Processing Method and Processing Device for Assembled Components

Abstract

Embodiments of the present disclosure provide a method and an apparatus for processing assembled components. In this processing method, a set of images collected for each assembled component during the process of assembling a plurality of assembled components is obtained. Among them, the set of images includes images collected for the assembled component at the end of each of a plurality of specified processes. Qualified assembled components and unqualified assembled components among the plurality of assembled components are determined. Then, the corresponding images of the unqualified assembled components and the qualified assembled components are compared to determine the features that cause the unqualified assembled components to be unqualified.

Description

Technical Field

[0001] Embodiments of the present disclosure relate to the field of computer technologies, and in particular, to a method and an apparatus for processing an assembly. Background Art

[0002] An assembly usually consists of multiple components. In some cases, the number of components included in an assembly may be hundreds or thousands. A factory that performs assembly operations on an assembly usually does not produce components, but purchases components from suppliers of each component (therefore, in this article, a component may also be referred to as incoming material or raw material). Then, corresponding processes can be performed by multiple devices in a pipeline operation mode to assemble the components into an assembly.

[0003] In some cases, a finished product (such as a mobile phone) that can be used by an end user can be referred to as an assembly. In other cases, an assembly may not be a finished product, but a module in a finished product. For example, a display module, a communication module, etc. in a mobile phone can also be referred to as an assembly.

[0004] In production management, in order to avoid shortages in the supply of incoming materials or for considerations of the cost of incoming materials, usually the same component may be provided by at least two suppliers. The tolerances of components from different suppliers may be different. For example, a component with a higher cost may be more finely crafted, so the tolerance of this component may be smaller. The tolerance of a component can be understood as the production error of the component, such as dimensional error, color error, brightness error, etc. In addition, in order to avoid production line failures or for considerations of equipment costs, usually the same type of assembly may be assembled on at least two assembly lines. The devices that perform the same process on these assembly lines may come from different equipment manufacturers, so the tolerances of the same process performed by different devices may also be different. For example, a device with a higher cost may operate more precisely, so the tolerance of the process it performs may be smaller. The tolerance of a process can be understood as an error in assembly accuracy, such as the connection angle error when connecting two components. Summary of the Invention

[0005] Embodiments described herein provide a method and an apparatus for processing an assembly, and a computer-readable storage medium storing a computer program.

[0006] According to a first aspect of the present disclosure, a method for processing an assembly is provided. In this processing method, a set of images collected for each assembly during the process of assembling multiple assemblies is obtained. Wherein, the set of images includes images collected for the assembly at the end of each of multiple specified processes. Qualified assemblies and unqualified assemblies among the multiple assemblies are determined. Then, the corresponding images of the unqualified assemblies and the qualified assemblies are compared to determine the features that cause the unqualified assemblies to be unqualified.

[0007] In some embodiments of the present disclosure, the processing method further includes: determining a component in the assembly associated with the feature; determining a first feature in the feature associated with the dimensional error of the component; determining a tolerance range of the component based on images of the defective assembly and the qualified assembly including the first feature.

[0008] In some embodiments of the present disclosure, the processing method further includes: determining a set of associated features in the feature; determining a plurality of components in the assembly associated with the set of associated features; determining a first feature in the set of associated features associated with the dimensional errors of the plurality of components; determining a tolerance range of each component in the plurality of components based on images of the defective assembly and the qualified assembly including the first feature and the costs of the plurality of components.

[0009] In some embodiments of the present disclosure, the processing method further includes: determining a component in the assembly associated with the feature; determining a second feature in the feature associated with the assembly error of the component; determining a target process among a plurality of specified processes associated with the second feature; determining a tolerance range of the target process based on images of the defective assembly and the qualified assembly including the second feature.

[0010] In some embodiments of the present disclosure, the processing method further includes: determining a set of associated features in the feature; determining a plurality of components in the assembly associated with the set of associated features; determining a second feature in the set of associated features associated with the assembly errors of the plurality of components; determining a plurality of target processes among a plurality of specified processes associated with the second feature; determining a tolerance range of each target process in the plurality of target processes based on images of the defective assembly and the qualified assembly including the second feature and the costs of the devices for performing the plurality of target processes.

[0011] In some embodiments of the present disclosure, the processing method further includes: determining a set of associated features in the feature; determining a plurality of components in the assembly associated with the set of associated features; determining a first feature in the set of associated features associated with the dimensional errors of the plurality of components; determining a second feature in the set of associated features associated with the assembly errors of the plurality of components; determining a plurality of target processes among a plurality of specified processes associated with the second feature; determining a tolerance range of each component in the plurality of components and a tolerance range of each target process in the plurality of target processes based on images of the defective assembly and the qualified assembly including the first feature and the second feature, the costs of the plurality of components, and the costs of the devices for performing the plurality of target processes.

[0012] In some embodiments of the present disclosure, the processing method further includes: determining a component in the assembly associated with the feature; determining a third feature in the feature associated with the dimension of the component; determining the dimension of the component based on images of the defective assembly and the qualified assembly including the third feature.

