A mold intelligent fault prediction and diagnosis method, device, equipment and medium

Abstract

The application relates to a mold intelligent fault prediction and diagnosis method, device, equipment and medium, a mold intelligent fault prediction and diagnosis method comprises the following steps: acquiring mold production requirements, determining corresponding detection areas and precision thresholds of each detection area; acquiring a cavity plane image of the mold in an open state; matching an image area corresponding to the detection area in the cavity plane image; analyzing the image area, generating and converting wear analysis data; judging whether the wear analysis data is greater than the precision threshold, if yes, generating a corresponding diagnosis workpiece, and if not, generating a corresponding fault range prediction result. By customizing the detection area and the precision threshold according to the mold production requirements, different production requirements can be flexibly adapted, so that the fault prediction is more accurate and effective; under the support of the visual recognition model, the fault area in the mold image can be accurately identified, and the accuracy and reliability of fault detection are significantly improved.

Description

Technical Field

[0001] This invention relates to the technical field of mold fault prediction and diagnosis, and in particular to a method, device, equipment and medium for intelligent mold fault prediction and diagnosis. Background Technology

[0002] Currently, mold fault diagnosis and prediction largely rely on manual experience or traditional sensor monitoring methods, predicting faults by periodically inspecting or sensing the physical condition of the mold. However, traditional technical methods have certain limitations, especially in quickly and efficiently detecting minute defects such as wear and cracks, where they often fall short of accuracy and timeliness.

[0003] Existing mold failure prediction systems typically monitor the mold's operating status in real time using sensors, combined with simple algorithms for fault detection. However, these systems generally cannot accurately detect complex mold surfaces and minute defects, and struggle to effectively integrate multiple data sources for comprehensive fault prediction. Therefore, existing technologies have certain shortcomings in terms of detection accuracy, real-time performance, and adaptability to different fault types. Summary of the Invention

[0004] To address the problem that traditional fault prediction methods rely on manual inspection and simple sensor data, lacking high-precision and intelligent diagnostic capabilities, this application provides a method, device, equipment, and medium for intelligent fault prediction and diagnosis of molds.

[0005] The above-mentioned objective of this application is achieved through the following technical solution:

[0006] A method for intelligent mold fault prediction and diagnosis, comprising:

[0007] Obtain the mold production requirements, and determine the corresponding detection areas and the accuracy threshold for each detection area based on the mold production requirements;

[0008] Using an industrial camera, acquire planar images of the mold cavity when it is open;

[0009] Based on the established visual recognition model, the image region corresponding to the detection region is matched in the cavity plane image;

[0010] The image region is analyzed to generate and convert wear analysis data at the same scale as the accuracy threshold;

[0011] judging whether the wear analysis data is greater than the precision threshold value, if yes, establishing a corresponding 3D printing model according to the wear analysis data, generating a corresponding diagnostic workpiece according to the 3D printing model based on additive manufacturing technology, and if no, generating and recording a corresponding fault range prediction result according to the wear analysis data.

[0012] By adopting the technical scheme, the detection area and the precision threshold value thereof can be customized according to different production requirements, thereby improving the accuracy and applicability of fault prediction; the mold cavity planar image can be acquired based on the industrial camera, thereby acquiring high-resolution and high-precision mold image information, thereby providing detailed visual data support for subsequent visual recognition and fault analysis; the fault area such as wear and crack can be accurately recognized from the mold image based on the established visual recognition model matching the image area, thereby improving the accuracy and reliability of fault detection; the detection data can be unified with the precision requirement by analyzing the image area and converting it into wear analysis data consistent with the precision threshold value, thereby providing standardized data support for subsequent fault diagnosis and repair decision-making.

[0013] In a preferred example, the application can be further configured to: in the step of acquiring the mold production requirement and determining the corresponding detection area and the precision threshold value of each detection area according to the mold production requirement, comprising:

[0014] A user input interface is pushed, the user input interface includes a plurality of block areas, each block area is provided with a plurality of interactive items, the interactive items include index codes, input explanation contents and interactive input boxes arranged in sequence, and the input explanation contents are used to explain the information to be input in the interactive input boxes;

[0015] If the confirmation submission instruction of the user is detected, the input content of the interactive input box is acquired;

[0016] According to the matching field of the block area, the input content of the same block area is determined as the detection content of the detection area;

[0017] According to the index code, the precision threshold value corresponding to the detection content is matched as the precision threshold value of the detection area.

[0018] By adopting the technical scheme, the detection area and the precision threshold value can be flexibly adjusted according to the actual requirements of mold production by pushing the user input interface and acquiring the user input content through the interactive items, thereby realizing the customized detection scheme and ensuring the accuracy and applicability of detection.

[0019] The application can be further configured in a preferred example as follows: in the step of matching the image region corresponding to the detection region in the cavity planar image based on the established visual recognition model, the application comprises:

[0020] Comparing the cavity planar image with the calibrated overall template image based on the established visual recognition model to divide the feature region, wherein the feature region comprises a template region, a bushing region, a cavity region, a pin region and a runner region;

[0021] Matching the corresponding feature region according to the index code;

[0022] Matching the corresponding bounding coordinates according to the detection region;

[0023] Based on the bounding coordinates, the corresponding region is framed and enlarged on the feature region to serve as the image region corresponding to the detection region.

[0024] By adopting the above technical solution, the feature region is divided by comparing the cavity planar image with the calibrated overall template image based on the established visual recognition model, which can accurately divide each important part of the mold (such as the template region and the bushing region), thereby improving the detection accuracy and ensuring that the wear and failure of different regions can be effectively monitored. By matching the corresponding feature region and the bounding coordinates and framing and enlarging the region, the specific region can be accurately focused for detailed analysis, thereby improving the detection efficiency and accuracy.

[0025] The application can be further configured in a preferred example as follows: in the step of analyzing the image region to generate and convert wear analysis data in the same scale as the precision threshold, the application comprises:

[0026] Matching the calibrated partial template image according to the detection region;

[0027] Comparing the image region with the calibrated partial template image to determine the wear region in the image region where there is a color difference, and determine the wear position coordinates, pixel unit area and gray value change data of the wear region;

[0028] Determining the image resolution of the industrial camera, and converting the pixel unit area of the wear region into a physical unit area according to the mapping relationship between the image resolution and the actual physical size;

[0029] Calculating the wear depth based on the calibration relationship between the gray value change data and the actual wear depth;

[0030] Substituting the wear position coordinates, physical unit area and wear depth into the trained wear analysis model to calculate the corresponding wear analysis data.

[0031] By adopting the technical scheme, the position of the wear region can be accurately found and marked in the image by matching the part of the template image calibrated according to the detection region, so as to improve the accuracy and reliability of wear detection; the color difference region is determined as the wear region by comparing the image region and the part of the template image calibrated, so as to accurately identify the tiny wear region, thereby avoiding missed detection or false detection; the consistency between the image data and the actual physical data can be realized by determining the image resolution and converting the wear region area according to the physical size mapping relationship, so as to improve the practicability and accuracy of fault prediction; the wear depth can be accurately calculated by the calibration relationship between the gray value change and the actual wear depth, so as to provide accurate reference data for subsequent repair.

