Injection molding method and device based on plastic mold
By performing three-dimensional modeling and simulation analysis of injection molds, combined with adaptive slow feature analysis and sparse automatic encoder model, the injection mold cooling system is optimized, which solves the problem that traditional designs are difficult to meet the cooling needs of complex shape products, and achieves more efficient Cooling and more stable product quality.
Patent Information
- Application Number
- CN202410912966.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-09
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2044-07-09
AI Technical Summary
The traditional injection mold design and optimization process depends on experience and repeated trials, making it difficult to cope with the manufacturing challenges of complex shape products. Especially in the design of cooling systems, traditional linear cooling channels cannot meet the needs of fast and uniform cooling, resulting in product deformation and curling. Quality issues such as chords.
By performing three-dimensional modeling of plastic molds, including cavity, runner system and cooling channels, the injection molding process simulation analysis is carried out, temperature, pressure and shear stress distribution data are extracted, and adaptive slow feature analysis and sparse automatic encoder model are used. , calculate temperature uniformity and cooling efficiency indicators, optimize cooling channel parameters, and realize intelligent design and optimization of cooling system.
It improves the simulation accuracy of the injection molding process, effectively captures the essential dynamic characteristics during the injection molding process, optimizes the cooling channel design, achieves the improvement of temperature uniformity and cooling efficiency, and solves product deformation and quality problems.
Smart Images

Figure CN118536364B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of injection molding management technology, and in particular to an injection molding method and device based on a plastic mold. Background Art
[0002] The traditional injection mold design and optimization process mainly relies on engineers' experience and repeated trials, which is not only time-consuming and labor-intensive, but also difficult to cope with the manufacturing challenges of complex-shaped products. Especially in the design of cooling systems, traditional linear cooling channels often cannot meet the needs of fast and uniform cooling, resulting in quality problems such as product deformation and warping.
[0003] In recent years, computer-aided engineering (CAE) technology has been widely used in injection mold design, but there are still some problems. The accuracy and reliability of CAE simulation results are affected by factors such as model simplification and material parameter accuracy, and it is difficult to fully reflect the complex situation in the actual production process. Secondly, the massive data generated by simulation analysis contains a lot of redundant information. How to effectively extract key features and use them to guide mold optimization design is still a challenge. Summary of the invention
[0004] The present application provides an injection molding method and device based on a plastic mold, which are used to realize intelligent design and optimization of the injection mold cooling system.
[0005] In a first aspect, the present application provides an injection molding method based on a plastic mold, the injection molding method based on a plastic mold comprising:
[0006] Perform 3D modeling on the plastic mold to obtain a 3D model of the mold including the cavity, runner system and cooling channel;
[0007] Performing injection molding process simulation analysis on the three-dimensional model of the mold to obtain temperature distribution data, pressure distribution data, and shear stress distribution data inside the mold;
[0008] Performing adaptive slow feature analysis on the temperature distribution data, the pressure distribution data and the shear stress distribution data to obtain a set of injection molding time-varying features;
[0009] Inputting the injection molding time-varying feature set into a preset sparse autoencoder model to calculate temperature uniformity and cooling efficiency, thereby obtaining temperature uniformity and cooling efficiency indicators;
[0010] According to the temperature uniformity and cooling efficiency indicators, parameter modeling and parameter optimization are performed on the cooling channel to obtain cooling channel optimization parameters.
[0011] In a second aspect, the present application provides an injection molding device based on a plastic mold, the injection molding device based on the plastic mold comprising:
[0012] A modeling module is used to perform three-dimensional modeling of the plastic mold to obtain a three-dimensional model of the mold including the cavity, runner system and cooling channel;
[0013] A simulation module is used to simulate and analyze the injection molding process of the three-dimensional model of the mold to obtain temperature distribution data, pressure distribution data and shear stress distribution data inside the mold;
[0014] An analysis module, used for performing an adaptive slow feature analysis on the temperature distribution data, the pressure distribution data and the shear stress distribution data to obtain a set of injection molding time-varying features;
[0015] A calculation module, used for inputting the injection molding time-varying feature set into a preset sparse autoencoder model to calculate temperature uniformity and cooling efficiency, and obtaining temperature uniformity and cooling efficiency indicators;
[0016] The optimization module is used to perform parameter modeling and parameter optimization on the cooling channel according to the temperature uniformity and cooling efficiency indicators to obtain the cooling channel optimization parameters.
[0017] The third aspect of the present application provides an injection molding device based on a plastic mold, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the injection molding device based on the plastic mold executes the above-mentioned injection molding method based on the plastic mold.
[0018] A fourth aspect of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned plastic mold-based injection molding method.
[0019] In the technical solution provided by the present application, the simulation accuracy of the entire injection molding process is improved by accurately modeling the plastic mold in three dimensions, including the cavity, runner system and cooling channel. The temperature, pressure and shear stress distribution data inside the mold are obtained by simulation analysis of the injection molding process. The adaptive slow feature analysis method can extract key slow-varying features from complex time-varying data and effectively capture the essential dynamic characteristics of the injection molding process. The sparse autoencoder model processes the injection molding time-varying features, realizes the effective compression and reconstruction of high-dimensional features, and retains the key information of the original data. By calculating the temperature uniformity and cooling efficiency indicators, the cooling channel is parameterized and optimized. Combined with the particle swarm algorithm, it can efficiently search for the optimal solution in the complex solution space. By performing frequency domain analysis on the temperature field, an energy concentration index is introduced. The multi-objective evaluation function is used to comprehensively consider the temperature uniformity and cooling efficiency, which can better balance multiple performance indicators in the optimization process, and realize the intelligent design and optimization of the injection mold cooling system. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.
