Material simulation test method and platform for mold flow analysis

Abstract

The invention provides a material simulation test method and platform for mold flow analysis, and relates to the technical field of material simulation testing, and the method comprises the steps: carrying out the product demand analysis of a target test material, and obtaining the product demand information; performing injection molding analysis to obtain an injection molding control parameter threshold; obtaining an injection molding simulation test model; carrying out an injection molding simulation test, and carrying out image pre-acquisition to obtain a first pre-acquired image; defect feature analysis is carried out, attention area distribution is carried out, and a target attention area is obtained; an injection molding defect area image is obtained, and defect feature information is obtained; and feedback testing and parameter optimization are carried out, optimized injection molding control parameters are obtained, and injection molding control is carried out. The technical problems that in the prior art, injection molding analysis generally depends on a fixed mathematical model and limited technological parameters, and problems in the molding process cannot be recognized and adjusted in real time, so that the production efficiency and the product quality are low are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of material simulation testing, and in particular to a material simulation testing method and platform for mold flow analysis. Background Art

[0002] Mold Flow Analysis (MFA), as a technology that simulates the flow, filling and solidification process of plastic melt in the mold, is usually used in combination with the actual injection molding process to optimize process parameters and predict possible molding defects. However, traditional mold flow analysis still faces many challenges in practical applications. Specifically, in the injection molding process, although traditional mold flow analysis technology can predict the behavior of materials during the molding process to a certain extent, it relies on fixed mathematical models and limited process parameters, making it inefficient and easily affected by human factors. It lacks automation and intelligence, and even if defects are detected, it is often impossible to accurately determine the cause of the defect, which makes it difficult to make timely process adjustments, resulting in low production efficiency and product quality. Summary of the invention

[0003] This application provides a material simulation test method and platform for mold flow analysis, aiming to solve the technical problem that the existing technology for injection molding analysis usually relies on fixed mathematical models and limited process parameters, and is unable to identify and adjust problems in the molding process in real time, resulting in low production efficiency and product quality.

[0004] The first aspect disclosed in the present application provides a material simulation test method for mold flow analysis, the method comprising: performing a product demand analysis on a target test material to obtain product demand information, wherein the product demand information includes surface finish, warpage, dimensional stability, and molding cycle; performing an injection molding analysis on the product demand information based on the material property information of the target test material to obtain an injection molding control parameter threshold; simulating the injection molding environment of the target test material to generate a simulated test environment, and combining the injection molding control parameter threshold to obtain an injection molding simulation test model; randomly selecting a first injection molding control parameter within the injection molding control parameter threshold, and performing the injection molding simulation test on the target test material through the injection molding simulation test model. The invention relates to an injection molding simulation test, and after the test is completed, an image of the target test material is pre-captured to obtain a first pre-captured image; based on a pre-constructed image defect feature comparison model, a defect feature analysis is performed on the first pre-captured image, and attention area distribution is performed based on the defect feature analysis result to obtain a target attention area; targeted image acquisition is performed based on the target attention area to obtain an injection molding defect area image, and similarity matching is performed in an injection molding defect feature library based on the injection molding defect area image to obtain defect feature information; based on the defect feature information, feedback testing and parameter optimization of the injection molding simulation test model are performed to obtain optimized injection molding control parameters, and injection molding control of the target test material is performed.

[0005] In a second aspect disclosed in the present application, a material simulation test platform for mold flow analysis is provided, and the platform is used for the material simulation test method for mold flow analysis mentioned above, and the platform includes: a product demand analysis module, and the product demand analysis module is used to perform product demand analysis on the target test material and obtain product demand information, wherein the product demand information includes surface finish, warpage, dimensional stability, and molding cycle; an injection molding analysis module, and the injection molding analysis module is used to perform injection molding analysis on the product demand information based on the material characteristic information of the target test material and obtain injection molding control parameter thresholds; a test model construction module, and the test model construction module is used to simulate the injection molding environment of the target test material, generate a simulated test environment, and obtain an injection molding simulation test model in combination with the injection molding control parameter thresholds; a simulation test module, and the simulation test module is used to randomly select a first injection molding control parameter within the injection molding control parameter threshold, and obtain an injection molding simulation test model through the injection molding control parameter thresholds. The injection molding simulation test model is used to perform an injection molding simulation test on the target test material, and after the test is completed, an image pre-collection of the target test material is performed to obtain a first pre-collected image; an attention area distribution module is used to perform defect feature analysis on the first pre-collected image based on a pre-constructed image defect feature comparison model, and perform attention area distribution based on the defect feature analysis result to obtain a target attention area; a similarity matching module is used to perform targeted image collection based on the target attention area to obtain an injection molding defect area image, and perform similarity matching in an injection molding defect feature library based on the injection molding defect area image to obtain defect feature information; an injection molding control module is used to perform feedback testing and parameter optimization of the injection molding simulation test model based on the defect feature information, obtain optimized injection molding control parameters, and perform injection molding control of the target test material.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] By analyzing the target materials and product requirements, key indicators such as surface finish, warpage, dimensional stability and molding cycle are clarified, and the control parameter threshold of injection molding is obtained in combination with material properties. This ensures that the injection molding test is based on sufficient data and analysis, avoiding blind trials in traditional process design; an injection molding simulation model is constructed, and simulation tests and optimization are carried out before actual production. The generation of the simulation model is based on the actual injection molding environment, which can accurately simulate the behavior of the material under different process conditions. Through the simulation test model, potential defects can be detected in a virtual environment, reducing the number of trial and error in actual production, improving production efficiency, and reducing production costs. ; A convolutional neural network is used to pre-build a defect feature comparison model, and the target attention area is identified in the image collected based on the simulation test results. The target attention area is matched with the defect feature similarity in the feature library, and the defect features in the molding process are automatically detected and identified. This automated image processing and defect detection technology improves the efficiency and accuracy of defect detection and avoids errors in manual detection. Through closed-loop feedback testing and parameter optimization mechanism, the process parameters can be optimized according to the feedback results of defect features, which means that the key parameters in injection molding can be automatically adjusted to obtain the best molding effect, realizing real-time and automated process parameter optimization and greatly reducing human intervention.