[0032] In a preferred example, the application can be further configured to: in the step of establishing a corresponding 3D printing model according to the wear analysis data, comprising:

[0033] calling a basic model, the basic model comprising a template model, a mold core model, a cavity model, a thimble model and a runner model;

[0034] cropping the basic model according to a preset rule to generate a corresponding preliminary model according to the wear position coordinates corresponding to the wear analysis data and the pixel unit area;

[0035] calculating the wear shape according to the gray value change data corresponding to the wear analysis data;

[0036] further cropping the preliminary model according to the wear shape and the wear depth to generate a corresponding 3D printing model.

[0037] By adopting the technical scheme, the preliminary model can be accurately generated by calling the basic model and according to the wear position coordinates corresponding to the wear analysis data and the pixel unit area, so as to provide accurate diagnosis basis for subsequent repair work; the shape of the repair region can be further accurately refined by calculating the wear shape according to the gray value change data, so as to avoid unnecessary over-repair or omission in the repair process; the 3D printing model generated can be highly consistent with the actual wear region by further cropping the preliminary model according to the wear shape and the depth, so as to improve the repair effect and work efficiency.

[0038] In a preferred example, the application can be further configured to: in the step of establishing a corresponding 3D printing model according to the wear analysis data, comprising:

[0039] obtaining mold design parameters;

[0040] The mold design parameters and the wear analysis data are substituted into the trained prediction model to generate corresponding trend results, the trend results including time trend and space trend;

[0041] According to the time trend, a time period grading alarm process is added;

[0042] According to the space trend, corresponding visualized space features are generated and pushed to the corresponding management terminal.

[0043] By adopting the above technical solution, by obtaining mold design parameters and substituting them into the trained prediction model with wear analysis data, comprehensive fault range prediction can be performed based on the design and wear data of the mold, so as to identify potential fault risks in advance; by generating corresponding time trend and space trend, the occurrence time and area of the fault can be accurately predicted, so as to help develop a more scientific maintenance plan and reduce unexpected downtime and maintenance costs; by adding a grading alarm process to the time trend, an alarm can be issued when the fault approaches the critical state, so that maintenance personnel can take action in time to prevent the fault from occurring; by generating and pushing the visualized features of the space trend, management personnel can intuitively understand the working state of the mold, thereby improving management efficiency and decision-making accuracy.

[0044] The above-mentioned second invention purpose of the present application is realized by the following technical solution:

[0045] A mold intelligent fault prediction and diagnosis device, the mold intelligent fault prediction and diagnosis device comprising:

[0046] A first acquisition module is configured to acquire mold production requirements, and determine corresponding detection areas and precision thresholds of each detection area according to the mold production requirements;

[0047] A second acquisition module is configured to acquire cavity plane images of a mold in an open state based on an industrial camera;

[0048] A matching module is configured to match image regions corresponding to the detection areas in the cavity plane images based on the established visual recognition model;

[0049] A generation module is configured to analyze the image regions, generate wear analysis data, and convert the wear analysis data into the same scale as the precision thresholds;

[0050] A judgment module is configured to judge whether the wear analysis data is greater than the precision thresholds, if yes, establish a corresponding 3D printing model according to the wear analysis data, generate a corresponding diagnosis workpiece based on additive manufacturing technology according to the 3D printing model, and if not, generate and record a corresponding fault range prediction result according to the wear analysis data.

[0051] The third purpose of the present application is achieved by the following technical solution:

[0052] A computer device comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the above-mentioned mold intelligent fault prediction and diagnosis method when executing the computer program.

[0053] The fourth purpose of the present application is achieved by the following technical solution:

[0054] A computer-readable storage medium stores a computer program, and the computer program implements the steps of the above-mentioned mold intelligent fault prediction and diagnosis method when executed by a processor.

[0055] In summary, the present application includes at least one of the following beneficial technical effects:

[0056] 1. By customizing the detection area and its precision threshold according to the mold production requirements, different production requirements can be flexibly adapted, so that the fault prediction is more accurate and effective; with the help of industrial cameras, high-resolution and high-precision cavity images are obtained, providing detailed image data for subsequent visual recognition and fault analysis, which makes it possible to accurately detect the wear, cracks and other faults of the mold; with the support of the visual recognition model, the fault area in the mold image can be accurately identified, significantly improving the accuracy and reliability of fault detection; at the same time, by converting the image analysis results into wear data consistent with the precision threshold, the uniformity and standardization of the data are ensured, thereby providing a reliable data basis for subsequent fault diagnosis and repair decisions, which helps to realize fine management and optimize maintenance decisions;

[0057] 2. In terms of diagnosis function, the entire system can automatically identify the fault type and fault range of the mold through precise image analysis and wear data conversion. When the detected wear area exceeds the preset precision threshold, the system can generate corresponding 3D printing models for actual repair. This process not only has high automation, but also can provide targeted repair solutions, avoiding errors or omissions that may occur in traditional manual diagnosis. By accurately calculating the shape and depth of the wear area, the system can generate diagnostic workpieces that highly conform to the actual wear condition, which can be used for mold repair verification, greatly improving repair efficiency and accuracy. At the same time, the system can also predict future possible faults through the prediction model, provide the time and spatial trend of fault occurrence, and then take maintenance measures in advance, reduce production downtime, reduce maintenance cost, and improve the service life and production efficiency of the mold. BRIEF DESCRIPTION OF DRAWINGS

[0058] Figure 1 is a flow chart of a mold intelligent fault prediction and diagnosis method in an embodiment of the present application.

[0059] Figure 2 is an implementation flow chart of step S10 in a mold intelligent fault prediction and diagnosis method in an embodiment of the present application.

[0060] Figure 3 is an implementation flow chart of step S30 in a mold intelligent fault prediction and diagnosis method in an embodiment of the present application.

[0061] Figure 4 is an implementation flow chart of step S40 in a mold intelligent fault prediction and diagnosis method in an embodiment of the present application.

[0062] Figure 5 is an implementation flow chart of step S50 in a mold intelligent fault prediction and diagnosis method in an embodiment of the present application.

[0063] Figure 6 is another implementation flow chart of step S50 in a mold intelligent fault prediction and diagnosis method in an embodiment of the present application.

[0064] Figure 7 is a structural schematic diagram of a mold applied by a mold intelligent fault prediction and diagnosis method in an embodiment of the present application.

[0065] Figure 8 is a structural schematic diagram of step S30 in a mold intelligent fault prediction and diagnosis method in an embodiment of the present application.

[0066] Figure 9 is a principle block diagram of a mold intelligent fault prediction and diagnosis device in an embodiment of the present application.

[0067] Figure 10 is a device schematic diagram in an embodiment of the present application. DETAILED DESCRIPTION

[0068] The present application will be further described in detail below in combination with the accompanying drawings.

[0069] In an embodiment, as shown in the accompanying drawings, the present application discloses a mold intelligent fault prediction and diagnosis method, and the mold intelligent fault prediction and diagnosis method comprises: Figure 1

[0070] ​S10, obtaining mold production requirements, determining corresponding detection areas and precision thresholds of each detection area according to mold production requirements; in this embodiment, first, the detailed production requirements of the mold are obtained through a user input interface or from a production management system, these requirements include the product type, production cycle, material type and quality requirements of the final product required by the mold production, and the working environment parameters of the mold, such as temperature, pressure and other working conditions, are also included in the production requirements, then, the system automatically calculates the detection areas of the mold, such as the cavity, core, runner and other areas, according to these requirements through a preset algorithm, the determination of the detection area is not only based on the design and function of the mold, but also considers the parts that may be worn or failed in the actual work of the mold, then, the system automatically sets the precision threshold of each area according to the importance and precision requirement of each detection area, for example, some key areas may require higher detection precision (such as ±0.01mm), while the precision threshold of other non-key areas may be set to a more relaxed standard (such as ±0.1mm), the setting of the precision threshold is based on the design tolerance of the mold and the precision requirement of the product, to ensure the effectiveness and accuracy of fault diagnosis.

