An automatic acquisition method for production process parameters of fabricated components based on the Internet of Things

Abstract

The application discloses a kind of based on thing networking's fabricated component production process parameter automatic acquisition method, it is related to the field of fabricated PC component, solve the existing fabricated component production process parameter automatic acquisition method and the problem that application effect is not good, including steps S1: acquisition target preselected process, the process component that target preselected process is completed is carried out image analysis, according to the profile abnormal area and the profile lack area in image in different process parameter quantization are marked according to analysis result, obtain process component anomaly acquisition data, step S2: according to process component anomaly acquisition data acquisition single component early warning and carry out early warning, to single early warning component carries out abnormal position overlap degree analysis, according to analysis result obtains process anomaly analysis data, step S3: according to process anomaly analysis data to target preselected process carries out abnormal early warning, the application can improve fabricated component production efficiency and production quality.

Description

Technical Field

[0001] This invention belongs to the field of prefabricated PC components and relates to image processing technology. Specifically, it is an automatic acquisition method for production process parameters of prefabricated components based on the Internet of Things. Background Technology

[0002] Existing methods for automatically collecting production process parameters of prefabricated components have the following drawbacks:

[0003] 1. Existing automatic acquisition methods for prefabricated component production process parameters only count the area of ​​abnormal contours and the area of ​​missing contours separately. They do not integrate the overlap of the two types of areas to construct a unified quantitative value for contour defect overlap. They cannot simultaneously distinguish the comprehensive severity of two types of deviations: component bulge and missing edges and corners. The standard for comparing the degree of defects of components of different specifications and at different work stations is different. It is difficult to accurately divide the benchmark range of process defects based on quantitative values. Defect classification and control lack reliable numerical basis.

[0004] 2. Existing component defect detection methods can only identify surface defects such as abnormal contours and missing contours for single finished products. They only complete the quantitative marking of defects in a single component and lack the statistical analysis logic for common defect data of multiple batches of components on the production line. They cannot summarize the frequency of recurrence of the same type of defects in continuous components in the same process to issue batch defect warnings to the corresponding production line in advance.

[0005] To address this, we propose an automatic data acquisition method for prefabricated component production process parameters based on the Internet of Things (IoT). Summary of the Invention

[0006] In view of the shortcomings of existing technologies, the purpose of this invention is to provide an automatic acquisition method for production process parameters of prefabricated components based on the Internet of Things, and this invention aims to ensure high-quality production of prefabricated components.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: an automatic acquisition method for production process parameters of prefabricated components based on the Internet of Things, comprising:

[0008] Step S1: Select the target pre-selected process from the production processes included in the prefabricated component, collect process parameter quantification images of the completed process components of the target pre-selected process, and perform image analysis on the process parameter quantification images. Based on the analysis results, mark the contour abnormal areas and contour missing areas in the different process parameter quantification images to obtain process component abnormal collection data.

[0009] Step S2: Based on the abnormal data collected from the process components, screen and issue warnings for individual early warning components, perform anomaly location overlap analysis on the individual early warning components, and obtain process abnormality analysis data based on the analysis results;

[0010] Step S3: Issue anomaly warnings for the target pre-selected processes based on process anomaly analysis data.

[0011] Furthermore, in step S1, the specific steps are as follows:

[0012] Step S11: Collect the production processes included in the prefabricated components to obtain multiple component production processes, and select the target pre-selected process from the collected component production processes;

[0013] Step S12: Collect the completed process components of the target pre-selected process, and arbitrarily select a sample process component from the collected process components. Collect and detect the external dimension parameters of the sample process component, and obtain the first to sixth process parameter quantization images corresponding to the sample process component based on the detection results.

[0014] Step S13: Obtain the first to sixth process parameter quantization images corresponding to each process component to obtain process component abnormal acquisition data.

[0015] Furthermore, in step S12, the specific steps are as follows:

[0016] Step S121: Acquire images of the outer surfaces of the components included in the sample process components to obtain surface images of the first to sixth components;

[0017] Step S122: Discretize the surface image of the first component into a number of image pixels, and set a pixel traversal window in the surface image of the first component;

[0018] Step S123: Randomly select a sample pixel from the discrete image pixels of the first component surface image, control the pixel traversal window to coincide with the sample pixel, perform pixel depth analysis on the sample pixel through the pixel traversal window, divide the sample pixel into component edge pixels or non-component edge pixels according to the analysis results, and extract the contour of the sample process component based on the component edge pixels to obtain the first component contour extraction image.