[0008] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A schematic diagram of a material simulation test method for mold flow analysis provided in an embodiment of the present application;

[0010] Figure 2 A schematic diagram of the structure of a material simulation test platform for mold flow analysis provided in an embodiment of the present application.

[0011] Explanation of the accompanying drawings: product demand analysis module 10, injection molding analysis module 20, test model construction module 30, simulation test module 40, attention area distribution module 50, similarity matching module 60, injection molding control module 70. DETAILED DESCRIPTION

[0012] The embodiments of the present application provide a material simulation test method and platform for mold flow analysis, thereby solving the technical problem that the prior art generally relies on fixed mathematical models and limited process parameters for injection molding analysis, and is unable to identify and adjust problems in the molding process in real time, resulting in low production efficiency and product quality.

[0013] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically introduced below in conjunction with the accompanying drawings of the specification. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0014] Embodiment 1, as Figure 1 As shown, an embodiment of the present application provides a material simulation test method for mold flow analysis, the method comprising:

[0015] Perform product demand analysis on the target test material to obtain product demand information, wherein the product demand information includes surface finish, warpage, dimensional stability, and molding cycle.

[0016] First, the target material for the test should be clearly defined, which can be a certain plastic material, composite material or other material used for injection molding. The design requirements of the product, including surface finish, warpage, dimensional stability, and molding cycle, are obtained through tools such as product specification sheets. Surface finish, such as smooth, matte or specific surface texture, can be achieved through different molding processes, mold polishing and other methods. Influencing factors include material fluidity, mold temperature, etc.; warpage refers to the deformation of the material caused by uneven stress distribution after injection molding. The control of warpage can be optimized by adjusting the mold temperature, holding pressure and injection speed; Dimensional stability is manifested in the change of product size, which is inevitable in the injection molding process. Therefore, it is necessary to clarify the dimensional accuracy requirements of the finished product, which can be optimized by analyzing the shrinkage rate of the material, mold design and process control parameters; molding cycle refers to the entire time period from the injection of the material into the mold to the cooling and removal of the finished product in injection molding. In order to improve production efficiency, the molding cycle is usually required to be as short as possible, but a balance must be struck between short cycle and product quality.

[0017] Based on the material property information of the target test material, an injection molding analysis is performed on the product demand information to obtain an injection molding control parameter threshold.

[0018] Obtain material property information of the target test material, including material physical properties, material mechanical properties, material thermal properties, etc. Among them, material physical properties include density, viscosity, melt index, glass transition temperature, thermal expansion coefficient, etc., which will affect the fluidity and cooling characteristics of the material during the injection molding process; material mechanical properties such as elastic modulus, tensile strength, impact strength, etc. are used to evaluate the strength performance of the material after molding; material thermal properties include material thermal conductivity, heat resistance, thermal stability, etc., which affect the control of cooling time and mold temperature during the molding process.

[0019] According to product requirements, multiple evaluation indicators are determined. Each indicator is used to analyze the performance of materials under different molding conditions. Through historical molding data, a multidimensional evaluation network model is trained in combination with multiple evaluation indicators to evaluate the impact of different injection molding conditions on product effects. The evaluation network is incrementally trained in combination with material property information to ensure that the network can accurately evaluate the molding effect of specific materials under different conditions. According to the evaluation network after incremental training, product demand information is analyzed to obtain the control threshold range of key process parameters related to injection molding, including injection speed, mold temperature, injection pressure, holding time, etc., to obtain the injection molding control parameter threshold.

[0020] The injection molding environment of the target test material is simulated to generate a simulated test environment, and an injection molding simulation test model is obtained in combination with the injection molding control parameter threshold.

[0021] Use injection molding simulation software to simulate the real injection molding environment, import the physical, mechanical and thermal performance parameters of the target test material into the simulation software, so as to accurately simulate the behavior of the material during the injection molding process, and set the mold parameters and process conditions in the simulation according to the actual mold design, such as gate location, mold structure, etc., to generate a simulated test environment.