[0021] Figure 1 This is a schematic diagram of an embodiment of an injection molding method based on a plastic mold in an embodiment of the present application;
[0022] Figure 2 It is a schematic diagram of an embodiment of an injection molding device based on a plastic mold in an embodiment of the present application. DETAILED DESCRIPTION
[0023] The embodiment of the present application provides an injection molding method and device based on a plastic mold. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0024] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In the embodiments of the present application, an embodiment of the injection molding method based on a plastic mold includes:
[0025] Step S101, performing three-dimensional modeling on the plastic mold to obtain a three-dimensional model of the mold including a cavity, a runner system and a cooling channel;
[0026] It is understandable that the execution subject of the present application may be an injection molding device based on a plastic mold, or may be a terminal or a server, which is not specifically limited here. The present application embodiment is described by taking a server as the execution subject as an example.
[0027] Specifically, the outer contour of the plastic mold is modeled to obtain the mold shell geometry model. According to the product design drawings, the cavity of the plastic mold is reverse engineered to obtain the cavity geometry model. Reverse engineering modeling can ensure that the cavity geometry model accurately matches the product design, thereby improving the precision of the mold and the quality of the injection molded product. The cavity geometry model is analyzed to obtain the mold parting surface data, and the best position and method of mold parting are determined to ensure that the mold can be opened and closed smoothly and the product can be demolded smoothly. The plastic mold is parted according to the mold parting surface data to obtain the initial geometry model of the movable mold and the fixed mold. The design of the movable mold and the fixed mold directly affects the efficiency of the injection molding process and the quality of the product. The initial geometry model of the movable mold and the fixed mold is Boolean operated to obtain the cavity structure model. Boolean operations make the cavity structure model more accurate and complete by merging, intersecting and subtracting geometric bodies. The runner system is designed according to the cavity structure model to obtain a runner system model including runners, runners and gates. The design of the runner system needs to consider the flow path and filling efficiency of the melt to ensure that the plastic melt can flow smoothly into the mold cavity to form the desired product shape. Perform flow analysis on the runner system model to obtain the melt flow path data. Flow analysis can help the system understand the flow of the melt in the runner system and identify potential problems such as uneven flow, bubbles and poor cooling. Design the initial layout of the cooling channel based on the melt flow path data to obtain the initial model of the cooling channel. The design of the cooling channel needs to consider how to cool the mold evenly in the shortest time to improve production efficiency and product quality. The initial model of the cooling channel is processed to obtain an adjustable cooling channel parameter set. The optimization of the cooling channel parameters can ensure that the cooling system can provide the best cooling effect under different production conditions, thereby stably controlling product quality. The cavity structure model, runner system model and cooling channel initial model are integrated to obtain a three-dimensional mold model including the cavity, runner system and cooling channel.
[0028] Step S102, performing injection molding process simulation analysis on the three-dimensional model of the mold to obtain temperature distribution data, pressure distribution data and shear stress distribution data inside the mold;
[0029] Specifically, the three-dimensional model of the mold is meshed by finite elements to obtain a mold mesh model. By dividing the mold into multiple small units, the behavior and changes of the material during the injection molding process can be simulated more accurately. The material properties of the mold mesh model are assigned according to the physical properties of the injection molding material to obtain a simulation model with material characteristics. The injection molding process parameters and boundary conditions are set for the simulation model to obtain an initial simulation condition set. The injection molding process parameters and boundary conditions include injection molding temperature, pressure, flow rate, etc. These parameters directly affect the simulation accuracy of the injection molding process. According to the initial simulation condition set, the simulation model is analyzed in the filling stage to obtain the melt flow front line data. The filling stage analysis simulates the flow process of the melt in the mold cavity. By analyzing the melt flow front line data, the flow conditions and potential problems in the filling process can be understood. The melt flow front line data is analyzed in time series to obtain a filling time distribution diagram. The filling time distribution diagram shows the flow of the melt at different time points. According to the filling time distribution diagram, the simulation model is analyzed in the holding stage to obtain the initial pressure field data inside the mold. The pressure holding phase analysis simulates the maintenance and change of pressure during the injection molding process. The initial pressure field data reflects the pressure distribution in different areas of the mold. The initial pressure field data is gradient calculated to obtain pressure distribution data. By calculating the pressure gradient, the stress concentration area in the mold is identified. According to the pressure distribution data, the simulation model is analyzed in the cooling phase to obtain the numerical solution of the heat conduction equation. The cooling phase analysis simulates the process of the melt cooling and solidification in the mold. The numerical solution of the heat conduction equation reflects the temperature change inside the mold. The numerical solution of the heat conduction equation is interpolated. Interpolation is a method in numerical analysis. By interpolating the numerical solution, smoother and more detailed temperature distribution data can be obtained. According to the temperature distribution data and pressure distribution data, the simulation model is rheologically analyzed to obtain the shear stress distribution data. The shear stress distribution data can reflect the flow resistance and stress concentration of the material in the mold.