[0013] In some embodiments of the present disclosure, the operation of comparing corresponding images of unqualified assemblies and qualified assemblies by multiple deep learning models to determine the features that cause the unqualified assemblies to be unqualified is performed. The first deep learning model among the multiple deep learning models is configured to compare the differences between a set of images collected for the same assembly to determine the regions of interest in the images collected at the end of each process. Multiple second deep learning models among the multiple deep learning models are configured to compare the same regions of interest in the corresponding images of multiple assemblies for multiple specified processes respectively to determine the features that cause the unqualified assemblies to be unqualified.

[0014] In some embodiments of the present disclosure, the multiple second deep learning models are trained respectively via the corresponding images of multiple test assemblies for which unqualified features and qualified features have been marked.

[0015] According to a second aspect of the present disclosure, a processing device for an assembly is provided. The processing device includes at least one processor; and at least one memory storing a computer program. When the computer program is executed by the at least one processor, the processing device is caused to implement the steps of the method according to the first aspect of the present disclosure.

[0016] According to a third aspect of the present disclosure, a computer-readable storage medium storing a computer program is provided, wherein the computer program, when executed by a processor, implements the steps of the method according to the first aspect of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings of the embodiments will be briefly described below. It should be understood that the following described drawings only relate to some embodiments of the present disclosure and do not limit the present disclosure, where:

[0018] Figure 1 is an exemplary flowchart of a processing method for an assembly according to an embodiment of the present disclosure;

[0019] Figure 2 is a schematic diagram of collecting images of an assembly at the end of each of multiple specified processes;

[0020] Figure 3 is an exemplary flowchart of further steps included in the processing method according to an embodiment of the present disclosure;

[0021] Figure 4 is another exemplary flowchart of further steps included in the processing method according to an embodiment of the present disclosure;

[0022] Figure 5It is yet another exemplary flowchart of steps further included in the processing method according to an embodiment of the present disclosure;

[0023] Figure 6 It is yet another exemplary flowchart of steps further included in the processing method according to an embodiment of the present disclosure;

[0024] Figure 7 It is yet another further exemplary flowchart of steps further included in the processing method according to an embodiment of the present disclosure; and

[0025] Figure 8 It is a schematic block diagram of a processing device for an assembly according to an embodiment of the present invention.

[0026] The elements in the drawings are schematic and not drawn to scale. Detailed Description

[0027] In order to make the objectives, technical solutions, and advantages of the embodiments of the present disclosure clearer, the technical solutions of the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some, but not all, of the embodiments of the present disclosure. All other embodiments obtained by those skilled in the art based on the described embodiments of the present disclosure without creative efforts also belong to the scope of protection of the present disclosure.

[0028] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those skilled in the art to which the subject matter of the present disclosure belongs. It will be further understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the specification and the relevant art, and will not be interpreted in an idealized or overly formal form unless expressly defined otherwise herein. Terms such as "first" and "second" are only used to distinguish one component (or a part of a component) from another component (or another part of a component).

[0029] As described above, in actual production, the same components used to assemble the same type of assembled parts may come from different suppliers, so the components may have different tolerances. When the tolerance of a component is too large, it may cause the failure of the function of the assembled part. The equipment used to perform the same process (the equipment that adds the same components in the same type of assembled parts to the assembled part) may also come from different equipment manufacturers, so the process may have different tolerances. When the tolerance of the process is too large, it will also cause the failure of the function of the assembled part. Further, due to the tolerances of both the raw materials and the process itself, there may be a combined tolerance caused by the tolerances of the raw materials and the process. When the combined tolerance is too large, it will also cause the failure of the function of the assembled part. The combined tolerance of the intermediate products obtained after individual assembly processes (processes) cannot be predicted before assembly, making it difficult to ensure the quality of the finally obtained assembled parts. Especially during the development of new products, the combined impact of new mechanism designs, new materials, and new processes on the product qualification rate will be amplified, and problems with unqualified product qualification rates may be encountered. How to quickly find the causes of problems and formulate corresponding control measures and specifications is an essential process for ensuring large-scale production in the later stage. This process extremely requires experienced engineers to invest a large amount of energy in the process of new product development.

[0030] Some factories may collect some data of some processes during the assembly process and store them in, for example, an Excel spreadsheet. These data are, for example, the operating parameters and production process parameters that can be collected by machines. Analyzing a large amount of data with the help of an Excel spreadsheet is very inefficient, usually taking a long time, and usually unable to quickly and systematically associate each assembly process. In addition, these data are usually not collected for each raw material in each process, so it may not be possible to quickly trace the combined effects of each process and raw material to determine the reasons for the unqualified assembled parts.

[0031] Further, in the process of actually producing products (assembling components into assembled parts), not only the quality of the products needs to be considered, but also the production cost. To obtain the optimal combination of product quality and cost, usually a large number of experienced engineers are required to conduct systematic experiments and analyses. This usually consumes a large amount of manpower and material resources in real-world scenarios. And in some scenarios, it is even impossible to conduct sufficient experiments and analyses due to resource limitations. If it is close to the product launch, the time left for engineers to conduct experiments and analyses is even more tense.

[0032] Embodiments of the present disclosure propose a processing method for assembled parts. Figure 1 An exemplary flowchart showing a processing method 100 for assembled parts according to an embodiment of the present disclosure is shown.