[0022] According to the injection molding control parameter threshold, the boundary conditions of injection molding are set, including injection speed, mold temperature, injection pressure, holding time, etc., to ensure that the simulation environment is consistent with the actual injection molding process, and an injection molding simulation test model is generated. In this simulation test model, the simulation software will simulate the entire molding process from the molten state to the solidification of the material through finite element analysis to provide guidance for actual production.

[0023] A first injection molding control parameter is randomly selected within the injection molding control parameter threshold, an injection molding simulation test of the target test material is performed through the injection molding simulation test model, and an image of the target test material is pre-captured after the test is completed to obtain a first pre-captured image.

[0024] Within the injection molding control parameter threshold, a parameter combination is randomly selected, such as an injection temperature of 240°C, a mold temperature of 90°C, an injection speed of 50mm / s, etc., as the first injection molding control parameter. The injection molding simulation test model is used to perform an injection molding simulation test on the target test material under the first injection molding control parameter. The simulation software simulates the material flow, filling, cooling, and curing process during the injection molding process. After the simulation test, the finished product generated by the simulation is pre-captured in a virtual manner. The first pre-captured image includes the morphology of the material surface and possible defect locations, such as surface scratches, dents, bubbles, flow marks, etc., for subsequent defect analysis.

[0025] Based on the pre-constructed image defect feature comparison model, defect feature analysis is performed on the first pre-collected image, and attention area distribution is performed based on the defect feature analysis result to obtain the target attention area.

[0026] The image defect feature comparison model uses a convolutional neural network to perform image feature extraction and defect detection. It specifically includes a standard feature extraction unit, a defect feature extraction unit, and a loss comparison analysis unit. By performing loss analysis on standard features and defect features, potential defect areas in the image are identified and marked as target attention areas. These areas are the focus of the image that require further detailed detection and analysis.

[0027] Targeted image acquisition is performed based on the target attention area to obtain an injection molding defect area image, and similarity matching is performed in an injection molding defect feature library based on the injection molding defect area image to obtain defect feature information.

[0028] High-resolution image acquisition is performed on the target attention area to capture more detailed information and generate injection molding defect area images containing defect features. These images have higher resolution and details than pre-acquired images, which is convenient for subsequent defect analysis. Multi-scale feature extraction is performed on the acquired defect area images to extract defect information at different scales. Based on the extracted multi-scale defect features, a similarity algorithm is used to perform traversal matching in the pre-built injection molding defect feature library. The defect feature library is established by accumulating a large number of defect cases encountered in actual production, including common injection molding defects such as bubbles, short shots, weld marks, warpage, etc. Through traversal matching, a similarity set is obtained, and the similarity set is sorted from high to low according to the similarity. The result with the highest similarity is selected to determine the type of the defect as the defect type of the target attention area. The defect feature information is obtained by combining the defect size and defect location.

[0029] Based on the defect feature information, feedback testing and parameter optimization of the injection molding simulation test model are performed to obtain optimized injection molding control parameters, and injection molding control of the target test material is performed.

[0030] Correlation analysis is performed based on defect feature information and injection molding control parameters. For example, warpage defects may be related to uneven mold cooling or too fast injection speed. Through correlation analysis, it is clear which parameters affect the defect characteristics and determine the direction of control parameters that need to be optimized. According to the optimization direction, the parameters in the simulation model are reset and retested in the simulation model to verify whether the new parameter combination can reduce or eliminate defects. Simulation tests are performed multiple times, and each injection molding parameter is adjusted in turn. By comparing the impact of each parameter on the molding results, the optimal parameter combination is found as the optimized injection molding control parameter.

[0031] Finally, the optimized injection molding control parameters are applied in actual injection molding production to ensure that the target test material meets the product requirements during the injection molding process, reduce or eliminate the generation of defects, and achieve improved production efficiency and finished product quality.

[0032] Furthermore, the method of performing injection molding analysis on the product demand information based on the material property information of the target test material to obtain the injection molding control parameter threshold includes:

[0033] Determine multiple injection molding effect evaluation indicators according to the product demand information; perform injection molding data mining to build an injection molding database, perform classification and identification training on the injection molding database based on the multiple injection molding effect evaluation indicators to obtain a multi-dimensional injection molding effect evaluation network; perform incremental training on the multi-dimensional injection molding effect evaluation network based on the material characteristic information to obtain an updated multi-dimensional injection molding effect evaluation network; perform injection molding control analysis on the product demand information based on the updated multi-dimensional injection molding effect evaluation network to obtain an injection molding control parameter threshold.

[0034] According to the product demand information, multiple indicators for evaluating the injection molding effect are determined, including surface finish, which is the product's demand for surface smoothness, such as no scratches or defects on the surface; warpage, which is the degree of deformation of the product after molding, directly affects the dimensional stability of the product; dimensional stability, which refers to whether the product can maintain the designed dimensional accuracy after injection molding, including key dimensions such as length, width, and height; molding cycle, which is the time from the injection of the material into the mold to the final cooling and solidification. According to the production efficiency requirements of the product, it is usually hoped that the molding cycle is as short as possible without affecting the product quality. Quantify each evaluation indicator and assign a weight to each indicator based on the importance of the product requirements. For example, some products may have extremely high requirements for warpage but lower requirements for surface finish.