[0071] S20, obtaining the cavity planar image of the mold in the open state based on the industrial camera; in this embodiment, a high-resolution industrial camera installed on a robot or automated device is used, which can shoot the planar image of the mold cavity when the mold is opened, during the shooting process, the industrial camera will automatically adjust the focal length, aperture and exposure time according to the actual size, shape and surface features of the mold, to ensure that the image is clear and distortion-free. In order to avoid the decline of image quality caused by uneven lighting conditions or reflection, the industrial camera will generally cooperate with a ring-shaped LED light source or a high-brightness directional light source to illuminate the mold surface, so as to ensure that a uniform and clear image is obtained. For example, in the application of injection mold, the industrial camera will capture detailed images of the mold cavity, such as the runner, injection port and tiny cracks and wear on the surface of the mold, in this case, the image data obtained by shooting can provide sufficient visual data support for subsequent fault diagnosis.

[0072] S30, based on the established visual recognition model, matching the image region corresponding to the detection region in the cavity plane image; in this embodiment, after obtaining the mold cavity plane image, the image is processed through the pre-trained visual recognition model, the model analyzes the texture, color, edge and other features in the image, and compares with the preset detection region template, so as to identify the image region corresponding to the detection region. For example, the model will identify the cavity, die core, runner and other regions and match them with the previously calibrated template image. The detected key areas such as the contact surface of the die core and the runner part of the cavity will be preferentially identified and analyzed in detail. In the image matching process, the visual recognition model uses deep learning technology such as convolutional neural network (CNN), which can automatically identify and locate the preset detection region in the image through image feature extraction and image matching algorithm, thereby improving the accuracy and efficiency of fault detection. For example, in the runner area of the mold, if part of it is blocked or has a crack, the model can accurately identify it, further helping fault diagnosis.

[0073] S40, analyzing the image region, generating and converting into wear analysis data with the same scale as the precision threshold; in this embodiment, after the visual recognition model matches the target detection region, the system analyzes the wear in the image region through image processing algorithm. First, the system will perform gray scale transformation and filtering processing on the image to improve the contrast and clarity of the wear area in the image. Then, the edge detection algorithm (such as Sobel or Canny operator) is used to identify the edge of the wear area, and the area, shape and depth of the wear are analyzed. Then, the system quantifies the features of the wear area according to the precision threshold, and converts it into wear analysis data consistent with the precision threshold. For example, if the detected wear area is a crack in the image with an area of 100 pixels², the system converts it to the actual physical area of 0.01 mm² according to the set resolution (e.g. 100 pixels = 1 mm), and calculates the corresponding wear depth by analyzing the gray value change of the area. In this way, the wear data can be unified to the physical scale, compared with the set precision threshold, and accurate data information is provided for subsequent diagnosis and repair.

[0074] S50, determine whether the wear analysis data is greater than the precision threshold, if yes, establish the corresponding 3D printing model according to the wear analysis data, generate the corresponding diagnostic workpiece based on the additive manufacturing technology according to the 3D printing model, if not, generate and record the corresponding fault range prediction result according to the wear analysis data. In this embodiment, after generating the wear analysis data, the system will compare with the preset precision threshold, if the data of the wear area (such as area, depth, etc.) exceeds the precision threshold, the system will automatically enter the generation stage of 3D printing model, based on the position, shape and depth in the wear analysis data, the corresponding 3D printing model is established, which can be printed through additive manufacturing technology (such as selective laser melting or fused deposition modeling), the printed workpiece is used to repair the wear area of the mold. For example, if the wear depth of a certain area exceeds the set threshold, the system will generate a 3D printing model with a compensation layer and send it to the printer for printing. If the wear data does not exceed the threshold, the system will predict the trend of future failure through data association, generate and record the fault range prediction result, which includes the time and area of failure, the system will save these data in the database and generate a prediction report for the user to refer to in subsequent mold maintenance.

[0075] In summary, by customizing the detection area and its precision threshold according to the mold production requirements, different production requirements can be flexibly adapted, so that the fault prediction is more accurate and effective; with the help of industrial camera to obtain high resolution and high precision cavity image, detailed image data is provided for subsequent visual recognition and fault analysis, which makes it possible to accurately detect the wear, crack and other faults of the mold; with the support of visual recognition model, the fault area in the mold image can be accurately identified, which significantly improves the accuracy and reliability of fault detection; at the same time, by converting the image analysis result into wear data consistent with the precision threshold, the uniformity and standardization of the data are ensured, which provides a reliable data basis for subsequent fault diagnosis and repair decision, and helps to realize fine management and optimize maintenance decision;

[0076] In terms of diagnostic capabilities, the entire system can automatically identify the type and extent of mold failures through precise image analysis and wear data conversion. When the detected wear area exceeds a preset accuracy threshold, the system can generate a corresponding 3D printed model for actual repair. This process is not only highly automated but also provides targeted repair solutions, avoiding errors or omissions that may occur in traditional manual diagnostics. By accurately calculating the shape and depth of the wear area, the system can generate diagnostic workpieces that closely match the actual wear conditions. These workpieces can be used for mold repair verification, greatly improving repair efficiency and accuracy. Simultaneously, the system can also predict future failures using predictive models, providing the time and spatial trends of failure occurrence, thereby enabling proactive maintenance measures to reduce production downtime, lower maintenance costs, and improve mold lifespan and production efficiency.

[0077] In one embodiment, such as Figure 2 As shown, in step S10, which involves obtaining the mold production requirements and determining the corresponding detection areas and the accuracy threshold for each detection area based on those requirements, the following steps are included:

[0078] S101. Push the user input interface. The user input interface includes multiple block areas, each block area has multiple interactive items. The interactive items include an index code, input explanation content, and an interactive input box arranged in sequence. The input explanation content is used to explain the information that needs to be entered in the interactive input box. In this embodiment, firstly, multiple block areas are displayed on the user interface. The design of the block areas allows the user to input relevant information according to the mold production requirements. The design of each block area is intended to cover a specific detection area and its corresponding requirements. For example, one area is used to input the mold type, and another area is used to input the mold precision requirements. Each block area contains multiple interactive items. Each item contains an index code, corresponding input explanation content, and an interactive input box. The index code is used to distinguish the specific information required for each detection area. The input explanation content provides specific guidance for the user input, such as "Please enter the mold cavity size (unit: mm)" or "Please enter the temperature range in the injection molding process". The interactive input box provides a place for the user to fill in information, ensuring that the user can accurately input according to the requirements.

[0079] S102. If the user's confirmation submission instruction is detected, the input content of the interactive input box is obtained. In this embodiment, after the user has filled in the content of all input boxes, the user clicks the submit button or confirms the instruction. The system will detect the confirmation instruction and trigger the input content acquisition operation. The system will extract the content filled in by the user in each interactive input box. This content will be temporarily stored and await further processing. For example, the user may have entered the size range of the mold, the required accuracy standard, etc. This data will serve as the basis for setting the accuracy threshold and matching the detection area in subsequent steps.