[0033] In Figure 1At block S102, a set of images collected for each assembly during the process of assembling multiple assemblies is obtained. Among them, the set of images includes the images collected for the assembly at the end of each of the multiple specified processes. Figure 2 Fig. shows a schematic diagram of collecting images of an assembly at the end of each of the multiple specified processes. The multiple specified processes may be all the processes during the assembly process, or multiple consecutive processes, or multiple processes that have a key impact on the qualification rate of the assembly.

[0034] At Figure 2 Block 210 shows an image of the operating table collected before starting the assembly operation. In some embodiments of the present disclosure, an image of the operating table before starting the assembly operation may be collected for comparison with the image collected at the end of the first specified process. Although no components are shown in block 210, in actual production, the image of the operating table collected can reflect image information such as the reflection or background color on the operating table. In some other embodiments of the present disclosure, the operating table may include a substrate for carrying other components. In the example of block 210, the substrate may be white, so it is not obvious in the figure. In some other embodiments of the present disclosure, the image shown at block 210 may not be collected either.

[0035] Block 220 shows an image collected at the end of specified process I. As shown in block 220, component A is obtained in specified process I and is placed in the upper left corner of the operating table. Block 230 shows an image collected at the end of specified process II. As shown in block 230, component A' is obtained in specified process II and is placed in the upper right corner of the operating table.

[0036] Block 240 shows an image collected at the end of specified process III. As shown in block 240, component B is obtained in specified process III and is placed in the center of the operating table. Block 250 shows an image collected at the end of specified process IV. As shown in block 250, component C is obtained in specified process IV and is placed directly below component B and connected to B. Similarly, blocks 260 to 290 show the images collected at the end of specified processes V to VIII respectively.

[0037] As described above, a set of images as shown in Figure 2 is collected for each assembly. The set of images can be stored in a storage device in association with the assembly number.

[0038] Return to Figure 1, at block S104, qualified and unqualified assemblies among multiple assemblies are determined. The way to determine whether an assembly is qualified can adopt traditional detection methods. An assembly with normal function is determined as a qualified assembly, and an assembly with abnormal function is determined as an unqualified assembly. The result of whether an assembly is qualified can be stored in a storage device in association with the assembly number.

[0039] At block S106, corresponding images of unqualified assemblies and qualified assemblies are compared to determine the features that cause the unqualified assemblies to be unqualified. In some embodiments of the present disclosure, the operation of comparing corresponding images of unqualified assemblies and qualified assemblies to determine the features that cause the unqualified assemblies to be unqualified can be performed by multiple deep learning models.

[0040] The first deep learning model among multiple deep learning models can be configured to compare the differences between a set of images collected for the same assembly to determine the regions of interest in the images collected at the end of each process. In Figure 2 the example, the first deep learning model can, for example, determine the difference between the images collected for any two consecutive processes to determine the regions of interest in the images collected at the end of the subsequent process among the two consecutive processes. For example, the first deep learning model can compare the images at block 220 and block 230, and determine that the image at block 230 has an additional component A' in the upper right corner compared to the image at block 220. Therefore, it can be determined that the region where component A' is located in the image at block 230 is the region of interest. Similarly, the first deep learning model can compare the images at block 240 and block 250, and determine that the image at block 250 has an additional component C below compared to the image at block 240. Therefore, it can be determined that the region where component C is located in the image at block 250 is the region of interest.

[0041] Multiple second deep learning models among multiple deep learning models can be configured to compare the same regions of interest in the corresponding images of multiple assemblies for multiple specified processes respectively to determine the features that cause the unqualified assemblies to be unqualified. In Figure 2 the example, there can be 8 second deep learning models, which are respectively used to compare the images of each assembly at blocks 220 to 290. For example, one of the second deep learning models can compare the region of interest in the upper right corner of the image at block 230 of each assembly to determine whether the component A' in the upper right corner includes the features that cause the assembly to be unqualified. Another second deep learning model can compare the region of interest below the image at block 250 of each assembly to determine whether the component C below includes the features that cause the assembly to be unqualified.

[0042] By determining the regions of interest in the images and only comparing the features in the regions of interest, the computing time and resources of the deep learning model can be saved, and the efficiency can be improved.

[0043] The above deep learning model can be, for example, a Mask R-CNN model, a B-MR-CNN model, a PointRend model, a Mask Transfiner model, a BPR model, a RefineMask model, a BCNet model, etc.

[0044] In some embodiments of the present disclosure, multiple second deep learning models can be trained respectively via corresponding images of multiple test assemblies for which unqualified features and qualified features have been marked. For example, images such as Figure 2 shown in frame 210 to frame 290 can be collected in advance for multiple test assemblies. The test assemblies can be the same as the actually produced assemblies, for example, the assemblies produced in the pre-production trial. It can be determined whether each test assembly is qualified. Then, for each feature of each component in each test assembly, it can be marked whether it is a qualified feature or an unqualified feature for the images shown in frame 220 to frame 290. After that, the same second deep learning model can be trained with the marked images corresponding to the same process. For example, one second deep learning model can be trained with the marked images such as those shown in frame 220 collected for each test assembly. Another second deep learning model can be trained with the marked images such as those shown in frame 230 collected for each test assembly. And so on.