[0035] Collect historical data and process data related to injection molding, including the relationship between different materials, molds, process parameters and finished product performance, and integrate the data to build an injection molding database. Classify and annotate the data in the database. For example, according to indicators such as surface finish, warpage deformation, dimensional stability, etc., classify the finished products into qualified, unqualified or other categories, and label them. Through machine learning algorithms such as neural networks and support vector machines, classify and train the annotated data. The goal is to build a classification model that can automatically identify and evaluate molding effects, and obtain a multi-dimensional injection molding effect evaluation network, which is used to predict molding effects under different process conditions.

[0036] Incremental training refers to updating and optimizing the model based on the existing model using new data, namely material property information, rather than training the model from scratch. Specifically, the material property information is combined with the existing injection molding process parameters and molding effect data to generate a new training data set. The new training data is used to perform incremental training on the multi-dimensional injection molding effect evaluation network. In this way, the model can be further optimized on the original basis to improve its adaptability to the target test materials and process conditions. After the incremental training is completed, the updated multi-dimensional injection molding effect evaluation network can be used to process the material type of the target test material, making the prediction effect more accurate.

[0037] The product demand information is input into the updated multi-dimensional injection molding effect evaluation network. The network automatically performs injection molding control analysis based on the input product demand. Each demand information corresponds to certain injection molding process parameters, such as injection speed, mold temperature, holding time, etc. The network automatically analyzes the impact of different injection molding process parameters on product demand through deep learning, and finally outputs the threshold range of key control parameters of injection molding to generate injection molding control parameter thresholds.

[0038] Furthermore, the method further comprises:

[0039] Injection molding equipment information is acquired, wherein the injection molding equipment information includes equipment control constraint parameters; and value constraints are imposed on injection molding control parameter thresholds based on the equipment control constraint parameters.

[0040] Determine the model and type of injection molding equipment used, and obtain equipment information through the equipment technical manual or equipment control panel, including the maximum operating range and performance limitations of the equipment, such as maximum injection pressure, maximum injection speed, mold temperature control range, machine response speed, etc.

[0041] Based on the equipment control constraint parameters, the actual value of the obtained injection molding control parameter threshold is constrained, that is, the physical limitations of the equipment will affect the selection range of the process parameters. For example, the maximum injection pressure of the equipment limits the injection pressure value during molding. If the injection pressure obtained in the simulation analysis exceeds the maximum injection pressure of the equipment, the pressure needs to be adjusted to ensure that the equipment can reach the required injection pressure in actual production. By adjusting the process parameters through constraints, it is ensured that the adjusted parameters not only meet the equipment constraints, but also can be close to the best molding effect in the simulation.

[0042] Furthermore, the method of performing defect feature analysis on the first pre-collected image based on the pre-built image defect feature comparison model, and distributing attention areas based on the defect feature analysis results to obtain the target attention area includes:

[0043] Determine a product standard image according to the product demand information; build the image defect feature comparison model based on a convolutional neural network, wherein the image defect feature comparison model includes a standard feature extraction unit, a defect feature extraction unit, and a loss comparison analysis unit, and the standard feature extraction unit and the defect feature extraction unit have the same structure and share weights; based on the standard feature extraction unit and the defect feature extraction unit, perform feature extraction on the product standard image and the first pre-collected image respectively, and based on the loss comparison analysis unit, perform pixel-level loss comparison analysis on the feature extraction result to obtain an image loss coefficient set; based on the image loss coefficient set, perform loss pixel identification in the first pre-collected image to obtain the target attention area.

[0044] Obtain standard products that meet all required information from the production line or design stage, and take or scan high-quality images of the products to generate product standard images as a benchmark. All subsequent comparison and analysis will use this image as a reference.

[0045] Convolutional neural network is a deep learning model used to process image data. It extracts the feature information of the image through structures such as convolutional layer, pooling layer and fully connected layer. The image defect feature comparison model is built based on convolutional neural network. The model consists of three core units, including standard feature extraction unit, defect feature extraction unit and loss comparison analysis unit. Among them, the standard feature extraction unit uses convolutional layer and pooling layer to process the input standard product image, extract high-level image features such as shape, texture, boundary, etc., and generate feature vectors of standard images for comparison with the features of defect images; the defect feature extraction unit has the same structure as the standard feature extraction unit, including convolutional layer and pooling layer, which is used to process the image to be inspected. In order to ensure the comparability of the two sets of features, this unit shares the same network weights with the standard feature extraction unit. This unit generates feature vectors of defect images for comparison with the features of standard images; the loss comparison analysis unit is used to compare standard features with defect features to identify potential defect areas.

[0046] Among them, the standard feature extraction unit and the defect feature extraction unit have the same structure and share the same weights. This weight sharing mechanism ensures the symmetry of the two units when extracting features, so that the model can more accurately compare the differences between the standard image and the image to be inspected. Sharing weights not only reduces the number of model parameters and improves training efficiency, but also enhances the model's comparison ability when detecting similar features.