[0080] S103. Match the corresponding detection area according to the matching field of the block area, and determine the input content of the same block area as the detection content of the detection area. In this embodiment, according to the content input by the user, the input content in each block area is matched with the specific detection area of ​​the mold through the preset matching rules. The matching process is to perform intelligent matching based on the field name of the block area and the input data. For example, if the user inputs "injection mold" in the "mold type" area, the input content will be matched with the corresponding detection area "cavity" and the content will be used as the detection requirement of the area. If the input content of a certain area is related to multiple detection areas, the most suitable detection area is determined according to the matching rules and associated with the corresponding detection content to ensure that the input information of each area can accurately correspond to the actual area that needs to be detected.

[0081] S104. Match the accuracy threshold corresponding to the detection content according to the index code, and use it as the accuracy threshold for the detection area. In this embodiment, the system compares the detection content input by the user with the preset accuracy standard database, and looks up the accuracy threshold corresponding to each detection area through the index code. The accuracy threshold is set based on the specific requirements of the mold and production needs. For example, a smaller accuracy threshold (such as ±0.01mm) may be set for some high-precision detection areas (such as cavities), while a more lenient accuracy threshold (such as ±0.1mm) may be set for some less important areas (such as the shell). According to these matching rules, the accuracy threshold will be accurately assigned to each detection area to ensure that subsequent fault detection and analysis can be carried out based on reasonable standards.

[0082] In one embodiment, such as Figure 3 As shown, in step S30, which is the step of matching the image region corresponding to the detection region in the cavity plane image based on the established visual recognition model, the following steps are included:

[0083] S301, based on the established visual recognition model, compare the cavity planar image and the calibrated overall template image to divide the feature area, the feature area includes the template area, the mold core area, the cavity area, the ejector pin area and the runner area; in this embodiment, the cavity planar image is the image information of the cavity plane of the mold, as shown in Figure 7 , the position of the cavity plane is shown, as shown in Figure 8 , the position of the feature area is shown, first, the visual recognition model trained by deep learning analyzes different features in the cavity planar image through image processing algorithm, the model compares the edge, texture, color and other features of the image with the calibrated overall template image, automatically identifies and divides each important component area in the image. For example, the template area generally corresponds to the outer frame of the mold, the model will match the outer edge in the image with the outer frame of the template when identifying; the mold core area is usually the key part of product forming, the model will automatically identify the contour of the mold core according to the feature points and lines in the image; the cavity area is usually the place where the product is actually formed in the mold, the model identifies and divides according to the shape and feature pattern of the cavity; the runner area is used to guide the flow of molten material, the system automatically identifies and distinguishes the runner area through the shape of the flow channel in the image; the ejector pin area is used to take out the product, the model will identify and demarcate the area where the ejector pin is located. Through this comparative analysis, different feature areas can be accurately identified and divided from the image, providing accurate area positioning for subsequent fault diagnosis.

[0084] S302, according to the index code, match the corresponding feature area; in this embodiment, the system uses the index code to match each area with the corresponding area information in the template according to the previously divided feature area. The index code is the unique identifier of each detection area, for example, the index code of the cavity area may be "01", the index code of the mold core area may be "02", the index code of the ejector pin area may be "03", the index code of the runner area may be "04", each index code corresponds to the specific function and structure position of the mold part. By searching for the matching area according to the index code, the system can ensure that the detection area and the parts of the mold design correspond one by one, avoiding detection misplacement or error area. For example, in the detection of a multi-cavity mold, the system identifies the runner area of a certain mold cavity according to the index code, and ensures the accurate identification of the area by matching the index code "04", preventing it from being mistakenly identified as other areas.

[0085] S303、According to the detection area, match the corresponding frame selection coordinates; in this embodiment, after completing the recognition of the feature area and the matching of the index code, the system will generate corresponding frame selection coordinates according to the specific position and range of each detection area in the image. The frame selection coordinates are the boundary positions of the calibration area in the image, which ensures that the subsequent processing is only concentrated in the area of interest. For example, in the cavity area of the mold, the system will calculate the pixel coordinates of the upper left corner and the lower right corner of the area based on the feature points in the image, such as the sharp parts and geometric contours of the mold edge, and determine the specific position of the rectangular frame selection area. In this way, the subsequent image processing algorithm can only analyze within the frame selection area, avoiding unnecessary background processing and improving efficiency and accuracy.

[0086] S304, based on the frame selection coordinates, frame selection and magnification of the corresponding area on the feature area to serve as the image area corresponding to the measurement area. In this embodiment, after generating the frame selection coordinates, the image processing algorithm will frame the feature area according to these coordinates and magnify the frame selection area. The purpose of magnification is to enhance the clarity of the detailed image, so that the subtle wear, cracks and other defects in the image can be more easily identified. For example, when detecting the flow channel area of the mold, if there are small cracks or blockages in the area, the system can display clearer image details after framing and magnifying the area, making the edges of the cracks more prominent and facilitating subsequent analysis and judgment. During the magnification process, the system will ensure that the image resolution is high enough to accurately measure and locate small defects, ensuring that wear, cracks and other problems can be accurately diagnosed.

[0087] Specifically, assume that during the inspection of an injection mold, the system needs to identify the cavity area of the mold and analyze the wear condition therein. In step S301, first, the system obtains the cavity planar image of the mold, then the system uses the pre-trained visual recognition model to divide different feature areas by comparing the cavity planar image with the calibrated overall template image. The calibration template contains various components of the mold, such as the template area, the insert area, the cavity area, the ejector pin area, and the runner area. The system first identifies the external contour of the mold, i.e., the template area, according to the texture and edge features in the image, and compares it with the edge in the template image to automatically demarcate the outline frame of the mold. Then, the model continues to analyze other parts of the cavity image, and by comparing the complex geometric structure in the mold, the model automatically identifies and divides the insert area, the cavity area, and the runner area, etc. For example, in the insert area, the system determines the specific location of the part for molding by identifying the relatively regular shape area in the image; in the cavity area, the system accurately divides the area by observing the hollow part in the image and the details inside the mold. In step S302, the system uses the pre-set index code to match according to these divided feature areas. Each area has a corresponding index code, for example, the index code of the insert area may be "01", and the index code of the cavity area may be "02". By matching the index code, the system ensures that the matching information of each area in the image conforms to the design requirements of the mold, avoiding the misidentification of one area as another. For example, when the system identifies the cavity area, based on the index code "02" of the area, the system will further confirm that it is a key area for molding products, rather than other areas of the mold, ensuring that appropriate wear detection is performed on this area in subsequent analysis. In step S303, the system generates bounding coordinates according to the features of each detection area. The bounding coordinates define the specific location and range of the detection area in the image. For example, after identifying the cavity area, the system automatically generates the coordinates of the area by calculating the edge features in the image, and frames the upper and lower boundaries of the cavity, and uses these coordinates for subsequent processing. At this time, the framed cavity area may contain details of multiple mold surfaces, and the system will accurately determine the top-left and bottom-right coordinates of the area to ensure that subsequent processing is concentrated only within this key area. In step S304, the system frames and enlarges the corresponding area based on the bounding coordinates. For example, the system finds that there is a small crack in a part of the cavity area, and after the bounding coordinates determine the location of the part, the system will enlarge the area for more detailed analysis. The purpose of enlargement is to improve the resolution of the area, making the crack in the image more clearly visible. After enlargement, the system can clearly observe the shape and depth of the crack, helping subsequent fault analysis.With the magnification, the edges of the crack are enlarged enough to be clear, and the system can accurately analyze the size and depth of the crack based on the grayscale changes in the image and edge detection techniques, which helps to determine whether the area needs to be repaired. This process significantly improves the system's ability to diagnose mold wear or defects. In this example, through precise analysis and image processing at each step, the system can accurately identify defects such as wear and cracks from the cavity area of the mold, ensuring effective monitoring and fault diagnosis of each key component of the mold.