[0030] Step S103, performing adaptive slow feature analysis on the temperature distribution data, the pressure distribution data and the shear stress distribution data to obtain a set of injection molding time-varying features;
[0031] Specifically, temperature distribution data, pressure distribution data and shear stress distribution data are adaptively sampled to obtain multi-scale time series. Adaptive sampling can dynamically adjust the sampling frequency according to the characteristics of the data to ensure that the main change characteristics of the data are captured at different time scales. A sliding time window is constructed according to the multi-scale time series, and a data subset sequence is generated based on the sliding time window. The sliding time window helps to analyze the change trend of data in different time periods. The data subset sequence is recursively calculated to obtain the time-varying covariance matrix estimate. The time-varying covariance matrix reflects the correlation and change of the data at different times. The generalized eigenvalue decomposition is performed on it to obtain adaptive eigenvectors and eigenvalue pairs. The eigenvectors and eigenvalue pairs reveal the main change direction and intensity of the data in different dimensions. The adaptive eigenvector is nonlinearly activated to obtain a nonlinear slow feature map. The nonlinear activation function can enhance the expression ability of the eigenvector so that it can capture the nonlinear characteristics of the data. A time consistency constraint is constructed according to the nonlinear slow feature map to obtain a slow feature optimization objective function. The time consistency constraint ensures the smoothness and continuity of the feature in time, and the optimization objective function is used to guide the optimization process of feature extraction. The slow feature optimization objective function is subjected to random gradient descent to obtain the optimal slow feature extractor. Stochastic gradient descent is an efficient optimization algorithm that gradually approaches the optimal solution through continuous iterative updates. The optimal slow feature extractor is used to extract features from the temperature distribution data, pressure distribution data, and shear stress distribution data to obtain temperature slow feature sequences, pressure slow feature sequences, and shear stress slow feature sequences. The temperature slow feature sequences, pressure slow feature sequences, and shear stress slow feature sequences are processed by the adaptive forgetting factor mechanism to obtain temperature time-varying features, pressure time-varying features, and shear stress time-varying features. The adaptive forgetting factor mechanism can dynamically adjust the value of the forgetting factor according to the changes in the data, so that the model can adapt to new data more quickly and improve the real-time and accuracy of feature extraction. The temperature time-varying features, pressure time-varying features, and shear stress time-varying features are fused to obtain the injection molding time-varying feature set. The features of different data sources are effectively integrated together, and the information of various features is fully utilized to improve the overall feature expression ability and prediction accuracy.
[0032] Step S104, inputting the injection molding time-varying feature set into a preset sparse autoencoder model to calculate temperature uniformity and cooling efficiency, and obtaining temperature uniformity and cooling efficiency indicators;
[0033] Specifically, the injection molding time-varying feature set is input into a preset sparse autoencoder model, and the injection molding time-varying feature set is normalized to obtain a normalized input vector, ensuring that the input data is on the same scale, thereby avoiding the instability of the model training caused by the difference in the numerical range. The normalized input vector is multiplied by the first layer weight matrix and the first bias is added to obtain the linear output of the first hidden layer. The normalized input vector is mapped to the high-dimensional space of the hidden layer through the weight matrix. The linear output of the first hidden layer is operated by the ReLU activation function to obtain the first nonlinear feature representation of the first hidden layer. The ReLU activation function can introduce nonlinear features, increase the expression ability of the model, and effectively alleviate the gradient disappearance problem. The linear output of the second hidden layer is obtained by multiplying the first nonlinear feature representation with the second layer weight matrix and adding the second bias. The nonlinear features of the first layer are further mapped to a higher-dimensional space. The Sigmoid activation function is operated on the linear output of the second hidden layer to obtain the second nonlinear feature representation of the second hidden layer. The Sigmoid activation function maps the output to the (0, 1) interval, making the feature representation smoother and more adaptable to subsequent processing. The KL divergence is calculated based on the second nonlinear feature representation to obtain the sparsity penalty term. The KL divergence is used to measure the difference between two probability distributions. By introducing the sparsity penalty term, the autoencoder can be forced to use as few activated neurons as possible in feature representation, thereby improving the sparsity and generalization ability of the model. The second nonlinear feature representation is multiplied by the decoder weight matrix and the third bias is added to obtain the reconstructed output vector. The feature representation of the hidden layer is remapped back to the input space to achieve data reconstruction. According to the reconstructed output vector, the temperature distribution statistics and cooling rate are calculated to obtain the temperature uniformity and cooling efficiency indicators. The temperature distribution statistics reflect the temperature difference between different areas in the mold. Temperature uniformity is an important indicator to measure the temperature control effect during the injection molding process. The cooling rate directly affects the production efficiency and product quality. The cooling efficiency index can help optimize the cooling channel design and cooling strategy.
[0034] The reconstructed output vector is feature parsed to obtain the temperature field matrix and the corresponding time vector. The temperature field matrix contains the temperature distribution information at different time points, while the time vector records the specific moments of these time points. The mean and standard deviation of the temperature at each moment are calculated according to the temperature field matrix to obtain the temperature statistical sequence. The temperature statistical sequence can reflect the overall distribution of temperature at different time points. The temperature statistical sequence is time-weighted averaged to obtain the comprehensive temperature uniformity index. This index can comprehensively reflect the uniformity of temperature during the injection molding process and is an important parameter for evaluating injection molding quality. The temperature gradient of each spatial point is calculated according to the temperature field matrix and the time vector to obtain the cooling rate field. The temperature gradient reflects the change of temperature in space, and the cooling rate field shows the cooling speed of different areas of the mold. The cooling rate field is spatially averaged to obtain the average cooling rate curve. The average cooling rate curve can show the change trend of the overall cooling process and is an important basis for evaluating the cooling efficiency. The characteristic time of the cooling process is calculated according to the average cooling rate curve to obtain the cooling efficiency index. The characteristic time refers to the time required for the cooling process to reach a certain level. This indicator can be used to evaluate the efficiency of the cooling system. The temperature field matrix is Fourier transformed to obtain the temperature field spectrum distribution. Fourier transform can convert time domain data into frequency domain, thereby revealing the periodicity and frequency characteristics of the temperature field. The energy concentration is calculated according to the temperature field spectrum distribution to obtain the temperature field frequency domain index. The energy concentration reflects the energy distribution of the main frequency components in the temperature field and is an important parameter for evaluating the stability of the temperature field. The temperature uniformity index, cooling efficiency index and temperature field frequency domain index are normalized to obtain a standardized index set. The dimensional differences between different indicators are eliminated so that different indicators can be compared and comprehensively evaluated on the same scale. A multi-objective evaluation function is constructed based on the standardized index set, and the temperature uniformity and cooling efficiency indicators are calculated based on the multi-objective evaluation function.