[0019] Furthermore, in step S12, the specific steps are as follows:

[0020] Step S124: Collect historical components that have been completed and passed quality inspection in the target pre-selected process, extract the contour of the top view projection corresponding to the historical component, obtain the edge contour line of the projected component, and scale the edge contour line of the projected component to the same size as the workpiece contour in the first component contour extraction image.

[0021] Step S125: Use the projection component edge contour lines to cover the component contour extraction image. In the component contour extraction image, mark the closed area enclosed by the pre-connected component contour lines as the actual component contour graphic, and mark the closed area enclosed by the projection component edge contour lines as the reference component edge contour graphic.

[0022] Step S126: In the first component contour extraction image, compare the actual component contour graphic with the reference component edge contour graphic, mark the graphic area that belongs only to the actual component contour graphic and is not covered by the reference component edge contour graphic as the contour abnormal area, and mark the graphic area that belongs only to the reference component edge contour graphic and is not covered by the actual component contour graphic as the contour missing area, and obtain the first process parameter quantization image.

[0023] Step S127: Perform process parameter quantization image acquisition on the surface images of the second component to the sixth component to obtain the process parameter quantization images of the second component to the sixth component.

[0024] Furthermore, in step S123, the specific steps are as follows:

[0025] Image pixels adjacent to the pixel traversal window are collected to obtain multiple window adjacent pixels. Pixel depth is collected for sample pixels that overlap with the pixel traversal window to obtain window pixel depth values. Pixel depth is collected for each window adjacent pixel to obtain multiple adjacent pixel depth values. The absolute difference between each adjacent pixel depth value and the window pixel depth value is calculated, and the average of the obtained absolute differences is calculated to obtain the neighborhood pixel deviation corresponding to the pixel traversal window.

[0026] Set a preset value for reasonable deviation of neighboring pixels. If the neighboring pixel deviation corresponding to the pixel traversal window is greater than or equal to the preset value for reasonable deviation of neighboring pixels, then the sample pixel is marked as a component edge pixel. If the neighboring pixel deviation corresponding to the pixel traversal window is less than the preset value for reasonable deviation of neighboring pixels, then the sample pixel is marked as a non-component edge pixel.

[0027] Furthermore, in step S123, the specific steps are as follows:

[0028] The pixel traversal window is used to traverse the image pixels except for the sample pixels. Based on the traversal results, all component edge pixels in the surface image of the first component are obtained. A feature edge pixel is selected from it, and the pixel distance between the feature edge pixel and the other component edge pixels is calculated. The obtained pixel distance values ​​are then compared.

[0029] Mark the component edge pixels corresponding to the minimum pixel distance value as edge connection pixels. Connect the feature edge pixels with the edge connection pixels to obtain the pre-connected component outline. Replace the feature edge pixels with the edge connection pixels, and re-select the edge connection pixels corresponding to the feature edge pixels and connect them. Repeat this process until the pre-connected component outline is closed to obtain the first component outline extraction image.

[0030] Furthermore, in step S2, the specific steps are as follows:

[0031] Step S21: Obtain abnormal acquisition data of process components, and obtain the first to sixth process parameter quantization images corresponding to each process component based on the abnormal acquisition data of process components;

[0032] Step S22: If there is a contour anomaly region in the first process parameter quantization image to the sixth process parameter quantization image, mark the process component as a single warning component and issue a contour anomaly warning for the single warning component;

[0033] Step S23: If there is a region with missing outline, mark the process component as a single warning component and issue a missing outline warning for the single warning component;

[0034] Step S24: If both contour abnormality areas and contour missing areas exist simultaneously, mark the process component as a single warning component, and issue contour abnormality warnings and contour missing warnings for the single warning component.

[0035] Step S25: If there are no abnormal contour areas or missing contour areas, there is no need to issue an early warning for the process components;

[0036] Step S26: Perform defect region overlap analysis on the first process parameter quantification image of the single early warning component, and obtain the defect overlap index at the first location based on the analysis results;

[0037] Step S27: Perform defect region overlap analysis on the second to sixth process parameter quantization images of the single early warning component to obtain the defect overlap index from the second position to the sixth position.

[0038] Step S28: Set the defect overlap index from the first position to the defect overlap index from the sixth position as process anomaly analysis data.