[0088] In an embodiment, as shown in FIG. 4, in step S40, i.e., the step of analyzing the image area, generating and converting the wear analysis data to the same scale as the precision threshold, includes: Figure 4

[0089] S401, according to the detection area, match the calibrated partial template image; in this embodiment, first, by defining the detection area, the feature areas in the cavity image are identified, which can be different working components in the mold, such as cavity, core, runner, etc. Then, the obtained cavity planar image is compared with the pre-calibrated template image. The template image is obtained by standardizing the shooting and image calibration of the mold, which contains the accurate positioning information of each important area in the mold, such as the position of the outer edge of the cavity, the runner, etc. in the mold design drawing has been calibrated. In practical application, the model uses image registration technology (such as feature point matching based on SIFT or SURF algorithm), extracts the feature points in the cavity image and the template image, calculates the matching degree between them and determines the position of the detection area. For example, if the runner part in the cavity image matches the runner part in the template image in feature point matching, the system will determine that this area is the runner area. The purpose of this step is to ensure the correct identification and division of the image area, and to avoid misaligned matching.

[0090] ​S402, compare the image region with the calibrated partial template image, determine the area with color difference in the image region as the wear area, and determine the wear position coordinates, pixel unit area, and gray value change data of the wear area; in this embodiment, once the detection region is matched, the image analysis software determines whether there is wear phenomenon according to the color difference and gray value change in the image. Generally, the wear area will present different color or gray value change from the surrounding area, for example, the wear of the mold surface will cause the change of the surface color or the surface texture, and the change of the gray value can reflect the physical change of the area. Through the color difference (such as the difference of RGB value) or the change of the gray value (for example, by calculating the local average gray value of the image) in the image, the system can determine the possible wear area in the image region. The system further analyzes the specific position coordinates of the wear area according to these differences, and calculates the pixel unit area thereof. For example, if the color of a part in the image changes from a deep color to a light gray color, it indicates that the area may be worn due to friction or high temperature. Then, the pixel unit area of the area is calculated and converted into the actual physical area, and the wear degree of the area is further determined by calculating the gray value change.

[0091] S403, determine the image resolution of the industrial camera, and convert the pixel unit area of the wear area into the physical unit area according to the mapping relationship between the image resolution and the actual physical size; in this embodiment, the resolution of the industrial camera determines the pixel density of the image, that is, the number of pixels contained in each unit physical size. For example, if the resolution of the industrial camera is 100 pixels / mm, it means that each 1 mm of actual size is represented as 100 pixels in the image. Based on this resolution, the system can convert the pixel area of the wear area in the image into the actual physical area. For example, assuming that the area of the wear area in the image is 500 pixels², based on the resolution of 100 pixels / mm, the system can convert it into the actual physical area by the following formula: physical area = wear pixel area / (resolution²) = 500 / (100²) = 0.05 mm². Through this mapping relationship, the system can accurately convert the size of the wear area in the image into the actual physical unit, ensuring that the wear data has actual meaning and meets the accuracy requirements.

[0092] S404. Based on the calibration relationship between grayscale value change data and actual wear depth, the wear depth is calculated. In this embodiment, after detecting the wear area and calculating its area, the system estimates the wear depth by measuring the change in grayscale value. There is a certain linear or non-linear relationship between grayscale value change and wear depth. This relationship is usually obtained by calibration on standard samples with different degrees of wear. For example, when wear occurs, the light reflectivity of the mold surface changes, resulting in a change in grayscale value in the image. Areas with lower grayscale values ​​usually indicate deeper wear. The system calculates the actual wear depth by comparing the grayscale change of this area in the image with the calibration data obtained through previous experiments. For example, assuming that the grayscale value change of a certain area is 10 units, based on the pre-calibrated relationship, the wear depth corresponding to a grayscale value change of 10 units is 0.02 mm, then the system can calculate the wear depth of this area as 0.02 mm.

[0093] S405. The wear location coordinates, physical unit area, and wear depth are all substituted into the trained wear analysis model to calculate the corresponding wear analysis data. In this embodiment, after calculating the area and depth of the wear region, the system uses the wear location coordinates, physical unit area, and wear depth as inputs to further analyze the pre-trained wear analysis model. This wear analysis model is trained based on a large amount of historical data and algorithms, and can predict the wear trend, the scope of influence, and the impact on mold performance. By substituting the wear data, the model outputs a comprehensive wear analysis data report. For example, the model can calculate whether the wear area will expand further at some point in the future, or whether repair is needed. In addition, the model can also assess the remaining service life of the mold based on the current wear level and provide corresponding maintenance suggestions. For example, if the depth and area of ​​the wear region exceed a preset threshold, the model may issue a warning and suggest mold repair to avoid affecting production.

[0094] In one embodiment, such as Figure 5 As shown, step S50, which is the step of establishing the corresponding 3D printing model based on the wear analysis data, includes:

[0095] S5011, call the basic model, the basic model includes the template model, the mold core model, the cavity model, the ejector pin model and the runner model; in this embodiment, in this step, first, by calling the pre-designed basic model, the basic model includes the template model, the mold core model, the cavity model, the ejector pin model and the runner model, each model corresponds to a specific area and function in the mold respectively. The template model represents the outer shape frame of the mold, the mold core model is used to bear and shape the shape of the product, the cavity model defines the forming space inside the mold, the ejector pin model is used for product ejection, and the runner model defines the channel of the molten material flow. The establishment of the basic model is based on the CAD drawing of the mold design, which contains accurate dimensions, shapes and geometric parameters. In this step, by calling these models, the subsequent cutting and modification operations are prepared to generate accurate repair models according to the wear analysis data.

[0096] S5012, according to the wear position coordinates and pixel unit area corresponding to the wear analysis data, the basic model is cut according to the preset rule to generate the corresponding preliminary model; in this embodiment, according to the wear analysis data generated in the previous step, first, the coordinates of the wear area and the calculated pixel unit area are located to determine the specific position and area to be repaired, based on this information, the basic model is modified and cut according to the preset cutting rule. For example, if the wear area is located on the contact surface of the mold core, and the pixel area of the area is calculated as 5mm², the area will be cut from the mold core model. The preset rule usually includes the size, shape and adaptability requirement of the cutting area, and the cut model is the preliminary model, which reflects the specific area and spatial form to be repaired.