[0035] Step S105: perform parameter modeling and parameter optimization on the cooling channel according to the temperature uniformity and cooling efficiency indicators to obtain the cooling channel optimization parameters.
[0036] Specifically, the cooling channel is parametrically modeled to obtain a parameter vector describing the shape of the cooling channel. The parameter vector includes key parameters such as the geometry, size, and position of the cooling channel, and can fully describe the structural characteristics of the cooling channel. The objective function is constructed according to the parameter vector, and the objective function is used to comprehensively consider the optimization objectives of temperature uniformity and cooling efficiency. The construction of the objective function needs to be combined with actual production requirements, and the temperature uniformity and cooling efficiency should be weighed to achieve the best balance between the two. Geometric constraints and process constraints are imposed on the objective function to obtain a constrained optimization problem expression. The geometric constraints include the spatial position and shape restrictions of the cooling channel in the mold, and the process constraints include the manufacturing process and material restrictions of the cooling channel. By imposing these constraints, it is ensured that the optimization results are operational and economical in actual production. An initialized particle swarm is created according to the constrained optimization problem expression, and the fitness of each particle in the initialized particle swarm is evaluated to obtain an initial fitness value set. Fitness evaluation is the core of the particle swarm optimization algorithm. By evaluating the fitness value of each particle, its performance in the objective function is determined. The individual optimal position and the global optimal position of the initialized particle swarm are updated according to the initial fitness value set to obtain an updated particle swarm. The individual optimal position refers to the best position found so far by each particle itself, and the global optimal position refers to the best position found so far by all particles in the entire particle swarm. By updating these positions, the particle swarm is guided to converge in the direction of the optimal solution. The updated particle swarm is updated with speed and position to obtain a new generation of particle swarm. Speed update is to adjust the movement speed of the particle according to the current speed, individual optimal position and global optimal position of the particle, while position update is to adjust the position of the particle according to the updated speed. The optimization iteration process is judged according to the preset termination condition to obtain a Boolean value used to determine whether to continue iteration. The termination condition can be the number of iterations, the degree of convergence of the objective function value or the limitation of computing resources. If the Boolean value is false, it means that the termination condition has not been met and iterative optimization needs to continue. Otherwise, the optimal particle is obtained, and the cooling channel optimization parameters of the cooling channel are obtained according to the optimal particle decoding. The decoding process is to convert the parameter vector of the optimal particle into a specific cooling channel design parameter, thereby guiding the actual mold cooling system design.
[0037] In the embodiment of the present application, the simulation accuracy of the entire injection molding process is improved by accurately modeling the plastic mold in three dimensions, including the cavity, runner system and cooling channel. The temperature, pressure and shear stress distribution data inside the mold are obtained by simulation analysis of the injection molding process. The adaptive slow feature analysis method can extract key slow-varying features from complex time-varying data, effectively capture the essential dynamic characteristics of the injection molding process, and the sparse autoencoder model processes the injection molding time-varying features, realizing the effective compression and reconstruction of high-dimensional features, while retaining the key information of the original data. By calculating the temperature uniformity and cooling efficiency indicators, the cooling channel is parameterized and optimized. Combined with the particle swarm algorithm, the optimal solution can be efficiently searched in the complex solution space. By performing frequency domain analysis on the temperature field, an energy concentration index is introduced. The multi-objective evaluation function is used to comprehensively consider the temperature uniformity and cooling efficiency, which can better balance multiple performance indicators in the optimization process, and realize the intelligent design and optimization of the injection mold cooling system.
[0038] In a specific embodiment, the process of executing step S101 may specifically include the following steps:
[0039] (1) Model the outer contour of the plastic mold to obtain the mold shell geometry model, and reverse engineer the plastic mold cavity according to the product design drawing to obtain the cavity geometry model;
[0040] (2) Analyze the parting surface of the cavity geometry model to obtain the mold parting surface data, and perform parting design on the plastic mold based on the mold parting surface data to obtain the initial geometry models of the movable mold and the fixed mold;
[0041] (3) Performing Boolean operations on the initial geometric models of the movable mold and the fixed mold to obtain the cavity structure model, and designing the runner system based on the cavity structure model to obtain the runner system model including the runner, runner, and gate;
[0042] (4) Performing flow analysis on the runner system model to obtain melt flow path data, and designing the initial layout of the cooling channel based on the melt flow path data to obtain an initial model of the cooling channel;
[0043] (5) Processing the initial model of the cooling channel to obtain an adjustable cooling channel parameter set;
[0044] (6) The cavity structure model, runner system model and cooling channel initial model are integrated to obtain a three-dimensional mold model including the cavity, runner system and cooling channel.
