into which the set value determined by the soldering condition determining unit is input, and the soldering unit so as to perform soldering under the conditions input to the condition setting unit. A soldering device characterized by having a soldering control unit for controlling. A soldering apparatus using the soldering condition determining apparatus according to claim 9.
The soldering unit of the device, the condition setting unit into which the set value determined by the soldering condition determining unit is input, and the soldering unit so as to perform soldering under the conditions input to the condition setting unit. Equipped with a soldering control unit to control
The soldering control unit controls the soldering unit so as to perform soldering under the conditions input to the condition setting unit even when there is the error associated with the conditions input to the condition setting unit. A soldering device characterized by this. A soldering apparatus using the soldering condition determining apparatus according to claim 10.
The soldering unit of the device, the condition setting unit into which the set value determined by the soldering condition determining unit is input, and the soldering unit so as to perform soldering under the conditions input to the condition setting unit. Equipped with a soldering control unit to control
The soldering control unit has the error associated with the condition input to the condition setting unit, and when the condition setting unit has corrected the error, the soldering control unit uses the set value which is a reference value for correcting the error. , A soldering apparatus characterized in that the soldering portion is controlled so as to perform soldering. Learning conditions for soldering to learn the moving speed of the nozzle that forms the jet of solder, the gap between the object to be machined and the tip of the nozzle, and the set values of the jet height of the solder for each processing target to be soldered. It ’s a device,
For each device to be soldered, a set value input unit for inputting the set value associated with the processing target and a non-defective product of the processing target processed by the set value input to the set value input unit. Alternatively, a good / bad judgment result input unit in which a good / bad judgment result of a defective product is input, and a learning unit that learns the set value which is a good product for each processing target based on the set value and the good / bad judgment result. The learning unit learns by associating each device with a different set value, which is a non-defective product even if the processing target is the same, as an error for each device, and serves as a reference value corrected for the error. It is a soldering condition determining device using the learning result of the soldering condition learning device characterized by determining the set value.
Using the processing target input unit into which the processing target information is input and the learning result of the learning unit, solder that determines the set value in which the processing target for which information is input to the processing target input unit is a good product. It is a soldering device using a soldering condition determining device, which is characterized by having a soldering condition determining unit.
The soldering unit of the device, the condition setting unit into which the set value determined by the soldering condition determining unit is input, and the soldering unit so as to perform soldering under the conditions input to the condition setting unit. Equipped with a soldering control unit to control
The soldering control unit has the error associated with the condition input to the condition setting unit, and when the condition setting unit has corrected the error, soldering is performed using the set value which is the reference value. A soldering apparatus characterized in that the soldering portion is controlled so as to be performed.

Images