[0039] Furthermore, in step S26, the specific steps are as follows:

[0040] Arbitrarily select a sample early warning component and a feature early warning component from the obtained individual early warning components. Collect the contour abnormal region of the sample early warning component in the first process parameter quantization image to obtain the first abnormal region. Collect the contour abnormal region corresponding to the feature early warning component to obtain the second abnormal region. Control the first abnormal region and the second abnormal region to coincide, and collect the area value of the overlapping region. Calculate the ratio of the obtained area value to the first abnormal region to obtain the abnormal area overlap degree between the sample early warning component and the feature early warning component.

[0041] Replace the feature warning component with each individual warning component, obtain the abnormal area overlap between the sample warning component and each individual warning component, calculate the product of the average abnormal area overlap and the area value of the first abnormal region, and obtain the abnormal overlap quantification value corresponding to the sample warning component.

[0042] Furthermore, in step S26, the specific steps are as follows:

[0043] The first missing region is obtained by collecting the contour missing region corresponding to the sample warning component, and the second missing region is obtained by collecting the contour missing region corresponding to the feature warning component. The first missing region and the second missing region are controlled to overlap, and the area value of the overlapping region is collected. The ratio of the obtained area value to the first missing region is calculated to obtain the missing area overlap degree between the sample warning component and the feature warning component.

[0044] Replace the feature warning component with each individual warning component, obtain the missing area overlap between the sample warning component and each individual warning component, calculate the product of the average missing area overlap and the area value of the first missing region, and obtain the missing overlap quantification value corresponding to the sample warning component.

[0045] The average value of the abnormal overlap quantization value and the missing overlap quantization value is used to obtain the contour defect overlap quantization value corresponding to the sample warning component;

[0046] The process of obtaining the contour defect overlap quantization value corresponding to the repeating sample early warning component involves obtaining the contour defect overlap quantization value corresponding to each individual early warning component, comparing the obtained contour defect overlap quantization values, and marking the maximum contour defect overlap quantization value as the first position defect overlap index.

[0047] Furthermore, step S3 specifically includes the following:

[0048] Obtain process anomaly analysis data, and obtain the defect overlap index from the first position to the sixth position based on the process anomaly analysis data. Compare the values ​​of the defect overlap index from the first position to the sixth position and mark the defect overlap index with the largest value as the process defect index.

[0049] Create a baseline range for the process defect index. If the process defect index is within the baseline range, there is no need to issue an anomaly warning for the target pre-selected process production line. If the process defect index is not within the baseline range, an anomaly warning is issued for the target pre-selected process production line.

[0050] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0051] 1. This invention integrates the area of ​​the abnormal contour region, the area of ​​the missing contour region, and the overlap index corresponding to the two types of regions, and generates a unified quantitative value of contour defect overlap through fusion calculation. It can simultaneously characterize the comprehensive severity of component bulge and chipped corner defects, establish a unified evaluation standard that is compatible with various specifications and inspection stations, and accurately delineate the benchmark range of process defect index based on standardized quantitative values. It provides an objective quantitative basis for defect classification and control, and solves the problems of one-sided evaluation and inconsistent comparison standards of the original single area index.

[0052] 2. Based on the quantitative identification of defects in single components, this invention adds statistical analysis logic for defects shared by multiple batches of components. It can continuously count the frequency of recurrence of the same type of defects in continuous components of the same process. Combined with the defect benchmark interval constructed by normal production data, it can judge the batch defect fluctuation status. It breaks through the limitations of traditional single-item detection and can identify abnormal trends in the early stage of concentrated batch defects. It can push batch defect warnings to the corresponding production line in advance, so that staff can adjust the production process in time to avoid batch defective products. Attached Figure Description

[0053] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0054] Figure 1 This is a diagram illustrating the implementation steps of the present invention;

[0055] Figure 2 These are six views of the process components of the present invention;

[0056] Figure 3 This is a schematic diagram of the abnormal contour region of the present invention. Detailed Implementation

[0057] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0058] Example 1

[0059] Please see Figure 1 This invention provides a technical solution: an automatic acquisition method for production process parameters of prefabricated components based on the Internet of Things, comprising the following steps:

[0060] Step S1: Select the target pre-selected process from the production processes included in the prefabricated component, collect process parameter quantification images of the completed process components of the target pre-selected process, and perform image analysis on the process parameter quantification images. Based on the analysis results, mark the contour abnormal areas and contour missing areas in the different process parameter quantification images to obtain process component abnormal collection data.