[0045] Specifically, the outer contour of the plastic mold is modeled to obtain the geometric model of the mold shell. Computer-aided design (CAD) software is used to create a three-dimensional model of the mold shell by inputting the size and shape information of the outer mold. According to the product design drawings, the cavity of the plastic mold is reverse-engineered to obtain the cavity geometry model. Reverse engineering modeling is to ensure that the mold cavity fully matches the product design by reversely mapping the geometric shape and features of the product to the mold. The cavity geometry model is analyzed to obtain the mold parting surface data, determine the optimal parting position of the mold, ensure that the mold can be opened and closed smoothly during the injection molding process, and the product can be demolded smoothly. Based on the parting surface data, the plastic mold is parted to obtain the initial geometry model of the movable mold and the fixed mold. Boolean operations are performed on the initial geometry models of the movable mold and the fixed mold to obtain the cavity structure model. Boolean operations include the merging, intersection and difference operations of geometric bodies. Through these operations, a complete cavity structure model is constructed. The runner system is designed according to the cavity structure model to obtain a runner system model including runners, runners and gates. The design of the runner system needs to consider the flow path and filling efficiency of the melt to ensure that the plastic melt can flow smoothly into the mold cavity to form the desired product shape. Through flow analysis, the melt flow path data can be obtained. Flow analysis usually uses computational fluid dynamics software to simulate the flow of the melt in the runner system and identify potential problems, such as uneven flow, bubbles, and poor cooling. According to the flow path data, the initial layout of the cooling channel is designed to obtain the initial model of the cooling channel. In order to optimize the cooling effect, the initial model of the cooling channel is processed to obtain an adjustable cooling channel parameter set. The cooling channel parameters include the diameter, length, position and flow rate of the cooling medium of the channel. By optimizing these parameters, the cooling efficiency can be significantly improved to ensure that the mold is cooled evenly in the shortest time. For example, assuming that in the initial layout of the cooling channel, the diameter of the cooling channel is , the length is , the flow rate of the cooling medium is , the cooling efficiency can be evaluated by the following formula:
[0046] ;
[0047] in, For cooling efficiency, is the temperature difference of the cooling medium, is the cross-sectional area of the cooling channel, is the specific heat capacity of the cooling medium. The cavity structure model, runner system model and cooling channel initial model are integrated to obtain a three-dimensional mold model including the cavity, runner system and cooling channel.
[0048] In a specific embodiment, the process of executing step S102 may specifically include the following steps:
[0049] (1) Perform finite element meshing on the mold 3D model to obtain a mold mesh model, and assign material properties to the mold mesh model based on the physical properties of the injection molding material to obtain a simulation model with material properties;
[0050] (2) Setting the injection molding process parameters and boundary conditions for the simulation model to obtain the initial simulation condition set, and performing filling phase analysis on the simulation model based on the initial simulation condition set to obtain melt flow front line data;
[0051] (3) Perform time series analysis on the melt flow front data to obtain a filling time distribution diagram, and perform pressure holding phase analysis on the simulation model based on the filling time distribution diagram to obtain the initial pressure field data inside the mold;
[0052] (4) Perform gradient calculation on the initial pressure field data to obtain pressure distribution data, and analyze the cooling stage of the simulation model based on the pressure distribution data to obtain the numerical solution of the heat conduction equation;
[0053] (5) The numerical solution of the heat conduction equation is interpolated to obtain temperature distribution data. The simulation model is subjected to rheological analysis based on the temperature distribution data and pressure distribution data to obtain shear stress distribution data.
[0054] Specifically, the finite element analysis software is used to mesh the mold 3D model. The model is divided into many small units, which can be tetrahedrons, hexahedrons or other geometric shapes. The fineness of the meshing will affect the accuracy and amount of calculation of the simulation. A finer mesh can provide higher simulation accuracy, but it also increases the calculation cost. After the meshing is completed, the material properties of the mold mesh model are assigned according to the physical properties of the injection molding material used. These physical properties include density , thermal conductivity Specific heat capacity Etc. Through the assignment process, a simulation model with material properties is obtained to ensure the accuracy and reliability of the simulation results. The injection molding process parameters and boundary conditions are set for the simulation model to obtain the initial simulation condition set. The injection molding process parameters include injection temperature , injection pressure , injection speed etc., and the boundary conditions include the fixed constraints of the mold, the temperature conditions of the cooling channel, etc. These parameters and conditions are set to simulate the real injection molding process and obtain accurate simulation results. According to the initial simulation condition set, the simulation model is analyzed in the filling stage to obtain the melt flow front line data. The filling stage analysis refers to the simulation of the flow process of the plastic melt in the mold cavity. The flow front position of the melt at different time points is obtained by calculation. These data can reflect the filling situation of the melt in the cavity and the potential defect location. The initial pressure field data is gradient calculated to obtain the pressure distribution data. The pressure distribution data can help understand the stress distribution inside the mold. According to the pressure distribution data, the simulation model is analyzed in the cooling stage to obtain the numerical solution of the heat conduction equation. The cooling stage analysis refers to the simulation of the cooling and solidification process of the melt in the mold. By calculating the numerical solution of the heat conduction equation, the temperature distribution inside the mold at different time points can be obtained. The basic form of the heat conduction equation is:
[0055] ;
[0056] in, is the temperature, For time, is the thermal diffusivity. The numerical solution of the heat conduction equation is interpolated to obtain the temperature distribution data. The discrete numerical solution is converted into continuous temperature distribution data for easy analysis and application. According to the temperature distribution data and pressure distribution data, the simulation model is subjected to rheological analysis to obtain the shear stress distribution data. Rheological analysis refers to the study of the stress and strain of the material during the flow and deformation process. By calculating the shear stress distribution data, the flow resistance and stress concentration of the material during the injection molding process can be understood. Shear stress The basic calculation formula is:
[0057] ;
[0058] in, is the shear stress, is the dynamic viscosity, is the velocity gradient.