[0045] The trained deep learning model can identify the difference between qualified assemblies and unqualified assemblies and determine the features that cause the unqualified assemblies to be unqualified.

[0046] In some embodiments of the present disclosure, after determining the features that cause the unqualified assemblies to be unqualified, the manufacturing scheme of the assemblies can be adjusted based on these features to improve the product qualification rate and / or reduce costs. Figures 3 to 7 An exemplary flowchart showing the further steps included in the processing method according to an embodiment of the present disclosure is shown. The following will respectively refer to Figures 3 to 7 the steps shown to introduce how to improve the product qualification rate and / or reduce costs. Figures 3 to 7 The steps shown can be executed Figure 1 after the box S106 in

[0047] Hereinafter, it can be assumed that in the Figure 2 example, the situations that cause some assemblies to be unqualified include:

[0048] (1) The top protrusion of component B is too wide, hindering the passage of wireless signals;

[0049] (2) The size mismatch between component B and component C causes poor contact;

[0050] (3) The placement angle of component A is inaccurate;

[0051] (4) Component E and component E' are not aligned;

[0052] (5) The distance between component D and component C is too close (it may be that the length of component D is too long, or the placement positions of components C and D are inaccurate).

[0053] In Figure 3 In the embodiment of, at block S302, the components in the assembly associated with the features that cause the non - conforming assembly to be non - conforming are determined. In the above - assumed case, by comparing the corresponding images of the conforming assembly and the non - conforming assembly, the components in the assembly associated with the features that cause the non - conforming assembly to be non - conforming are component A, component B, component C, component D, component E, and component E'.

[0054] At block S304, the first feature associated with the dimensional error of the component in this feature is determined. In the above example, the components associated with the features that cause the non - conforming assembly to be non - conforming are component A, component B, component C, component D, component E, and component E'. The first feature associated with the dimensional error of the component in the features that cause the non - conforming assembly to be non - conforming may include: the width of the top protrusion of component B, the width of the bottom protrusion of component B, the width of the groove of component C, and the size of component D. In the above example, assume that the standard value of the width of the top protrusion of component B of the non - conforming assembly is 10 mm, while its actual value is 10.5 mm, resulting in the weakening of the wireless signal passing through its two sides. It can be determined that the dimensional error of the width of the top protrusion of component B is the reason for the non - conformity of this assembly. Assume that the standard value of the width of the bottom protrusion of component B of the non - conforming assembly is 10 mm, while its actual value is 10.5 mm, resulting in its inability to be inserted into the groove of component C with a standard width of 10.1 mm. It can be determined that the dimensional error of the width of the bottom protrusion of component B is the reason for the non - conformity of this assembly. Assume that the standard value of the width of the groove of component C of the non - conforming assembly is 10.1 mm, while its actual value is 9.5 mm, resulting in its inability to accommodate the protrusion of component B with a standard width of 10 mm. Then it can be determined that the dimensional error of the width of the groove of component C is the reason for the non - conformity of this assembly. The standard value of the distance between component D and component C of the non - conforming assembly is 2.5 mm, while its actual value is 2.0 mm. This is because the length of component D exceeds the length standard value. Therefore, it can be determined that the dimensional error of the length of component D is the reason for the non - conformity of this assembly.

[0055] At block S306, the tolerance range of the component is determined based on the images of the unqualified assembly and the qualified assembly including the first feature. In the above example, the width of the top protrusion of component B of the qualified assembly and the unqualified assembly can be compared through the image at block 240. For example, the width of the top protrusion of component B in the qualified assembly is between 9.7 mm and 10.3 mm, while the minimum width of the top protrusion of component B in the unqualified assembly is 10.4 mm. Then, it can be determined that the tolerance range of the width of the top protrusion of component B can be ±0.3 mm. In other words, when the width of the top protrusion of component B is less than 10.3 mm, it can be ensured that it does not affect the performance of the assembly. Therefore, when purchasing component B, for example, component B with a tolerance range of ±0.3 mm should be selected.

[0056] Similarly to the method of determining the tolerance range of the width of the top protrusion of component B, the tolerance range of the width of the bottom protrusion of component B, the tolerance range of the width of the groove of component C, and the tolerance range of the length of component D can be determined.

[0057] In the above example, the size mismatch between component B and component C results in poor contact. In some embodiments of the present disclosure, the tolerance ranges of both can be determined by comprehensively considering the sizes of component B and component C. Figure 4 A processing flow for this situation is shown.

[0058] At block S402, a set of associated features among the features that cause the unqualified assembly to be unqualified is determined. In the above example, it can be determined that the size matching between component B and component C is a set of associated features (the first set of associated features). The position matching between component E and component E' is a set of associated features (the second set of associated features). The relative position between component C and component D and the length of component D are a set of associated features (the third set of associated features).

[0059] At block S404, multiple components in the assembly associated with this set of associated features are determined. In the above assumed example, it can be determined that the multiple components associated with the first set of associated features are component B and component C, the multiple components associated with the second set of associated features are component E and component E', and the multiple components associated with the third set of associated features are component C and component D.