[0061] In step S1, the specific steps are as follows:

[0062] The production processes included in the prefabricated components are collected to obtain multiple component production processes, and a target pre-selected process is selected from the collected component production processes;

[0063] It should be noted here that:

[0064] In this application, the component production process involved here specifically refers to the standardized operation steps in a prefabricated component factory, from the entry of raw materials to the loading and transportation of components, which are sequential, clearly divided, and have independent process control requirements. The component production process involved here includes, but is not limited to, the material placement and casting process, the vibration molding process, and the demolding process, and the target pre-selected process involved here is the demolding process.

[0065] Collect the completed process components of the target pre-selected process, and arbitrarily select a sample process component from the collected process components. Collect and detect the external dimension parameters of the sample process component, and obtain the first to sixth process parameter quantization images corresponding to the sample process component based on the detection results.

[0066] Specifically as follows:

[0067] Images of the outer surfaces of the components included in the sample process are acquired to obtain surface images of the first to sixth components.

[0068] It should be noted here that:

[0069] Please see Figure 2In this application, the first to sixth component surface images mentioned herein correspond to the six views of the process component, the first component surface image mentioned herein is the top view of the sample process component, the second component surface image mentioned herein is the front view of the sample process component, the third component surface image mentioned herein is the left view of the sample process component, the fourth component surface image mentioned herein is the right view of the sample process component, the fifth component surface image mentioned herein is the bottom view of the sample process component, and the sixth component surface image mentioned herein is the rear view of the sample process component.

[0070] In this application, when acquiring images of the outer surface of a component, a contrast background device must be set up on the opposite side of the acquisition to form a clear visual distinction between the background and the component body, so as to facilitate the recognition of image edge contours.

[0071] Discretize the surface image of the first component into a number of image pixels, and set a pixel traversal window in the surface image of the first component;

[0072] Arbitrarily select a sample pixel from the discrete image pixels of the first component surface image, control the pixel traversal window to coincide with the sample pixel, perform pixel depth analysis on the sample pixel through the pixel traversal window, divide the sample pixel into component edge pixels or non-component edge pixels according to the analysis results, and extract the contour of the sample process component based on the component edge pixels to obtain the first component contour extraction image.

[0073] Specifically as follows:

[0074] Image pixels adjacent to the pixel traversal window are collected to obtain multiple window adjacent pixels. Pixel depth is collected for sample pixels that overlap with the pixel traversal window to obtain window pixel depth values. Pixel depth is collected for each window adjacent pixel to obtain multiple adjacent pixel depth values. The absolute difference between each adjacent pixel depth value and the window pixel depth value is calculated, and the average of the obtained absolute differences is calculated to obtain the neighborhood pixel deviation corresponding to the pixel traversal window.

[0075] Set a preset value for reasonable deviation of neighboring pixels. If the neighboring pixel deviation corresponding to the pixel traversal window is greater than or equal to the preset value for reasonable deviation of neighboring pixels, then mark the sample pixel as a component edge pixel. If the neighboring pixel deviation corresponding to the pixel traversal window is less than the preset value for reasonable deviation of neighboring pixels, then mark the sample pixel as a non-component edge pixel.

[0076] It should be noted here that:

[0077] In this application, a first component surface image is acquired for a historical process component that has completed the target pre-selection process, resulting in a first historical surface image. The neighboring pixel deviations corresponding to the pixels covered by the non-contour area of ​​the component body in the first historical surface image are acquired. The sum of the average and standard deviation of the obtained neighboring pixel deviations is calculated to obtain a preset value for reasonable neighboring pixel deviation.

[0078] The pixel traversal window is used to traverse all image pixels except for the sample pixels. Based on the traversal results, all component edge pixels in the surface image of the first component are obtained. A feature edge pixel is selected from these pixels, and the pixel distance between the feature edge pixel and the other component edge pixels is calculated. The obtained pixel distance values ​​are compared, and the component edge pixel corresponding to the smallest pixel distance value is marked as an edge connection pixel. The feature edge pixel and the edge connection pixel are connected to obtain the pre-connected component contour line. The feature edge pixel is replaced with the edge connection pixel, and the edge connection pixels corresponding to the feature edge pixel are re-selected and connected. This process is repeated until the pre-connected component contour line is closed, and the first component contour extraction image is obtained.