[0059] In a specific embodiment, the process of executing step S103 may specifically include the following steps:
[0060] (1) Adaptively sample temperature distribution data, pressure distribution data, and shear stress distribution data to obtain multi-scale time series;
[0061] (2) Construct a sliding time window based on the multi-scale time series and generate a data subset sequence based on the sliding time window;
[0062] (3) Recursively calculate the data subset sequence to obtain the time-varying covariance matrix estimate, and perform generalized eigenvalue decomposition on the time-varying covariance matrix estimate to obtain adaptive eigenvectors and eigenvalue pairs;
[0063] (4) Perform nonlinear activation on the adaptive feature vector to obtain a nonlinear slow feature map, and construct a time consistency constraint based on the nonlinear slow feature map to obtain a slow feature optimization objective function;
[0064] (5) Performing random gradient descent on the slow feature optimization objective function to obtain an optimal slow feature extractor, and performing feature extraction on the temperature distribution data, pressure distribution data, and shear stress distribution data based on the optimal slow feature extractor to obtain a temperature slow feature sequence, a pressure slow feature sequence, and a shear stress slow feature sequence;
[0065] (6) The temperature slow feature sequence, pressure slow feature sequence and shear stress slow feature sequence are processed by an adaptive forgetting factor mechanism to obtain the temperature time-varying feature, pressure time-varying feature and shear stress time-varying feature;
[0066] (7) The temperature time-varying features, pressure time-varying features, and shear stress time-varying features are fused to obtain the injection molding time-varying feature set.
[0067] Specifically, the temperature distribution data, pressure distribution data, and shear stress distribution data are adaptively sampled. According to the characteristics of data changes, the sampling frequency is dynamically adjusted to capture the main change characteristics of the data at different time scales, and a time series reflecting the multi-scale characteristics of the data is obtained. A sliding time window is constructed based on the multi-scale time series, and a data subset sequence is generated based on the sliding time window. The sliding time window is a commonly used time series analysis method. By setting a fixed-length window on the time series and sliding the window on the sequence, a series of continuous data subsets are generated. Set the sliding window length to , then in the time series The generated data subset sequence is . Recursively calculate these data subset sequences to obtain the time-varying covariance matrix estimate. The time-varying covariance matrix reflects the correlation and changes of the data at different times. Covariance matrix It can be calculated by the following formula:
[0068] ;
[0069] in, is the mean vector of the data in the window, It is The data vector at each time point. Perform generalized eigenvalue decomposition on the time-varying covariance matrix to obtain adaptive eigenvectors and eigenvalue pairs. Decomposing the covariance matrix into eigenvalues and eigenvectors reveals the main direction and intensity of changes in the data in different dimensions. Assume that the eigenvalues of the covariance matrix are , the eigenvector is , the generalized eigenvalue decomposition can be performed by the following formula:
[0070] ;
[0071] The adaptive feature vector is nonlinearly activated to obtain a nonlinear slow feature map. The nonlinear activation function can enhance the expressive power of the feature vector so that it can capture the nonlinear characteristics of the data. Commonly used nonlinear activation functions include ReLU and Sigmoid functions. Assume that the activation function is , then the nonlinear activation can be expressed as:
[0072] ;
[0073] Based on the nonlinear slow feature mapping, the time consistency constraint is constructed to obtain the slow feature optimization objective function. The time consistency constraint ensures the smoothness and continuity of the feature in time. The optimization objective function is used to guide the optimization process of feature extraction. The optimization objective function can be expressed as:
[0074] ;
[0075] in, is the optimization objective function, It is The feature representation of each time point is is the regularization parameter. The slow feature optimization objective function is subjected to stochastic gradient descent to obtain the optimal slow feature extractor. Stochastic gradient descent is an efficient optimization algorithm that can gradually approach the optimal solution through continuous iterative updates. The optimal slow feature extractor is used to extract features from the temperature distribution data, pressure distribution data, and shear stress distribution data to obtain temperature slow feature sequences, pressure slow feature sequences, and shear stress slow feature sequences. The feature sequences are processed by the adaptive forgetting factor mechanism to obtain temperature time-varying features, pressure time-varying features, and shear stress time-varying features. The adaptive forgetting factor mechanism can dynamically adjust the value of the forgetting factor according to the changes in the data, so that the model can adapt to new data more quickly and improve the real-time and accuracy of feature extraction. Assume that the forgetting factor is , it can be updated by the following formula:
[0076]
[0077] The temperature time-varying features, pressure time-varying features and shear stress time-varying features are fused to obtain a set of injection molding time-varying features. Feature fusion can effectively integrate features from different data sources, make full use of information from various features, and improve the overall feature expression ability and prediction accuracy. In a specific embodiment, the process of executing step S104 can specifically include the following steps:
[0078] (1) Inputting the injection molding time-varying feature set into a preset sparse autoencoder model, and normalizing the injection molding time-varying feature set to obtain a normalized input vector;
[0079] (2) The linear output of the first hidden layer is obtained by multiplying the normalized input vector with the first layer weight matrix and adding the first bias;
[0080] (3) Performing a ReLU activation function operation on the linear output of the first hidden layer to obtain the first nonlinear feature representation of the first hidden layer;
[0081] (4) According to the first nonlinear feature representation, multiplying it with the second layer weight matrix and adding the second bias, the linear output of the second hidden layer is obtained;
[0082] (5) Performing a Sigmoid activation function operation on the linear output of the second hidden layer to obtain a second nonlinear feature representation of the second hidden layer;
[0083] (6) Calculate the KL divergence based on the second nonlinear feature representation to obtain the sparsity penalty term, multiply the second nonlinear feature representation with the decoder weight matrix and add the third bias to obtain the reconstructed output vector;
[0084] (7) Based on the reconstructed output vector, the temperature distribution statistics and cooling rate are calculated to obtain the temperature uniformity and cooling efficiency indicators.