[0047] The standard feature extraction unit extracts high-level features from the product standard image, and uses convolutional layers and pooling layers to extract key feature information of the image, including texture, edge, shape, etc., and converts the extracted features into feature vectors to represent the feature structure of the image; the defect feature extraction unit extracts features from the first pre-collected image in the same way as the standard feature extraction unit. Through this unit, feature information in the pre-collected image is extracted and a corresponding feature vector is generated.

[0048] The loss comparison analysis unit compares the feature vector of the standard image and the feature vector of the pre-collected image pixel by pixel and calculates the difference between them. This process uses a loss function to measure the difference between the two images in the feature space and converts the difference of each pixel into a loss coefficient through the loss function. The larger the loss coefficient, the greater the deviation between the pixel and the standard image, and the defect may exist. Through the pixel-by-pixel loss calculation, a set of image loss coefficients is generated, and each loss coefficient corresponds to the position of a pixel in the image.

[0049] According to the preset image loss coefficient threshold, abnormal pixels are screened out in the image loss coefficient set and marked on the first pre-acquired image. When marking pixels, if the loss coefficients of multiple adjacent pixels exceed the threshold, they are identified as the same defect area. These adjacent loss pixels are connected through clustering or morphological operations to form target attention areas. These areas are the areas in the pre-acquired image that need further focus and analysis and may have surface defect problems.

[0050] Furthermore, the method of identifying loss pixels in the first pre-captured image based on the image loss coefficient set to obtain the target attention area includes:

[0051] Preset an image loss coefficient threshold, traverse the image loss coefficient set based on the image loss coefficient threshold, and obtain an abnormal image loss coefficient set, wherein the abnormal image loss coefficient set has a loss pixel coordinate set; identify loss pixels based on the loss pixel coordinate set, and connect the identified pixels to obtain the target attention area.

[0052] A threshold value of image loss coefficient is preset as a standard for judging whether it is an abnormal pixel. It can be determined based on experimental data or specific quality requirements. Pixels exceeding this threshold value are judged to be defective, and pixels below this threshold value are judged to be normal areas.

[0053] The image loss coefficient set is traversed based on the image loss coefficient threshold. During the traversal process, the loss coefficient of each pixel is compared with the preset threshold. If the loss coefficient of a pixel exceeds the threshold, the pixel is judged to be abnormal and is an abnormal loss pixel. By traversing all pixel loss coefficients, all pixels exceeding the threshold are screened out, and the loss coefficients and corresponding coordinate positions of these pixels are recorded to form an abnormal image loss coefficient set. The loss pixel coordinate set can be used to accurately locate areas where defects may exist.

[0054] Using the loss pixel coordinate set, each abnormal pixel is identified in the first pre-acquired image, and the adjacent loss pixels are connected. Morphological operations in image processing, such as dilation and erosion, can be used to connect adjacent abnormal pixels together to form a complete area. By connecting multiple abnormal pixel points, the boundary of the defective area is determined to form a coherent area as the target attention area. The shape, size and position of this area reflect the scope of the potential defect.

[0055] Furthermore, the method of performing similarity matching in an injection molding defect feature library based on the injection molding defect area image to obtain defect feature information includes:

[0056] Perform multi-scale defect feature extraction on the injection molding defect area image to obtain multi-scale defect features; perform similarity calculation and traversal in the injection molding defect feature library based on the multi-scale defect features to obtain a similarity set; sort the similarity set from high to low, obtain the injection molding defect feature corresponding to the maximum similarity according to the sorting result, and obtain the defect type; obtain the defect area position and the defect area size according to the loss pixel coordinate set; and determine the defect feature information in combination with the defect type, defect area position and defect area size.

[0057] Multi-scale feature extraction is a commonly used technology in image processing. It refers to analyzing and processing images at different scales, such as size and detail level. This can capture feature information of various sizes and complexities in the image. In injection molding defect detection, defects may have different sizes and shapes, so it is necessary to capture these features through multi-scale methods to avoid focusing on a single scale and missing important information. For example, the image pyramid technology is used to extract image features at different resolutions by downsampling the image multiple times. Each level corresponds to a different scale, which can capture the features of large-scale and small-scale defects. The obtained multi-scale defect features include geometric features, texture features, color features, etc.

[0058] The injection molding defect feature library is a database constructed from a large amount of known injection molding defect data. Each defect sample is marked and has a corresponding feature vector. These features include the type, shape, texture, etc. of the defect. The multi-scale defect features are used to calculate the similarity with the feature vectors in the feature library. For example, the cosine similarity method is used to judge the similarity by calculating the cosine value of the angle between two vectors. The closer the value is to 1, the more similar the two vectors are. Since the feature vector contains features at different scales, different weights can be assigned to features at different scales to ensure a more comprehensive similarity calculation. Each defect sample in the feature library is traversed and similarity calculation is performed to finally obtain a similarity set, in which each value represents the similarity between the current defect feature and a defect sample in the feature library. The larger the value, the higher the similarity.