[0079] Collect historical components that have been completed and passed quality inspection in the target pre-selected process, extract the contour of the top view projection corresponding to the historical component, obtain the edge contour line of the projected component, and scale the edge contour line of the projected component to the same scale as the workpiece contour of the first component contour extraction image.

[0080] The component contour extraction image is covered by the projection component edge contour lines. In the component contour extraction image, the closed area enclosed by the pre-connected component contour lines is marked as the actual component contour graphic, and the closed area enclosed by the projection component edge contour lines is marked as the reference component edge contour graphic.

[0081] It should be noted here that:

[0082] When the projection component edge contour line covers the first component contour extraction image, the geometric position and shape of the actual component contour graphic are consistent with those of the reference component edge contour, and the geometric center point of the actual component contour graphic coincides with the geometric center point of the reference component edge contour graphic.

[0083] Please see Figure 3 In the first component contour extraction image, the actual component contour graphic is compared with the reference component edge contour graphic. The graphic area that belongs only to the actual component contour graphic and is not covered by the reference component edge contour graphic is marked as the contour abnormal area, and the graphic area that belongs only to the reference component edge contour graphic and is not covered by the actual component contour graphic is marked as the contour missing area, thus obtaining the first process parameter quantization image.

[0084] It should be noted here that:

[0085] In this application, the contour abnormal area is specifically the component area where the actual component contour graphic exists but the reference component edge contour graphic does not exist. It is manifested as the outer contour of the sample process component protruding outward and the size deviation. Typical phenomena include, but are not limited to, mold bulging, side protrusion, and excess slurry at the corners.

[0086] The missing outline area is a component area where the outline of the reference component exists but the outline of the actual component does not. It is manifested as the component outline being missing inward and the size being smaller. Typical phenomena include, but are not limited to, missing edges and corners, grout leakage and depression, and edge and corner damage.

[0087] Process parameter quantization image acquisition is performed on the surface images of the second component to the sixth component respectively to obtain the second process parameter quantization image to the sixth process parameter quantization image;

[0088] The first to sixth process parameter quantization images corresponding to each process component are obtained to obtain the process component abnormal acquisition data;

[0089] Step S2: Based on the abnormal data collected from the process components, screen and issue warnings for individual early warning components, perform anomaly location overlap analysis on the individual early warning components, and obtain process abnormality analysis data based on the analysis results;

[0090] In step S2, the specific steps are as follows:

[0091] Obtain abnormal data of process components, and obtain the first to sixth process parameter quantization images corresponding to each process component based on the abnormal data of process components;

[0092] If there are contour anomaly areas in the first to sixth process parameter quantization images, the process components are marked as individual early warning components, and contour anomaly early warnings are issued for the individual early warning components.

[0093] If there are areas with missing outlines, the process components will be marked as individual warning components, and a missing outline warning will be issued for the individual warning components.

[0094] If both contour abnormality areas and contour missing areas exist simultaneously, the process component will be marked as a single warning component, and contour abnormality warnings and contour missing warnings will be issued for the single warning component.

[0095] If there are no abnormal contour areas or missing contour areas, there is no need to issue an early warning for the process components;

[0096] The overlap of defect areas is analyzed by quantifying the first process parameters of the single early warning component, and the overlap index of defects at the first location is obtained based on the analysis results.

[0097] Specifically as follows:

[0098] Arbitrarily select a sample early warning component and a feature early warning component from the obtained individual early warning components. Collect the contour abnormal region of the sample early warning component in the parameter quantization image of the first process to obtain the first abnormal region. Collect the contour abnormal region corresponding to the feature early warning component to obtain the second abnormal region. Control the first abnormal region and the second abnormal region to keep them overlapping. Collect the area value of the overlapping region. Calculate the ratio of the obtained area value to the first abnormal region to obtain the abnormal area overlap degree between the sample early warning component and the feature early warning component.

[0099] Replace the feature warning component with each individual warning component, obtain the abnormal area overlap between the sample warning component and each individual warning component, calculate the product of the average abnormal area overlap and the area value of the first abnormal region, and obtain the abnormal overlap quantification value corresponding to the sample warning component.