[0097] S5013, calculate the wear shape according to the gray value change data corresponding to the wear analysis data; in this embodiment, according to the gray value change data in the wear analysis data, the shape of the wear area is calculated. Gray value usually represents the degree and change of wear, and the gray value of the area will change with the occurrence of wear. By comparing the gray value in the image, the shape of the wear can be accurately described. For example, if the gray value of a certain area changes greatly, it means that the area has obvious wear. The system can infer the shape of the wear area by the intensity and mode of the gray value change, and map it to the preliminary model. In this way, a more accurate wear area shape can be obtained, and data basis can be provided for subsequent model cutting and repair.

[0098] S5014、According to the wear shape and the wear depth, further trim the preliminary model to generate the corresponding 3D printing model. In this embodiment, after obtaining the wear shape, it is also necessary to further trim the preliminary model based on the depth data of the wear area. The wear depth is usually calculated by gray scale change, which represents the height change of the wear surface relative to the original surface. According to the wear depth, the repaired part of the trimmed preliminary model is accurately modified. The model will generate a repaired part with corresponding depth compensation based on the wear depth, ensuring that the repaired area matches the original model. For example, if the depth of the wear area is 0.2mm, the model will generate a corresponding compensation structure according to the depth. The compensation part will accurately fill the wear area, forming a complete 3D printing model. The final 3D printing model will serve as the basis for repairing the mold, and through additive manufacturing technology (such as 3D printing), the actual repaired workpiece will be printed.

[0099] Specifically, assume that during the inspection of an injection mold, the system needs to identify the wear condition of the cavity area and generate a repaired 3D printing model. In step S5011, the system first calls multiple part models of the mold from the preset base model, including the template model, the core model, the cavity model, the ejector pin model, and the runner model. For example, the cavity model represents the molding space inside the mold, the core model is the part used to shape the product, and the runner model represents the channel for the flow of molten material. These models are accurately designed based on CAD drawings, and the system retrieves the corresponding part models according to the working requirements and design parameters of the mold, providing accurate references for subsequent repair work. When entering step S5012, the system begins to trim the base model based on the wear analysis data obtained earlier. Assume that the system discovers through image analysis that a part of the cavity area has experienced significant wear due to high temperature and long-term use, with an area of approximately 5mm². The system determines the exact location of the wear area based on the position coordinates in the wear analysis and extracts the area from the cavity model according to the preset trimming rules. This trimming rule is based on the size and shape of the wear, and the system modifies the model accordingly according to these rules to generate a preliminary model containing the wear compensation area. This preliminary model shows the damaged area of the mold and provides a basis for subsequent repair design. Next, in step S5013, the system uses the gray value changes in the wear analysis data to further refine the shape of the wear area. Assume that in the image of the cavity area, the gray value changes significantly, indicating that the surface wear of this part is relatively deep. Through the gray value changes, the system can infer the shape of the wear area, and the area with a large gray value change often means the expansion of the wear area or the increase in surface roughness. The system accurately draws the specific shape of the wear area based on these gray value change data. For example, if a surface crack in a certain area appears as a significant drop in gray value in the image, the system can infer the direction, length, and morphology of the crack based on the gray value changes and map this information to the preliminary model to provide more accurate shape data for subsequent repair design. When entering step S5014, the system combines the depth information of the wear with the wear shape to further trim the preliminary model. In this example, the system determines the depth of the wear area to be approximately 0.2mm based on the previous calculation of the gray value changes. To accurately repair, the system will further adjust the preliminary model based on this depth value, filling the wear area with a compensation structure to restore the repair area to the same height as other areas. At this time, the system will generate a corresponding compensation area on the preliminary model based on the wear shape and depth, and the shape and depth of the compensation area will match the actual wear area to ensure that the generated repair model fully meets the actual needs.The final 3D printing model can accurately reflect the mold wear condition, and a precise repair workpiece can be printed by additive manufacturing technology (such as 3D printing), which will be used to replace the worn part of the mold, ensuring the long-term stable operation of the mold.

[0100] In an embodiment, as shown in FIG. 5, the step S50 of generating and recording the corresponding failure range prediction result according to the wear analysis data association includes: Figure 6

[0101] S5021, obtaining mold design parameters; in this embodiment, the step of obtaining mold design parameters is first to extract relevant design data from the database of mold design, which may come from CAD drawings, mold design documents, process flow or production management system. For example, for injection mold, design parameters include the size of the cavity, the material of the mold, the cooling system design of the mold, the shape of the core, the structure of the runner, the setting of injection pressure and temperature, etc., which directly affect the performance and wear pattern of the mold. Assuming that the cavity design of the mold requires an accuracy of ±0.02mm, and the material used is H13 steel, these design data will be extracted and used as input for the subsequent prediction model, so that the system can accurately predict the type and range of wear according to the design requirements.

[0102] S5022, the mold design parameters and wear analysis data are input into the trained prediction model to generate corresponding trend results, including time trend and space trend; in this embodiment, in this step, the design parameters and wear analysis data are input into the trained prediction model, which is generally trained based on a large amount of historical data and machine learning algorithms (such as random forest, support vector machine, neural network, etc.) to predict the future failure behavior of the mold. The model will generate two main trend results: time trend and space trend according to the design characteristics and wear data of the mold. The time trend describes the change trend of wear on the time axis, for example, whether the wear degree of the mold increases with the increase of the number of uses, or whether there is a periodic wear phenomenon; the space trend describes how the wear of different regions changes in space during use, which regions may have a large range of wear due to high stress or high temperature. For example, based on the results of the previous analysis, the model may predict that in the next 5000 injection cycles, the wear of the cavity area will increase by 0.02mm each time, while the flow area may have concentrated wear at the inlet due to the influence of material flow.

[0103] ​S5023, according to the time trend, add time period hierarchical alarm process; in this embodiment, according to the time trend, the system will set different time period early warning mechanism for the wear trend, and then set hierarchical alarm. When the prediction model shows that the wear degree reaches a certain threshold, the system will trigger the alarm according to the set time period and threshold. For example, if the model predicts that the cavity area will wear more than 0.1mm in the next production cycle, the system will issue a warning at the previous stage before reaching the threshold, reminding the maintenance personnel to check and maintain in advance. The time period hierarchical alarm process may include multiple levels, such as reminding regular inspection for slight wear, suggesting maintenance for moderate wear, and warning that it may affect production and need to be repaired immediately for severe wear. This alarm process can make maintenance work in advance, avoiding the suddenness of mold failure.

[0104] S5024, according to the spatial trend, generate corresponding visual spatial features and push them to the corresponding management terminal. In this embodiment, according to the spatial trend, the system will convert the predicted wear area and its distribution on the mold into visual spatial feature map, and push these information to the management terminal through the graphical interface. For example, the model may predict that the inlet part of the runner area will be subject to greater wear, and a corner of the cavity may be subject to wear due to uneven cooling, and the system will generate a heat map or three-dimensional image to mark the risk level of these key areas. In the heat map, the areas with more serious wear are marked in red, and the areas with less serious wear are marked in yellow or green. These graphical data will be pushed to the terminal of the maintenance management personnel in real time, helping them to prioritize the processing of key areas in actual operation. The management personnel can intuitively understand the health status of the mold through these graphical spatial features, and develop more accurate maintenance and repair strategies.