[0059] The similarity set is sorted from high to low. After sorting, the defect sample corresponding to the maximum similarity is at the top of the sorted list, indicating that the current defect is most similar to the defect sample in the feature library. The type of the defect sample is directly output as the defect type of the current defect.

[0060] By analyzing the loss pixel coordinate set, the outermost boundary of the defect area is determined. Based on the boundary of the loss pixel, the size of the defect area, such as width and height, and the position of the defect area in the entire image are calculated. The position is usually represented by the coordinates of the center point of the area. The defect type, defect area location and defect area size are integrated to determine the defect feature information. The defect feature information includes the type, location and size of the defect, such as defect features such as warpage, bubbles or weld marks.

[0061] Furthermore, the method of performing feedback testing and parameter optimization of the injection molding simulation test model based on the defect feature information to obtain optimized injection molding control parameters includes:

[0062] A parameter association analysis is performed based on the defect feature information, and an optimization direction is determined within the injection molding control parameter threshold according to the parameter association analysis result; a feedback test of the injection molding simulation test model is performed according to the optimization direction, and a parameter comparison and optimization is performed during the feedback test to obtain the optimized injection molding control parameters.

[0063] By analyzing previous simulation test results, historical production data or process data in the literature, the relationship between different injection molding process parameters and defect types is analyzed. For example, a higher injection speed may lead to more obvious weld marks, while a lower injection speed may lead to short shot defects; a higher mold temperature helps reduce warping, but too high a temperature may lead to a long cooling time. For the identified related problems, the specific direction of adjustment is clarified, such as reducing the injection speed, increasing the mold temperature or adjusting the cooling time. Sometimes multiple parameters jointly affect a defect, so it is necessary to comprehensively consider the joint adjustment strategy of multiple parameters.

[0064] According to the determined optimization direction, the control parameters of the injection molding simulation test model are adjusted. After the parameters are adjusted, simulation tests are performed through simulation software to simulate the impact of the optimized process on the molding results. By comparing the molding quality of each simulation result, the parameters are gradually adjusted. After multiple feedback tests and parameter comparisons, the optimal optimized injection molding control parameters are determined. These parameters can minimize defects and improve production efficiency while ensuring product quality.

[0065] In summary, the material simulation test method for mold flow analysis provided in the embodiment of the present application has the following technical effects:

[0066] By analyzing the target materials and product requirements, key indicators such as surface finish, warpage, dimensional stability and molding cycle are clarified, and the control parameter threshold of injection molding is obtained in combination with material properties. This ensures that the injection molding test is based on sufficient data and analysis, avoiding blind trials in traditional process design; an injection molding simulation model is constructed, and simulation tests and optimization are carried out before actual production. The generation of the simulation model is based on the actual injection molding environment, which can accurately simulate the behavior of the material under different process conditions. Through the simulation test model, potential defects can be detected in a virtual environment, reducing the number of trial and error in actual production, improving production efficiency, and reducing production costs. ; A convolutional neural network is used to pre-build a defect feature comparison model, and the target attention area is identified in the image collected based on the simulation test results. The target attention area is matched with the defect feature similarity in the feature library, and the defect features in the molding process are automatically detected and identified. This automated image processing and defect detection technology improves the efficiency and accuracy of defect detection and avoids errors in manual detection. Through closed-loop feedback testing and parameter optimization mechanism, the process parameters can be optimized according to the feedback results of defect features, which means that the key parameters in injection molding can be automatically adjusted to obtain the best molding effect, realizing real-time and automated process parameter optimization and greatly reducing human intervention.

[0067] Embodiment 2 is based on the same inventive concept as the material simulation test method for mold flow analysis in the above embodiment. Figure 2As shown, the embodiment of the present application provides a material simulation test platform for mold flow analysis, the platform comprising:

[0068] A product demand analysis module 10, the product demand analysis module 10 is used to perform product demand analysis on the target test material and obtain product demand information, wherein the product demand information includes surface finish, warpage, dimensional stability, and molding cycle; an injection molding analysis module 20, the injection molding analysis module 20 is used to perform injection molding analysis on the product demand information based on the material property information of the target test material and obtain an injection molding control parameter threshold; a test model construction module 30, the test model construction module 30 is used to simulate the injection molding environment of the target test material, generate a simulated test environment, and obtain an injection molding simulation test model in combination with the injection molding control parameter threshold; a simulation test module 40, the simulation test module 40 is used to randomly select a first injection molding control parameter within the injection molding control parameter threshold, and perform an injection molding simulation of the target test material through the injection molding simulation test model. The test is performed, and after the test is completed, an image of the target test material is pre-collected to obtain a first pre-collected image; an attention area distribution module 50 is used to perform defect feature analysis on the first pre-collected image based on a pre-constructed image defect feature comparison model, and to distribute attention areas based on the defect feature analysis results to obtain a target attention area; a similarity matching module 60 is used to perform targeted image acquisition based on the target attention area to obtain an injection molding defect area image, and to perform similarity matching in an injection molding defect feature library based on the injection molding defect area image to obtain defect feature information; an injection molding control module 70 is used to perform feedback testing and parameter optimization of the injection molding simulation test model based on the defect feature information, to obtain optimized injection molding control parameters, and to perform injection molding control of the target test material.