[0060] At block S406, a first feature associated with the dimensional errors of the multiple components in this set of associated features is determined. In the above assumed example, in the first set of associated features, the first feature associated with the dimensional errors of the components is the width of the bottom protrusion of component B and the width of the groove of component C. The second set of associated features does not involve a first feature associated with the dimensional errors of the components. In the third set of associated features, the first feature associated with the dimensional errors of the components is the length of component D.

[0061] At block S408, the tolerance range for each of the plurality of components is determined based on the images of the non-conforming assemblies and the conforming assemblies including the first feature, as well as the costs of the plurality of components. In the above assumed example, for the first set of associated features, the tolerance ranges of component B and component C can be determined based on the images of the conforming assembly and the non-conforming assembly at block 250. In an example where the standard value of the width of the bottom protrusion of component B is 10 mm and the standard value of the width of the groove of component C is 10.1 mm, if the tolerance range of the width of the bottom protrusion of component B is ±0.05 mm, then the tolerance range of the width of the groove of component C can be ±0.05 mm. If the tolerance range of the width of the bottom protrusion of component B is ±0.03 mm, then the tolerance range of the width of the groove of component C can be ±0.07 mm. Assume that the price of component B with a tolerance range of ±0.05 mm is 10 yuan cheaper than the price of component B with a tolerance range of ±0.03 mm, and the price of component C with a tolerance range of ±0.05 mm is 2 yuan more expensive than the price of component C with a tolerance range of ±0.07 mm. Then, based on the costs of components B and C, the tolerance range of component B can be set to ±0.05 mm, and the tolerance range of component C can be set to ±0.05 mm. That is, purchasing component B with a tolerance range of ±0.05 mm and component C with a tolerance range of ±0.05 mm can avoid the dimensional mismatch of components B and C and can also save the cost of incoming materials.

[0062] In some embodiments of the present disclosure, the errors in the manufacturing process may cause assembly errors of the components, which may in turn cause the assemblies to be non-conforming. Figure 5 A processing flow for this situation is shown.

[0063] At Figure 5 block S502, the components in the assembly that are associated with the feature that causes the non-conforming assembly to be non-conforming are determined. In the above assumed situation, by comparing the corresponding images of the conforming assembly and the non-conforming assembly, it can be determined that the components in the assembly that are associated with the feature that causes the non-conforming assembly to be non-conforming are component A, component B, component C, component D, component E, and component E'.

[0064] At block S504, a second feature associated with the assembly error of the component in the feature is determined. In the above example, the components associated with the feature that causes the non-conforming assembly to be non-conforming are component A, component B, component C, component D, component E, and component E'. The second features associated with the assembly error of the component in the feature that causes the non-conforming assembly to be non-conforming may include: the placement angle of component A is incorrect, component E and component E' are not aligned, and the placement positions of component D and component C are incorrect.

[0065] At block S506, the target process associated with the second feature in the plurality of specified processes is determined. In the above example, referring to Figure 2It can be known that the target processes associated with the second feature may include: Process I associated with the inaccurate placement angle of Component A, Processes VII and VIII associated with the misalignment between Component E and Component E', and Processes IV and V associated with the inaccurate placement positions of Component D and Component C.

[0066] At block S508, the tolerance range of the target process is determined based on the images of the non-conforming assembly and the conforming assembly including the second feature. In the above assumed example, the placement angles of Component A of the conforming assembly and the non-conforming assembly can be compared through the images at block 220. For example, the standard placement angle of Component A should be that the long side of Component A is parallel to the y-axis. During the actual assembly process, it is possible that the placement angle of Component A has deviated. The angles between the long side of Component A and the y-axis in the conforming assembly are all within ±2°, while the angles between the long side of Component A and the y-axis in the non-conforming assembly all exceed the range of ±2°. Then it can be determined that when the angle between the long side of Component A and the y-axis is within ±2°, it will not affect the efficacy of the assembly. In other words, the tolerance range of Process I for assembling Component A can be ±2°. Therefore, when purchasing equipment for executing Process I, for example, equipment with a tolerance range of ±2° should be selected.

[0067] Similarly to the method of determining the tolerance range of the equipment for executing Process I, the tolerance ranges of the equipment for executing Process IV, the equipment for executing Process V, the equipment for executing Process VII, and the equipment for executing Process VIII can be determined.

[0068] In the above assumed example, the misalignment between Component E and Component E' results in a non-conforming assembly. In some embodiments of the present disclosure, the tolerance ranges of the equipment for executing Process VII and the equipment for executing Process VIII can be determined by comprehensively considering the positions of Component E and Component E'. Figure 6 A processing flow for this situation is shown.

[0069] At block S602, a set of associated features among the features that cause the non-conforming assembly to be non-conforming is determined. In the above example, it can be determined that the size matching between Component B and Component C is a set of associated features (the first set of associated features). The position matching between Component E and Component E' is a set of associated features (the second set of associated features). The relative position between Component C and Component D and the length of Component D are a set of associated features (the third set of associated features).

[0070] At block S604, multiple components in the assembly associated with this set of associated features are determined. In the above assumed example, it can be determined that the multiple components associated with the first set of associated features are Component B and Component C, the multiple components associated with the second set of associated features are Component E and Component E', and the multiple components associated with the third set of associated features are Component C and Component D.