[0105] It should be understood that the size of the serial number of each step in the above embodiment does not mean the order of execution, and the execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0106] In an embodiment, a mold intelligent fault prediction and diagnosis device is provided, which corresponds to the mold intelligent fault prediction and diagnosis method described above. As shown in the figure, the mold intelligent fault prediction and diagnosis device includes a first acquisition module, a second acquisition module, a matching module, a generation module and a judgment module. The function modules are described in detail as follows: Figure 9

[0107] The first acquisition module is used for acquiring mold production requirements, determining corresponding detection areas and precision thresholds of each detection area according to the mold production requirements;

[0108] ​The second acquisition module is configured to acquire a cavity planar image of the mold in the open state based on the industrial camera.

[0109] The matching module is configured to match an image region corresponding to the detection region in the cavity planar image based on the established visual recognition model.

[0110] The generating module is configured to analyze the image region, generate and convert wear analysis data in the same scale as the precision threshold.

[0111] The judging module is configured to judge whether the wear analysis data is greater than the precision threshold. If yes, a corresponding 3D printing model is established according to the wear analysis data, and a corresponding diagnostic workpiece is generated based on the additive manufacturing technology according to the 3D printing model. If no, a corresponding fault range prediction result is generated and recorded according to the wear analysis data.

[0112] Optionally, the first acquisition module comprises:

[0113] The pushing unit is configured to push a user input interface, the user input interface comprising a plurality of block regions, each of the block regions being provided with a plurality of interactive items, the interactive items comprising an index code, input explanation content and an interactive input box arranged in sequence, the input explanation content being used to explain information to be input in the interactive input box.

[0114] The first acquisition unit is configured to acquire input content of the interactive input box if a confirmation submission instruction of the user is detected.

[0115] The first determination unit is configured to determine input content of the same block region as detection content of the detection region according to matching of a matching field of the block region.

[0116] The first matching unit is configured to match a precision threshold corresponding to the detection content according to the index code, so as to serve as a precision threshold of the detection region.

[0117] Optionally, the matching module comprises:

[0118] The first comparison unit is configured to compare the cavity planar image and a calibrated overall template image based on the established visual recognition model, so as to divide a feature region, the feature region comprising a template region, a mold core region, a cavity region, a ejector pin region and a runner region.

[0119] The second matching unit is configured to match a corresponding feature region according to the index code.

[0120] The third matching unit is configured to match a corresponding frame selection coordinate according to the detection region.

[0121] a frame selection unit, configured to frame and zoom a corresponding region on the feature region as a corresponding image region of the measurement region based on the frame selection coordinates;

[0122] Optionally, the generating module comprises:

[0123] a fourth matching unit, configured to match a calibrated partial template image according to the detection region;

[0124] a second comparison unit, configured to compare the image region and the calibrated partial template image, determine a wear region in which color difference exists in the image region, and determine a wear position coordinate, a pixel unit area and a gray value change data of the wear region;

[0125] a second determination unit, configured to determine an image resolution of the industrial camera, and convert the pixel unit area of the wear region into a physical unit area according to a mapping relationship between the image resolution and an actual physical size;

[0126] a first calculation unit, configured to calculate a wear depth based on a calibrated relationship between the gray value change data and the actual wear depth;

[0127] a second calculation unit, configured to input the wear position coordinate, the physical unit area and the wear depth into a trained wear analysis model to calculate corresponding wear analysis data;

[0128] Optionally, the judging module comprises:

[0129] a calling unit, configured to call a basic model, the basic model comprising a template model, a mold core model, a cavity model, a thimble model and a runner model;

[0130] a first generating unit, configured to crop the basic model according to a wear position coordinate corresponding to the wear analysis data and the pixel unit area according to a preset rule to generate a corresponding preliminary model;

[0131] a third calculation unit, configured to calculate a wear shape according to a gray value change data corresponding to the wear analysis data;

[0132] a second generating unit, configured to further crop the preliminary model according to the wear shape and the wear depth to generate a corresponding 3D printing model;

[0133] Optionally, the judging module further comprises:

[0134] a second obtaining unit, configured to obtain a mold design parameter;

[0135] a third generating unit configured to input the mold design parameter and the wear analysis data into the trained prediction model to generate a corresponding trend result, the trend result including a time trend and a space trend;

[0136] an adding unit configured to add a time period classification alarm process according to the time trend;

[0137] a fourth generating unit configured to generate a corresponding visual space feature according to the space trend and push the visual space feature to a corresponding management terminal.

[0138] The specific limitations of the mold intelligent fault prediction and diagnosis device can be referred to the limitations of the mold intelligent fault prediction and diagnosis method described above, and will not be repeated here. Each module in the mold intelligent fault prediction and diagnosis device described above can be realized by software, hardware, and a combination thereof, in whole or in part. The above-mentioned modules can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory in the computer device in software form, so as to be called and executed by the processor to perform the operations corresponding to each module.

[0139] In one embodiment, a computer device is provided, which can be a server, and an internal structure diagram thereof can be as shown in Figure 10 The computer device includes a processor, a memory, a network interface, and a database connected through a system bus. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The network interface of the computer device is configured to communicate with an external terminal through a network connection. The computer program is executed by the processor to implement a mold intelligent fault prediction and diagnosis method.

[0140] In one embodiment, a computer device is provided, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the following steps when executing the computer program:

[0141] S10, obtaining mold production requirements, and determining corresponding detection areas and precision thresholds of each detection area according to the mold production requirements;

[0142] S20, obtaining a cavity plane image of a mold in an open state based on an industrial camera;

[0143] S30, matching an image region corresponding to a detection area in the cavity plane image based on the established visual recognition model;

[0144] S40. Analyze the image region to generate and convert wear analysis data at the same scale as the accuracy threshold;

[0145] S50. Determine whether the wear analysis data is greater than the accuracy threshold. If so, establish a corresponding 3D printing model based on the wear analysis data. Based on additive manufacturing technology, generate a corresponding diagnostic workpiece based on the 3D printing model. If not, generate and record the corresponding fault range prediction results based on the wear analysis data.

[0146] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0147] S10. Obtain mold production requirements, and determine the corresponding inspection areas and the accuracy threshold for each inspection area based on the mold production requirements.

[0148] S20. Using an industrial camera, acquire a planar image of the mold cavity in the open state;

[0149] S30. Based on the established visual recognition model, match the image region corresponding to the detection region in the cavity plane image;

[0150] S40. Analyze the image region to generate and convert wear analysis data at the same scale as the accuracy threshold;

[0151] S50. Determine whether the wear analysis data is greater than the accuracy threshold. If so, establish a corresponding 3D printing model based on the wear analysis data. Based on additive manufacturing technology, generate a corresponding diagnostic workpiece based on the 3D printing model. If not, generate and record the corresponding fault range prediction results based on the wear analysis data.

[0152] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiment methods. Any reference to memory, storage, database or other medium used in the embodiments provided by the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0153] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above-mentioned division of functional units and modules is exemplified, and in actual application, the above-mentioned functions can be completed by different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the above-described functions.