[0069] Furthermore, the platform also includes an injection molding control parameter threshold acquisition module to perform the following operation steps:

[0070] Determine multiple injection molding effect evaluation indicators according to the product demand information; perform injection molding data mining to build an injection molding database, perform classification and identification training on the injection molding database based on the multiple injection molding effect evaluation indicators to obtain a multi-dimensional injection molding effect evaluation network; perform incremental training on the multi-dimensional injection molding effect evaluation network based on the material characteristic information to obtain an updated multi-dimensional injection molding effect evaluation network; perform injection molding control analysis on the product demand information based on the updated multi-dimensional injection molding effect evaluation network to obtain an injection molding control parameter threshold.

[0071] Furthermore, the platform also includes a value constraint module to perform the following operation steps:

[0072] Injection molding equipment information is acquired, wherein the injection molding equipment information includes equipment control constraint parameters; and value constraints are imposed on injection molding control parameter thresholds based on the equipment control constraint parameters.

[0073] Furthermore, the platform also includes a target attention area acquisition module to perform the following operation steps:

[0074] Determine a product standard image according to the product demand information; build the image defect feature comparison model based on a convolutional neural network, wherein the image defect feature comparison model includes a standard feature extraction unit, a defect feature extraction unit, and a loss comparison analysis unit, and the standard feature extraction unit and the defect feature extraction unit have the same structure and share weights; based on the standard feature extraction unit and the defect feature extraction unit, perform feature extraction on the product standard image and the first pre-collected image respectively, and based on the loss comparison analysis unit, perform pixel-level loss comparison analysis on the feature extraction result to obtain an image loss coefficient set; based on the image loss coefficient set, perform loss pixel identification in the first pre-collected image to obtain the target attention area.

[0075] Furthermore, the platform further includes a attention area acquisition module to perform the following operation steps:

[0076] Preset an image loss coefficient threshold, traverse the image loss coefficient set based on the image loss coefficient threshold, and obtain an abnormal image loss coefficient set, wherein the abnormal image loss coefficient set has a loss pixel coordinate set; identify loss pixels based on the loss pixel coordinate set, and connect the identified pixels to obtain the target attention area.

[0077] Furthermore, the platform also includes a defect feature information acquisition module to perform the following operation steps:

[0078] Perform multi-scale defect feature extraction on the injection molding defect area image to obtain multi-scale defect features; perform similarity calculation and traversal in the injection molding defect feature library based on the multi-scale defect features to obtain a similarity set; sort the similarity set from high to low, obtain the injection molding defect feature corresponding to the maximum similarity according to the sorting result, and obtain the defect type; obtain the defect area position and the defect area size according to the loss pixel coordinate set; and determine the defect feature information in combination with the defect type, defect area position and defect area size.

[0079] Furthermore, the platform also includes an optimized injection molding control parameter acquisition module to perform the following operation steps:

[0080] A parameter association analysis is performed based on the defect feature information, and an optimization direction is determined within the injection molding control parameter threshold according to the parameter association analysis result; a feedback test of the injection molding simulation test model is performed according to the optimization direction, and a parameter comparison and optimization is performed during the feedback test to obtain the optimized injection molding control parameters.

[0081] Through the above-mentioned detailed description of a material simulation test method for mold flow analysis in this specification, those skilled in the art can clearly know a material simulation test platform for mold flow analysis in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0082] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A material simulation test method for mold flow analysis, characterized in that: The method comprises: Perform product demand analysis on the target test material to obtain product demand information, wherein the product demand information includes surface finish, warpage, dimensional stability, and molding cycle; Performing injection molding analysis on the product demand information based on the material property information of the target test material to obtain injection molding control parameter thresholds; Simulating the injection molding environment of the target test material to generate a simulated test environment, and combining the injection molding control parameter threshold to obtain an injection molding simulation test model; randomly selecting a first injection molding control parameter within the injection molding control parameter threshold, performing an injection molding simulation test on the target test material through the injection molding simulation test model, and performing image pre-capture of the target test material after the test is completed to obtain a first pre-captured image; Based on a pre-built image defect feature comparison model, defect feature analysis is performed on the first pre-collected image, and attention area distribution is performed based on the defect feature analysis result to obtain a target attention area; Based on the target attention area, targeted image acquisition is performed to obtain an injection molding defect area image, and based on the injection molding defect area image, similarity matching is performed in an injection molding defect feature library to obtain defect feature information; Based on the defect feature information, feedback testing and parameter optimization of the injection molding simulation test model are performed to obtain optimized injection molding control parameters, and injection molding control of the target test material is performed.