[0100] The first missing region is obtained by collecting the contour missing region corresponding to the sample warning component, and the second missing region is obtained by collecting the contour missing region corresponding to the feature warning component. The first missing region and the second missing region are controlled to overlap, and the area value of the overlapping region is collected. The ratio of the obtained area value to the first missing region is calculated to obtain the missing area overlap degree between the sample warning component and the feature warning component.

[0101] Replace the feature warning component with each individual warning component, obtain the missing area overlap between the sample warning component and each individual warning component, calculate the product of the average missing area overlap and the area value of the first missing region, and obtain the missing overlap quantification value corresponding to the sample warning component.

[0102] The average value of the abnormal overlap quantization value and the missing overlap quantization value is used to obtain the contour defect overlap quantization value corresponding to the sample warning component;

[0103] It should be noted here that:

[0104] In this application, the contour defect overlap quantification value is calculated by comprehensively considering multiple contour defect indicators. First, the total area of ​​the contour abnormal area and the total area of ​​the contour missing area corresponding to a single warning component are calculated. Then, the area overlap of the contour abnormal area and the area overlap of the contour missing area between the actual component and the reference component are calculated. The above four indicators are weighted and fused to obtain a unified contour defect overlap quantification value. This value can simultaneously reflect the severity of two types of defects: the component contour convexity deviation and the contour missing, as well as the degree of deviation between the defect area and the standard contour, so as to achieve a unified quantitative comparison of the contour defect degree of different components.

[0105] The process of obtaining the contour defect overlap quantization value corresponding to the repeat sample early warning component involves obtaining the contour defect overlap quantization value corresponding to each individual early warning component, comparing the obtained contour defect overlap quantization values, and marking the maximum contour defect overlap quantization value as the first position defect overlap index.

[0106] Repeat the process of obtaining the first position defect overlap index, and perform defect region overlap analysis on the second to sixth process parameter quantization images of the single early warning component to obtain the second to sixth position defect overlap indices.

[0107] Set the defect overlap index from the first position to the sixth position as the process anomaly analysis data;

[0108] Step S3: Issue anomaly warnings for the target pre-selected processes based on process anomaly analysis data;

[0109] In step S3, the specific details are as follows:

[0110] Obtain process anomaly analysis data, and obtain the defect overlap index from the first position to the sixth position based on the process anomaly analysis data. Compare the values ​​of the defect overlap index from the first position to the sixth position and mark the defect overlap index with the largest value as the process defect index.

[0111] Create a benchmark range for the process defect index. If the process defect index is within the benchmark range, there is no need to issue an anomaly warning for the target pre-selected process production line. If the process defect index is not within the benchmark range, an anomaly warning is issued for the target pre-selected process production line.

[0112] It should be noted here that:

[0113] For the target pre-selected process production line, the process defect index is collected at historical moments when no abnormality warning is issued. The maximum process defect index is set as the upper limit of the process defect index benchmark interval. The upper limit of the process defect index benchmark interval is 0, which means that there are no process defects.

[0114] Compared to the problems described in the background technology, the present invention integrates the area of ​​the abnormal contour region, the area of ​​the missing contour region, and the overlap index corresponding to the two types of regions, and generates a unified quantitative value of contour defect overlap through fusion calculation. This can simultaneously characterize the comprehensive severity of component bulge and chipped corner defects, establish a unified evaluation standard that is compatible with various specifications and inspection stations, and accurately delineate the benchmark range of process defect index based on standardized quantitative values. This provides an objective quantitative basis for defect classification and control, and solves the problems of one-sided evaluation and inconsistent comparison standards of the original single area index.

[0115] Furthermore, based on the quantitative identification of defects in a single component, this invention adds statistical analysis logic for defects shared by multiple batches of components. It can continuously count the frequency of recurrence of the same type of defects in continuous components of the same process, and judge the batch defect fluctuation status by combining the defect benchmark interval constructed with normal production data. It breaks through the limitations of traditional single-item detection and can identify abnormal trends in the early stage of concentrated batch defects, and push batch defect warnings to the corresponding production line in advance, so that staff can adjust the production process in time to avoid batch defective products.

[0116] The preferred embodiments of the present invention disclosed above are only for illustrating the present invention. These preferred embodiments do not describe all details exhaustively, nor do they limit the invention to mere implementation methods. Obviously, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the present invention, thereby enabling those skilled in the art to better understand and utilize the present invention.