[0085] Specifically, the injection molding time-varying feature set is normalized and the data is scaled to a standard range, usually by scaling each feature to the interval [0, 1] or [-1, 1], thereby eliminating the influence of different feature dimensions. With the first layer weight matrix Multiply and add the first bias , and get the linear output of the first hidden layer:
[0086] ;
[0087] in, is the linear output of the first hidden layer, is the weight matrix of the first layer, is the bias vector of the first layer. The ReLU activation function is performed on the linear output of the first hidden layer to obtain the first nonlinear feature representation of the first hidden layer. The ReLU activation function is defined as:
[0088] ;
[0089] in, is the nonlinear feature representation of the first hidden layer. According to the first nonlinear feature representation With the second layer weight matrix Multiply and add the second bias , and get the linear output of the second hidden layer:
[0090] ;
[0091] The Sigmoid activation function is operated on the linear output of the second hidden layer to obtain the second nonlinear feature representation of the second hidden layer. The Sigmoid activation function is defined as:
[0092] ;
[0093] in, is the nonlinear feature representation of the second hidden layer. The KL divergence is calculated based on the second nonlinear feature representation to obtain the sparsity penalty term. The KL divergence is used to measure the difference between two probability distributions. Assume that the sparsity target is , then the KL divergence is defined as:
[0094] ;
[0095] The sparsity penalty term is the sum of the KL divergences of all hidden units. and the decoder weight matrix Multiply and add the third bias , and get the reconstructed output vector:
[0096] ;
[0097] in, is the reconstructed output vector, is the weight matrix of the decoder, is the bias vector of the decoder. According to the reconstructed output vector , calculate the temperature distribution statistics and cooling rate to obtain the temperature uniformity and cooling efficiency indicators. The temperature distribution statistics can be obtained by calculating the mean and variance of the reconstructed temperature field. The cooling rate can be obtained by calculating the rate of change of temperature over time, assuming that the time step is , then the cooling rate Defined as:
[0098] ;
[0099] in, is the temperature variation. Through these calculation results, the temperature uniformity and cooling efficiency indicators are obtained. Through the sparse autoencoder model, the main features in the injection molding time-varying feature set are extracted, and the temperature distribution and cooling rate are calculated to evaluate the temperature uniformity and cooling efficiency during the injection molding process.
[0100] In a specific embodiment, the process of calculating the temperature distribution statistics and the cooling rate according to the reconstructed output vector to obtain the temperature uniformity and cooling efficiency index may specifically include the following steps:
[0101] (1) Perform feature analysis on the reconstructed output vector to obtain the temperature field matrix and the corresponding time vector;
[0102] (2) Calculate the temperature mean and standard deviation at each moment according to the temperature field matrix to obtain the temperature statistical series, and perform time-weighted averaging on the temperature statistical series to obtain the comprehensive temperature uniformity index;
[0103] (3) Calculate the temperature gradient of each spatial point according to the temperature field matrix and the time vector to obtain the cooling rate field, and perform spatial averaging on the cooling rate field to obtain the average cooling rate curve;
[0104] (4) Calculate the characteristic time of the cooling process based on the average cooling rate curve to obtain the cooling efficiency index;
[0105] (5) Perform Fourier transform on the temperature field matrix to obtain the temperature field spectrum distribution, and calculate the energy concentration based on the temperature field spectrum distribution to obtain the temperature field frequency domain index;
[0106] (6) Normalizing the temperature uniformity index, cooling efficiency index and temperature field frequency domain index to obtain a standardized index set;
[0107] (7) A multi-objective evaluation function is constructed based on the standardized indicator set, and the temperature uniformity and cooling efficiency indicators are calculated based on the multi-objective evaluation function.
[0108] Specifically, the reconstructed output vector is converted into a temperature field matrix and a corresponding time vector. Assume that the reconstructed output vector Contains temperature data at multiple times, the temperature field matrix It can be expressed as , where each is the temperature distribution data at a certain moment. The time vector The time point corresponding to each moment is recorded. According to the temperature field matrix Calculate the mean and standard deviation of temperature at each moment to obtain the temperature statistical series. The calculation formula is as follows:
[0109] ;
[0110] in, The temperature field matrix is represented in moment, The temperature value of a spatial point, Represents the total number of spatial points. Perform time-weighted averaging on the temperature statistical series to obtain the comprehensive temperature uniformity index. The formula for time-weighted averaging is:
[0111] ;
[0112] According to the temperature field matrix and the time vector , calculate the temperature gradient of each spatial point and obtain the cooling rate field. The temperature gradient calculation formula is:
[0113] ;
[0114] in, Indicates moment, The temperature gradient at a spatial point, and Respectively represent The time point of the moment and the previous moment. The cooling rate field is spatially averaged to obtain the average cooling rate curve. The formula for spatial average is:
[0115] ;
[0116] in, Indicates The average cooling rate at each moment. According to the average cooling rate curve, the characteristic time of the cooling process is calculated to obtain the cooling efficiency index. Assume that the time point when the cooling process reaches a steady state is , then the cooling efficiency index can be expressed as the time required for cooling to reach steady state:
[0117] ;
[0118] Perform Fourier transform on the temperature field matrix to obtain the temperature field spectrum distribution. The formula for Fourier transform is:
[0119] ;
[0120] Among them, F represents the frequency spectrum distribution of the temperature field, represents frequency. The energy concentration is calculated according to the temperature field spectrum distribution, and the temperature field frequency domain index is obtained. The energy concentration can be measured by the concentration of spectrum energy. Assuming that the spectrum energy is , then the calculation formula for energy concentration is:
[0121] ;
[0122] The temperature uniformity index, cooling efficiency index and temperature field frequency domain index are normalized to obtain a standardized index set. A multi-objective evaluation function is constructed based on the standardized index set, and the temperature uniformity and cooling efficiency indexes are calculated based on the multi-objective evaluation function. The multi-objective evaluation function can comprehensively consider the weights and influences of multiple indicators.