[0071] At block S606, a second feature associated with the assembly errors of multiple components among the set of associated features is determined. In the above hypothetical example, the second feature associated with the assembly errors of multiple components in the second set of associated features is the placement of component E and component E'. The second feature associated with the assembly errors of multiple components in the third set of associated features is the placement of component C and component D. The first set of associated features does not involve the second feature associated with the assembly errors of components.

[0072] At block S608, multiple target processes associated with the second feature among multiple specified processes are determined. In the above hypothetical example, the placement of component E and component E' is performed by processes VII and VIII. Thus, processes VII and VIII can be determined as the target processes. The placement of component C and component D is performed by processes IV and V. Thus, processes IV and V can be determined as the target processes.

[0073] At block S610, according to the images of the non-conforming assembled parts and the conforming assembled parts including the second feature and the costs of the devices for performing multiple target processes, the tolerance range of each target process among the multiple target processes is determined. In the above hypothetical example, for the second set of associated features, the tolerance ranges of processes VII and VIII can be determined according to the images of the conforming assembled parts and the non-conforming assembled parts at block 290. Assume that if the tolerance range of the device for performing process VII is ±4°, then the tolerance range of the device for performing process VIII needs to be ±1° so that the placement of component E and component E' does not affect the efficacy of the assembled part. If the tolerance range of the device for performing process VII is ±3°, then the tolerance range of the device for performing process VIII can be ±2°. Assume that the device for performing process VII with a tolerance range of ±4° is 100,000 cheaper than the device for performing process VII with a tolerance range of ±3°, and the device for performing process VIII with a tolerance range of ±1° is 200,000 more expensive than the device for performing process VIII with a tolerance range of ±2°. Then, according to the costs of the device for performing process VII and the device for performing process VIII, the tolerance range of the device for performing process VII can be set to ±3°, and the tolerance range of the device for performing process VIII can be set to ±2°. In this way, it can not only make the placement of component E and component E' not affect the efficacy of the assembled part, but also save the cost of the device.

[0074] Similar to the way of determining the tolerance ranges of the device for performing process VII and the device for performing process VIII, the tolerance ranges of the device for performing process IV and the device for performing process V can be determined.

[0075] In some cases, the combination of the dimensional errors and the assembly errors of components may cause the assembled parts to be non-conforming. Figure 7 A processing flow for this situation is shown.

[0076] At block S702, a set of associated features among the features that cause the non - conforming assembled part to be non - conforming is determined. In the above example, it can be determined that the dimensional matching between part B and part C is a set of associated features (the first set of associated features). The positional matching between part E and part E' is a set of associated features (the second set of associated features). The relative position between part C and part D and the length of part D are a set of associated features (the third set of associated features).

[0077] At block S704, a plurality of parts in the assembled part that are associated with the set of associated features are determined. In the above assumed example, it can be determined that the plurality of parts associated with the first set of associated features are part B and part C, the plurality of parts associated with the second set of associated features are part E and part E', and the plurality of parts associated with the third set of associated features are part C and part D.

[0078] At block S706, a first feature in the set of associated features that is associated with the dimensional errors of the plurality of parts is determined. In the above assumed example, the first feature in the first set of associated features that is associated with the dimensional errors of the plurality of parts is the width of the bottom protrusion of part B and the width of the groove of part C. The second set of associated features does not involve a first feature associated with the dimensional errors of the parts. The first feature in the third set of associated features that is associated with the dimensional errors of the parts is the length of part D.

[0079] At block S708, a second feature in the set of associated features that is associated with the assembly errors of the plurality of parts is determined. In the above assumed example, the second feature in the second set of associated features that is associated with the assembly errors of the plurality of parts is the placement of part E and part E'. The second feature in the third set of associated features that is associated with the assembly errors of the plurality of parts is the placement of part C and part D. The first set of associated features does not involve a second feature associated with the assembly errors of the parts.

[0080] At block S710, a plurality of target processes in the plurality of specified processes that are associated with the second feature are determined. In the above assumed example, the assembly of part E and part E' in the second set of associated features is performed by processes VII and VIII. The assembly of part C and part D in the third set of associated features is performed by processes IV and V.

[0081] At block S712, based on the images of the unqualified and qualified assemblies including the first feature and the images including the second feature, the costs of multiple components, and the costs of the equipment for performing multiple target processes, the tolerance range of each component among the multiple components and the tolerance range of each target process among the multiple target processes are determined. In the above assumed example, the third set of associated features includes both the first feature and the second feature. For the third set of associated features, the tolerance range of the length of component D and the tolerance ranges of processes IV and V can be determined according to the images of the qualified and unqualified assemblies at block 260. Specifically, if the tolerance range of component D is small, its cost is high, but the requirements for the tolerance ranges of processes IV and V are low. Thus, the cost of the equipment for performing processes IV and V is low. If the tolerance range of component D is large, its cost is low, but the requirements for the tolerance ranges of processes IV and V are high. Thus, the cost of the equipment for performing processes IV and V is high. The tolerance range of component D and the tolerance ranges of processes IV and V can be determined by comprehensively considering the cost of component D and the cost of the equipment for performing processes IV and V. In this way, both the qualification rate of the assembly can be ensured and the comprehensive cost can be saved.