Claims

1. A method for automatically collecting production process parameters of prefabricated components based on the Internet of Things, characterized in that, include: Step S1: Collect target pre-selected process, collect process parameter quantization images of the completed process components of the target pre-selected process, and perform image analysis on the process parameter quantization images. Set a pixel traversal window to collect pixel deviations in the neighborhood of the process parameter quantization images. Based on the pixel deviations in the neighborhood of the pixels, identify the edge pixels of the components and filter their corresponding edge-connecting pixels. Connect the edge pixels of the components with the edge-connecting pixels in sequence to obtain the pre-connected component outline. Collect historical components and extract the outlines of their top view projections to obtain the projection component edge outlines. Perform overlap region analysis on the pre-selected component outlines and the projection component edge outlines, and mark the outline abnormal areas and outline missing areas in the process parameter images respectively to obtain the process component abnormal collection data. Step S2: Based on the abnormal data collected from the process components, screen and issue warnings for individual early warning components, and perform anomaly location overlap analysis on the individual early warning components to obtain process abnormality analysis data; Step S3: Issue anomaly warnings for the target pre-selected processes based on process anomaly analysis data.

2. The method for automatically collecting production process parameters of prefabricated components based on the Internet of Things according to claim 1, characterized in that, The specific steps in step S1 are as follows: Step S11: Collect the production processes included in the prefabricated components to obtain multiple component production processes, and select the target pre-selected process from the collected component production processes; Step S12: Collect the completed process components of the target pre-selected process, randomly select a sample process component from the collected process components, collect and analyze the image of the sample process component, and obtain the process parameter quantification image based on the analysis results. Step S13: Obtain the process parameter quantization image corresponding to each process component to obtain process component abnormal acquisition data.

3. The method for automatically collecting production process parameters of prefabricated components based on the Internet of Things according to claim 2, characterized in that, The specific steps in step S12 are as follows: Step S121: Acquire images of the outer surface of the components included in the sample process components to obtain surface images of the components; Step S122: Discretize the component surface image into several image pixels and select sample pixels, and set a pixel traversal window in the component surface image; Step S123: Control the pixel traversal window to coincide with the sample pixels, perform pixel depth analysis on the sample pixels through the pixel traversal window, divide the sample pixels into component edge pixels or non-component edge pixels, and extract the contour of the sample process component based on the component edge pixels to obtain the component contour extraction image.

4. The method for automatically collecting production process parameters of prefabricated components based on the Internet of Things according to claim 3, characterized in that, The specific steps in step S12 are as follows: Step S124: Collect historical components that have been completed and passed quality inspection in the target pre-selected process, extract the contour of the top view projection corresponding to the historical component, obtain the edge contour line of the projected component, and scale the edge contour line of the projected component to the same proportion as the component contour extraction image. Step S125: Use the projection component edge contour lines to cover the component contour extraction image. In the component contour extraction image, mark the closed area enclosed by the pre-connected component contour lines as the actual component contour graphic, and mark the closed area enclosed by the projection component edge contour lines as the reference component edge contour graphic. Step S126: In the component contour extraction image, compare the actual component contour graphic with the reference component edge contour graphic. Mark the graphic area that belongs only to the actual component contour graphic and is not covered by the reference component edge contour graphic as the contour abnormal area, and mark the graphic area that belongs only to the reference component edge contour graphic and is not covered by the actual component contour graphic as the contour missing area, so as to obtain the process parameter quantification image.

5. The method for automatically collecting production process parameters of prefabricated components based on the Internet of Things according to claim 3, characterized in that, In step S123, the specific steps are as follows: Image pixels adjacent to the pixel traversal window are collected to obtain multiple window adjacent pixels. Pixel depth is collected for the sample pixels to obtain the window pixel depth value. Pixel depth is collected for the corresponding pixels of the window to obtain multiple adjacent pixel depth values. The absolute difference between each adjacent pixel depth value and the window pixel depth value is calculated, and the average of the obtained absolute differences is calculated to obtain the neighborhood pixel deviation corresponding to the pixel traversal window. Set a preset value for reasonable deviation of neighboring pixels. If the neighboring pixel deviation corresponding to the pixel traversal window is greater than or equal to the preset value for reasonable deviation of neighboring pixels, then the sample pixel is marked as a component edge pixel. If it is less than the preset value, then the sample pixel is marked as a non-component edge pixel.