[0154] The above-mentioned embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A method for intelligent fault prediction and diagnosis of molds, characterized in that, The intelligent mold fault prediction and diagnosis method includes: Obtain the mold production requirements, and determine the corresponding detection areas and the accuracy threshold for each detection area based on the mold production requirements; Using an industrial camera, acquire planar images of the mold cavity when it is open; Based on the established visual recognition model, the image region corresponding to the detection region is matched in the cavity plane image; The image region is analyzed to generate and convert wear analysis data at the same scale as the accuracy threshold; If the wear analysis data is greater than the accuracy threshold, a corresponding 3D printing model is established based on the wear analysis data. Based on additive manufacturing technology, a corresponding diagnostic workpiece is generated based on the 3D printing model. If not, the corresponding fault range prediction result is generated and recorded based on the wear analysis data. The step of analyzing the image region and generating and converting wear analysis data at the same scale as the accuracy threshold includes: Based on the detection area, match the calibrated portion of the template image; By comparing the image region with the calibrated partial template image, the areas in the image region that have color differences are identified as wear areas, and the wear location coordinates, pixel unit area, and grayscale value change data of the wear areas are determined. The image resolution of the industrial camera is determined, and the pixel unit area of ​​the wear area is converted into physical unit area based on the mapping relationship between the image resolution and the actual physical size. The wear depth is calculated based on the calibration relationship between grayscale value change data and actual wear depth; The wear location coordinates, physical unit area, and wear depth are all substituted into the trained wear analysis model to calculate the corresponding wear analysis data.

2. The intelligent fault prediction and diagnosis method for molds according to claim 1, characterized in that, The step of obtaining mold production requirements and determining the corresponding detection area and the accuracy threshold of each detection area based on the mold production requirements includes: A user input interface is pushed out. The user input interface includes multiple block areas. Each block area has multiple interactive items. The interactive items include index codes, input explanation content and interactive input boxes arranged in sequence. The input explanation content is used to explain the information that needs to be entered in the interactive input box. If a user's confirmation submission command is detected, the input content of the interactive input box is obtained; Based on the matching field of the block region, the corresponding detection region is matched, and the input content of the same block region is determined as the detection content of the detection region. Based on the index code, a precision threshold corresponding to the detected content is matched and used as the precision threshold for the detected region.

3. The intelligent fault prediction and diagnosis method for molds according to claim 2, characterized in that, The step of matching the image region corresponding to the detection region in the cavity plane image based on the established visual recognition model includes: Based on the established visual recognition model, the cavity planar image and the calibrated overall template image are compared to divide the feature regions, which include the template region, mold core region, cavity region, ejector pin region and runner region. Match the corresponding feature region based on the index code; Based on the detection area, match the corresponding bounding box coordinates; Based on the selected coordinates, a region is selected and enlarged on the feature region to serve as the image region corresponding to the measured region.

4. The intelligent mold fault prediction and diagnosis method according to claim 1, characterized in that, The step of establishing a corresponding 3D printing model based on the wear analysis data includes: Call the basic model, which includes the template model, mold core model, cavity model, ejector pin model, and flow channel model; Based on the wear location coordinates corresponding to the wear analysis data and the pixel unit area, the basic model is cropped according to preset rules to generate a corresponding preliminary model. The wear shape is calculated based on the grayscale value change data corresponding to the wear analysis data; Based on the wear shape and wear depth, the preliminary model is further trimmed to generate the corresponding 3D printed model.

5. The intelligent fault prediction and diagnosis method for molds according to claim 1, characterized in that, The step of generating and recording the corresponding fault range prediction results based on the wear analysis data includes: Obtain mold design parameters; Substitute the mold design parameters and the wear analysis data into the trained prediction model to generate corresponding trend results, which include time trends and spatial trends. Based on the aforementioned time trend, a time-based hierarchical alarm process is added; Based on the spatial trends, corresponding visual spatial features are generated and pushed to the corresponding management terminal.

6. A mold intelligent fault prediction and diagnosis device, characterized in that, The intelligent mold fault prediction and diagnosis device includes: The first acquisition module is used to acquire mold production requirements and determine the corresponding detection area and the accuracy threshold of each detection area based on the mold production requirements. The second acquisition module is used to acquire a planar image of the cavity of the mold in the open state based on an industrial camera; The matching module is used to match the image region corresponding to the detection region in the cavity plane image based on the established visual recognition model; The generation module is used to analyze the image region, generate and convert wear analysis data at the same scale as the accuracy threshold; The judgment module is used to determine whether the wear analysis data is greater than the accuracy threshold. If it is, a corresponding 3D printing model is established based on the wear analysis data. Based on additive manufacturing technology, a corresponding diagnostic workpiece is generated based on the 3D printing model. If not, the corresponding fault range prediction result is generated and recorded based on the wear analysis data. The generation module includes: The fourth matching unit is used to match a calibrated portion of the template image based on the detection area; The second comparison unit is used to compare the image region and the calibrated partial template image, identify the region with color difference in the image region as the wear region, and determine the wear location coordinates, pixel unit area and gray value change data of the wear region. The second determining unit is used to determine the image resolution of the industrial camera and, based on the mapping relationship between the image resolution and the actual physical size, convert the pixel unit area of ​​the wear area into the physical unit area. The first calculation unit is used to calculate the wear depth based on the calibration relationship between grayscale value change data and actual wear depth; The second calculation unit is used to input the wear location coordinates, physical unit area, and wear depth into the trained wear analysis model to calculate the corresponding wear analysis data.

7. The intelligent mold fault prediction and diagnosis device according to claim 6, characterized in that, The first acquisition module includes: A push unit is used to push a user input interface, which includes multiple block areas. Each block area has multiple interactive items. The interactive items include an index code, an input explanation, and an interactive input box arranged in sequence. The input explanation is used to explain the information that needs to be entered in the interactive input box. The first acquisition unit is used to acquire the input content of the interactive input box if a user's confirmation submission instruction is detected. The first determining unit is used to match the corresponding detection region according to the matching field of the block region, and determine the input content of the same block region as the detection content of the detection region. The first matching unit is used to match the precision threshold corresponding to the detected content according to the index code, so as to use it as the precision threshold of the detection area; The matching module includes: The first comparison unit is used to compare the cavity planar image and the calibrated overall template image based on the established visual recognition model to divide the feature regions, which include the template region, mold core region, cavity region, ejector pin region and flow channel region. The second matching unit is used to match the corresponding feature region according to the index code; The third matching unit is used to match the corresponding bounding box coordinates based on the detection area; The selection unit is used to select and enlarge the corresponding area on the feature region based on the selection coordinates, so as to serve as the image region corresponding to the measurement region. The judgment module includes: The calling unit is used to call the basic model, which includes the template model, the core model, the cavity model, the ejector pin model, and the flow channel model; The first generation unit is used to generate a corresponding preliminary model by cropping the basic model according to preset rules based on the wear location coordinates corresponding to the wear analysis data and the pixel unit area; The third calculation unit is used to calculate the wear shape based on the gray value change data corresponding to the wear analysis data; The second generation unit is used to further trim the preliminary model according to the wear shape and the wear depth to generate the corresponding 3D printing model; The judgment module also includes: The second acquisition unit is used to acquire mold design parameters; The third generation unit is used to input the mold design parameters and the wear analysis data into the trained prediction model to generate corresponding trend results, including time trends and spatial trends. An add unit is used to add a time-based hierarchical alarm process based on the time trend; The fourth generation unit is used to generate corresponding visualized spatial features based on the spatial trend and push them to the corresponding management terminal.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the intelligent mold fault prediction and diagnosis method as described in any one of claims 1 to 5.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the intelligent mold fault prediction and diagnosis method as described in any one of claims 1 to 5.

Images