[0082] In some embodiments of the present disclosure, the size of a component can also be selected with the aid of the corresponding images of the unqualified and qualified assemblies. In this case, the component associated with the feature that causes the unqualified assembly to be unqualified in the assembly can be determined. Then, the third feature associated with the size of the component in this feature is determined. Next, the size of the component is determined according to the images of the unqualified and qualified assemblies including the third feature. For example, in the Figure 2 example, if it is not determined during the new product R & D stage how much the area of component A should be, several candidate components A with different areas can be tried. Then, the area of component A is determined according to the qualification rates of the assemblies in these cases.

[0083] Through the processing method of the embodiments of the present disclosure, not only the tolerance range of the size of a component can be determined, but also the size of the component can be determined.

[0084] Figure 8 FIG. shows a schematic block diagram of a processing device 800 for an assembly according to an embodiment of the present disclosure. As Figure 8 shown, the device 800 may include a processor 810 and a memory 820 storing a computer program. When the computer program is executed by the processor 810, the device 800 can be made to execute as Figure 1Steps of the method 100 shown. In one example, the device 800 can be a computer device or a cloud computing node. The device 800 can obtain a set of images collected for each assembly during the process of assembling multiple assemblies. Among them, the set of images includes the images collected for the assembly at the end of each of the multiple specified processes. The device 800 can determine the qualified assemblies and unqualified assemblies among the multiple assemblies. Then, the device 800 can compare the corresponding images of the unqualified assemblies and the qualified assemblies to determine the features that cause the unqualified assemblies to be unqualified.

[0085] In some embodiments of the present disclosure, the device 800 can determine the components in the assembly associated with the feature. The device 800 can determine a first feature in the feature associated with the dimensional error of the component. The device 800 can determine the tolerance range of the component based on the images of the unqualified assembly and the qualified assembly including the first feature.

[0086] In some embodiments of the present disclosure, the device 800 can determine a set of associated features in the feature. The device 800 can determine multiple components in the assembly associated with the set of associated features. The device 800 can determine a first feature in the set of associated features associated with the dimensional errors of the multiple components. The device 800 can determine the tolerance range of each of the multiple components based on the images of the unqualified assembly and the qualified assembly including the first feature and the costs of the multiple components.

[0087] In some embodiments of the present disclosure, the device 800 can determine the components in the assembly associated with the feature. The device 800 can determine a second feature in the feature associated with the assembly error of the component. The device 800 can determine the target process among the multiple specified processes associated with the second feature. The device 800 can determine the tolerance range of the target process based on the images of the unqualified assembly and the qualified assembly including the second feature.

[0088] In some embodiments of the present disclosure, the device 800 can determine a set of associated features in the feature. The device 800 can determine multiple components in the assembly associated with the set of associated features. The device 800 can determine a second feature in the set of associated features associated with the assembly errors of the multiple components. The device 800 can determine multiple target processes among the multiple specified processes associated with the second feature. The device 800 can determine the tolerance range of each of the multiple target processes based on the images of the unqualified assembly and the qualified assembly including the second feature and the costs of the devices for performing the multiple target processes.

[0089] In some embodiments of the present disclosure, the apparatus 800 may determine a set of associated features among the features. The apparatus 800 may determine a plurality of components in the assembly associated with the set of associated features. The apparatus 800 may determine a first feature among the set of associated features associated with the dimensional errors of the plurality of components. The apparatus 800 may determine a second feature among the set of associated features associated with the assembly errors of the plurality of components. The apparatus 800 may determine a plurality of target processes among the plurality of specified processes associated with the second feature. The apparatus 800 may determine the tolerance range of each component among the plurality of components and the tolerance range of each target process among the plurality of target processes based on the images including the first feature and the images including the second feature of the non-conforming assembly and the conforming assembly, the costs of the plurality of components, and the costs of the equipment for performing the plurality of target processes.

[0090] In some embodiments of the present disclosure, the apparatus 800 may determine the components in the assembly associated with the feature. The apparatus 800 may determine a third feature among the features associated with the dimensions of the components. The apparatus 800 may determine the dimensions of the components based on the images including the third feature of the non-conforming assembly and the conforming assembly.

[0091] In an embodiment of the present disclosure, the processor 810 may be, for example, a central processing unit (CPU), a microprocessor, a digital signal processor (DSP), a processor based on a multi-core processor architecture, etc. The memory 820 may be any type of memory implemented using data storage technology, including but not limited to random access memory, read-only memory, semiconductor-based memory, flash memory, disk memory, etc.

[0092] In addition, in an embodiment of the present disclosure, the apparatus 800 may also include an input device 830, such as a camera, a keyboard, a mouse, etc., for acquiring images and numbers of the assembly. Additionally, the apparatus 800 may further include an output device 840, such as a display, etc., for outputting processing results.

[0093] In other embodiments of the present disclosure, a computer-readable storage medium storing a computer program is also provided, wherein the computer program, when executed by a processor, can implement the steps of the method as Figure 1 and Figures 3 to 7 shown.

[0094] In summary, the processing method according to the embodiments of the present disclosure can determine the characteristics of the unqualified assembly by means of the images collected for multiple specified processes, and select the components of the assembly and the equipment for performing the processes based on the characteristics. Further, the embodiments of the present disclosure also consider the combination relationship between the characteristics that cause the unqualified assembly, and select the components of the assembly and the equipment for performing the processes according to the combination relationship to improve the qualification rate of the assembly and / or reduce costs. The embodiments of the present disclosure are also helpful to improve production efficiency and reduce the consumption of human resources in the actual production process.