2. A material simulation test method for mold flow analysis according to claim 1, characterized in that: The method of performing injection molding analysis on the product demand information based on the material property information of the target test material to obtain the injection molding control parameter threshold includes: Determine a plurality of injection molding effect evaluation indicators according to the product demand information; Performing injection molding data mining to construct an injection molding database, and performing classification and identification training on the injection molding database based on the multiple injection molding effect evaluation indicators to obtain a multi-dimensional injection molding effect evaluation network; Performing incremental training on the multi-dimensional injection molding effect evaluation network based on the material property information to obtain an updated multi-dimensional injection molding effect evaluation network; Based on the updated multi-dimensional injection molding effect evaluation network, injection molding control analysis is performed on the product demand information to obtain injection molding control parameter thresholds.

3. A material simulation test method for mold flow analysis as claimed in claim 2, characterized in that: The method further comprises: Acquiring injection molding equipment information, wherein the injection molding equipment information includes equipment control constraint parameters; The injection molding control parameter threshold is constrained based on the equipment control constraint parameter.

4. A material simulation test method for mold flow analysis according to claim 1, characterized in that: The method of performing defect feature analysis on the first pre-collected image based on the pre-built image defect feature comparison model, and distributing attention areas based on the defect feature analysis results to obtain the target attention area includes: Determine a product standard image according to the product demand information; The image defect feature comparison model is built based on a convolutional neural network, wherein the image defect feature comparison model includes a standard feature extraction unit, a defect feature extraction unit, and a loss comparison analysis unit, and the standard feature extraction unit and the defect feature extraction unit have the same structure and share weights; Based on the standard feature extraction unit and the defect feature extraction unit, feature extraction is performed on the product standard image and the first pre-collected image respectively, and based on the loss comparison and analysis unit, pixel-level loss comparison and analysis is performed on the feature extraction results to obtain an image loss coefficient set; Lost pixels are identified in the first pre-captured image based on the image loss coefficient set to obtain the target attention area.

5. A material simulation test method for mold flow analysis as claimed in claim 4, characterized in that: The method of identifying loss pixels in the first pre-collected image based on the image loss coefficient set to obtain the target attention area includes: Preset an image loss coefficient threshold, traverse the image loss coefficient set based on the image loss coefficient threshold, and obtain an abnormal image loss coefficient set, wherein the abnormal image loss coefficient set has a loss pixel coordinate set; Lost pixels are identified based on the lost pixel coordinate set, and the identified pixels are connected to obtain the target attention area.

6. A material simulation test method for mold flow analysis as claimed in claim 5, characterized in that: The method of performing similarity matching in an injection molding defect feature library based on the injection molding defect area image to obtain defect feature information includes: Extract multi-scale defect features from the injection molding defect area image to obtain multi-scale defect features; Based on the multi-scale defect features, similarity calculation and traversal are performed in the injection molding defect feature library to obtain a similarity set; Sorting the similarity sets from high to low, obtaining the injection molding defect feature corresponding to the maximum similarity according to the sorting result, and obtaining the defect type; Acquire the defect area position and defect area size according to the loss pixel coordinate set; The defect characteristic information is determined in combination with the defect type, defect area position and defect area size.

7. A material simulation test method for mold flow analysis according to claim 1, characterized in that: Based on the defect feature information, feedback testing and parameter optimization of the injection molding simulation test model are performed to obtain optimized injection molding control parameters, and the method includes: Performing parameter association analysis based on the defect feature information, and determining an optimization direction within the injection molding control parameter threshold according to the parameter association analysis result; A feedback test of the injection molding simulation test model is performed according to the optimization direction, and parameter comparison and optimization are performed during the feedback test to obtain the optimized injection molding control parameters.

8. A material simulation test platform for mold flow analysis, characterized in that: A material simulation test method for mold flow analysis according to any one of claims 1 to 7, wherein the platform comprises: A product demand analysis module, which is used to perform product demand analysis on target test materials and obtain product demand information, wherein the product demand information includes surface finish, warpage, dimensional stability, and molding cycle; An injection molding analysis module, the injection molding analysis module is used to perform injection molding analysis on the product demand information based on the material property information of the target test material, and obtain an injection molding control parameter threshold; A test model construction module, the test model construction module is used to simulate the injection molding environment of the target test material, generate a simulated test environment, and obtain an injection molding simulation test model in combination with the injection molding control parameter threshold; A simulation test module, wherein the simulation test module is used to randomly select a first injection molding control parameter within the injection molding control parameter threshold, perform an injection molding simulation test on the target test material through the injection molding simulation test model, and perform image pre-capture of the target test material after the test is completed to obtain a first pre-captured image; an attention region distribution module, the attention region distribution module being used to perform defect feature analysis on the first pre-collected image based on a pre-built image defect feature comparison model, and to perform attention region distribution based on the defect feature analysis result to obtain a target attention region; A similarity matching module, wherein the similarity matching module is used to perform targeted image acquisition based on the target attention area to obtain an injection molding defect area image, and perform similarity matching in an injection molding defect feature library based on the injection molding defect area image to obtain defect feature information; An injection molding control module is used to perform feedback testing and parameter optimization of the injection molding simulation test model based on the defect feature information, obtain optimized injection molding control parameters, and perform injection molding control of the target test material.

Images