6. The method for automatically acquiring production process parameters of prefabricated components based on the Internet of Things according to claim 5, characterized in that, In step S123, the specific steps are as follows: The pixel traversal window is used to traverse the image pixels except for the sample pixels. Based on the traversal results, all component edge pixels in the surface image of the first component are obtained. One feature edge pixel is selected from them, and the pixel distance between the feature edge pixel and the other component edge pixels is calculated. The obtained pixel distance values ​​are compared. Mark the component edge pixels corresponding to the minimum pixel distance value as edge connection pixels. Connect the feature edge pixels with the edge connection pixels to obtain the pre-connected component outline. Replace the feature edge pixels with the edge connection pixels and re-select the edge connection pixels corresponding to the feature edge pixels for connection. Continue in this way until the pre-connected component outline is closed to obtain the component outline extraction image.

7. The method for automatically collecting production process parameters of prefabricated components based on the Internet of Things according to claim 1, characterized in that, In step S2, the specific steps are as follows: Step S21: Obtain abnormal data of process components, and obtain the process parameter quantization image corresponding to each process component based on the abnormal data of process components; Step S22: If there is a contour abnormality area in the process parameter quantification image, mark the process component as a single warning component and issue a contour abnormality warning for the single warning component. Step S23: If there is a region with missing outline, mark the process component as a single warning component and issue a missing outline warning for the single warning component; Step S24: If both contour abnormality areas and contour missing areas exist simultaneously, mark the process component as a single warning component, and issue contour abnormality warnings and contour missing warnings for the single warning component. Step S25: If there are no abnormal contour areas or missing contour areas, there is no need to issue an early warning for the process components; Step S26: Perform defect area overlap analysis on the quantitative image of each process parameter of the single early warning component, obtain the defect overlap index at multiple locations, and obtain process anomaly analysis data.

8. The method for automatically collecting production process parameters of prefabricated components based on the Internet of Things according to claim 7, characterized in that, In step S26, the specific steps are as follows: Arbitrarily select one sample early warning component and one feature early warning component from the obtained individual early warning components. Collect the contour abnormal region of the sample early warning component in the corresponding process parameter quantization image to obtain the first abnormal region. Collect the contour abnormal region of the feature early warning component in the corresponding process parameter quantization image to obtain the second abnormal region. Control the first abnormal region and the second abnormal region to keep them overlapping. Collect the area value of the overlapping region and calculate the ratio of the obtained area value to the first abnormal region to obtain the abnormal area overlap degree. Replace the feature warning component with each individual warning component, obtain the abnormal area overlap between the sample warning component and each individual warning component, calculate the product of the average abnormal area overlap and the area value of the first abnormal region, and obtain the abnormal overlap quantification value corresponding to the sample warning component.

9. The method for automatically collecting production process parameters of prefabricated components based on the Internet of Things according to claim 8, characterized in that, In step S26, the specific steps are as follows: The first missing region is obtained by collecting the contour missing region corresponding to the sample warning component, and the second missing region is obtained by collecting the contour missing region corresponding to the feature warning component. The first missing region and the second missing region are controlled to overlap. The area value of the overlapping region is collected, and the ratio of the obtained area value to the first missing region is calculated to obtain the missing area overlap degree between the sample warning component and the feature warning component. Replace the feature warning component with each individual warning component, obtain the missing area overlap between the sample warning component and each individual warning component, calculate the product of the average missing area overlap and the area value of the first missing region, and obtain the missing overlap quantification value corresponding to the sample warning component. The average of the abnormal overlap quantization values ​​and the missing overlap quantization values ​​is used to obtain the contour defect overlap quantization value; Obtain the contour defect overlap quantization value corresponding to each individual early warning component, and mark the maximum contour defect overlap quantization value as the position defect overlap index.

10. The method for automatically collecting production process parameters of prefabricated components based on the Internet of Things according to claim 1, characterized in that, In step S3, the specific details are as follows: Obtain process anomaly analysis data, obtain multiple location defect overlap indices based on the process anomaly analysis data, compare the values ​​of multiple location defect overlap indices, and mark the location defect overlap index with the largest value as the process defect index. Create a baseline range for the process defect index. If the process defect index is within the baseline range, there is no need to issue an anomaly warning for the target pre-selected process production line. If the process defect index is not within the baseline range, an anomaly warning is issued for the target pre-selected process production line.

Images