[0095] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of devices and methods according to multiple embodiments of the present disclosure. In this regard, each block in the flowchart or block diagram may represent a module, a segment of a program, or a part of an instruction, and the module, the segment of a program, or the part of an instruction includes one or more executable instructions for implementing the specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than marked in the accompanying drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and the combination of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0096] Unless otherwise clearly specified in the context, the singular forms of the words used in this specification and the appended claims include the plural, and vice versa. Thus, when referring to the singular, the corresponding plural is usually included. Similarly, the terms "comprising" and "including" will be interpreted as inclusive rather than exclusive. Likewise, the term "including" and "or" should be interpreted as inclusive, unless such an interpretation is explicitly prohibited in this specification. Where the term "example" is used in this specification, especially when it is located after a group of terms, the "example" is merely exemplary and illustrative, and should not be considered exclusive or extensive.

[0097] Further aspects and scopes of adaptability become apparent from the description provided herein. It should be understood that the various aspects of the present application can be implemented alone or in combination with one or more other aspects. It should also be understood that the description herein and the specific embodiments are for illustrative purposes only and are not intended to limit the scope of the present application.

[0098] The above has described several embodiments of the present disclosure in detail. However, obviously, those skilled in the art can make various modifications and variations to the embodiments of the present disclosure without departing from the spirit and scope of the present disclosure. The protection scope of the present disclosure is defined by the appended claims.

Claims

1. A processing method for assembled parts, comprising: Obtaining a set of images collected for each assembled part during the assembly of multiple assembled parts, where the set of images includes images collected for the assembled part at the end of each of multiple specified processes and before the start of the next process, and images collected for the operating table before the start of the assembly operation; Determining qualified and unqualified assembled parts among the multiple assembled parts according to whether the function of each assembled part is normal; and Performing an operation of comparing the corresponding images of the unqualified assembled parts and the qualified assembled parts by multiple deep learning models to determine the features that cause the unqualified assembled parts to be unqualified. The first deep learning model among the multiple deep learning models is configured to compare the differences between a set of images collected for the same assembled part to determine the region of interest in the images collected at the end of each process. Multiple second deep learning models among the multiple deep learning models are configured to compare the same region of interest in the corresponding images of the multiple assembled parts for the multiple specified processes respectively to determine the features that cause the unqualified assembled parts to be unqualified.

2. The processing method according to claim 1, further comprising: Determining the components in the assembled part associated with the feature; Determining a first feature among the features associated with the dimensional error of the component; Determining the tolerance range of the component according to the images of the unqualified assembled part and the qualified assembled part including the first feature.

3. The processing method according to claim 1, further comprising: Determining a set of associated features among the features; Determining multiple components in the assembled part associated with the set of associated features; Determining a first feature among the set of associated features associated with the dimensional error of the multiple components; Determining the tolerance range of each of the multiple components according to the images of the unqualified assembled part and the qualified assembled part including the first feature and the costs of the multiple components.

4. The processing method according to claim 1, further comprising: Determining the components in the assembled part associated with the feature; Determining a second feature among the features associated with the assembly error of the component; Determining the target process among the multiple specified processes associated with the second feature; Determining the tolerance range of the target process according to the images of the unqualified assembled part and the qualified assembled part including the second feature.

5. The processing method according to claim 1, further comprising: Determining a set of associated features among the features; Determining multiple components in the assembled part associated with the set of associated features; Determining a second feature among the set of associated features associated with the assembly error of the multiple components; Determining multiple target processes among the multiple specified processes associated with the second feature; Determining the tolerance range of each of the multiple target processes according to the images of the unqualified assembled part and the qualified assembled part including the second feature and the costs of the devices performing the multiple target processes.

6. The processing method according to claim 1, further comprising: Determining a set of associated features among the features; Determine multiple components in the assembly that are associated with the set of associated features; Determine a first feature in the set of associated features that is associated with the dimensional errors of the multiple components; Determine a second feature in the set of associated features that is associated with the assembly errors of the multiple components; Determine multiple target processes in the multiple specified processes that are associated with the second feature; Determine the tolerance range of each component in the multiple components and the tolerance range of each target process in the multiple target processes based on the images including the first feature and the images including the second feature of the unqualified assembly and the qualified assembly, the costs of the multiple components, and the costs of the equipment for performing the multiple target processes.

7. The processing method according to claim 1, further comprising: Determine the components in the assembly that are associated with the feature; Determine a third feature in the feature that is associated with the dimensions of the component; Determine the dimensions of the component based on the images including the third feature of the unqualified assembly and the qualified assembly.

8. The processing method according to claim 1, wherein The multiple second deep learning models are respectively trained via the corresponding images of multiple test assemblies for which unqualified features and qualified features have been marked.

9. A processing device for an assembly, comprising: At least one processor; And At least one memory storing a computer program; Wherein, when the computer program is executed by the at least one processor, the processing device is caused to execute the steps of the processing method according to any one of claims 1 to 8.

Images