[0123] In a specific embodiment, the process of executing step S105 may specifically include the following steps:
[0124] (1) Parametric modeling of the cooling channel is performed to obtain a parameter vector describing the shape of the cooling channel;
[0125] (2) Constructing an objective function based on the parameter vector. The objective function is used to comprehensively consider the optimization goals of temperature uniformity and cooling efficiency.
[0126] (3) Apply geometric constraints and process constraints to the objective function to obtain a constrained optimization problem expression;
[0127] (4) Create an initialized particle swarm based on the constrained optimization problem expression, and evaluate the fitness of each particle in the initialized particle swarm to obtain an initial fitness value set;
[0128] (5) Update the individual optimal position and global optimal position of the initialized particle swarm according to the initial fitness value set to obtain the updated particle swarm;
[0129] (6) Update the velocity and position of the updated particle swarm to obtain a new generation of particle swarm;
[0130] (7) The optimization iteration process is judged according to the preset termination condition to obtain a Boolean value used to determine whether to continue the iteration;
[0131] (8) Conditional judgment is performed on the Boolean value. If the Boolean value is false, the optimal particle is obtained, and the cooling channel optimization parameters of the cooling channel are obtained according to the decoding of the optimal particle.
[0132] Specifically, the shape of the cooling channel is described to obtain a parameter vector. Assume that the shape of the cooling channel can be described by a set of geometric parameters, including the diameter of the channel ,length ,Location And the bending angle Etc. Through these parameters, construct a vector To describe the shape of the cooling channel. According to the parameter vector Construct the objective function. The objective function needs to comprehensively consider temperature uniformity and cooling efficiency to optimize the design of the cooling channel. Set the objective function as ,in Including temperature uniformity items and cooling efficiency term , the objective function formula can be expressed as:
[0133] ;
[0134] in, and are weight coefficients, indicating the importance of temperature uniformity and cooling efficiency respectively. Temperature uniformity It reflects whether the temperature distribution of the cooling channel in the mold is uniform and the cooling efficiency It reflects the efficiency of the cooling system. Geometric constraints and process constraints are imposed on the objective function to obtain the constrained optimization problem expression. Geometric constraints include the spatial position and shape restrictions of the cooling channel in the mold, and process constraints include the manufacturing process and material restrictions of the cooling channel. For example, the diameter of the cooling channel Cannot be less than a minimum value ,Location The boundary of the mold cannot be exceeded, etc. The constrained optimization problem can be expressed as:
[0135] minimize subject to , ;
[0136] in, represents an inequality constraint, Represents an equality constraint. According to the constrained optimization problem expression, an initial particle swarm is created, and the fitness of each particle in the initial particle swarm is evaluated to obtain an initial fitness value set. The particle swarm optimization algorithm finds the optimal solution to the optimization problem by simulating group behavior. Each particle represents a potential solution, and its position is determined by the parameter vector Represents. Fitness evaluation is based on the objective function Calculate the fitness value of each particle to measure its quality. Update the individual optimal position and global optimal position of the initialized particle swarm based on the initial fitness value set. Individual optimal position It refers to the optimal position found by each particle itself, the global optimal position, and the global optimal position. It refers to the optimal position found in the entire particle swarm. The updated particle swarm updates its speed and position through the following formula:
[0137] ;
[0138] ;
[0139] in, It is a particle In the The speed at the iteration, It is a particle In the The position at the iteration, is the inertia weight and is the learning factor, and is a random number. The optimization iteration process is judged according to the preset termination condition to obtain a Boolean value used to determine whether to continue iteration. The termination condition can be that the number of iterations reaches the maximum value, or the global optimal position does not change significantly in multiple iterations. If the termination condition is false, the iteration continues; if it is true, the iteration stops and the optimal particle is obtained. The position of the optimal particle Represents the optimization parameters of the cooling channel.
[0140] The above describes the injection molding method based on the plastic mold in the embodiment of the present application. The following describes the injection molding device based on the plastic mold in the embodiment of the present application. Figure 2 In the embodiments of the present application, an injection molding device based on a plastic mold includes:
[0141] A modeling module 201 is used to perform three-dimensional modeling on the plastic mold to obtain a three-dimensional model of the mold including a cavity, a runner system and a cooling channel;
[0142] The simulation module 202 is used to simulate and analyze the injection molding process of the three-dimensional model of the mold to obtain temperature distribution data, pressure distribution data and shear stress distribution data inside the mold;
[0143] An analysis module 203 is used to perform an adaptive slow feature analysis on the temperature distribution data, the pressure distribution data and the shear stress distribution data to obtain a set of injection molding time-varying features;
[0144] The calculation module 204 is used to input the injection molding time-varying feature set into a preset sparse autoencoder model to calculate the temperature uniformity and cooling efficiency, and obtain the temperature uniformity and cooling efficiency indicators;
[0145] The optimization module 205 is used to perform parameter modeling and parameter optimization on the cooling channel according to the temperature uniformity and cooling efficiency indicators to obtain the optimized parameters of the cooling channel.
[0146] Through the cooperation of the above components, the simulation accuracy of the entire injection molding process is improved by accurately modeling the plastic mold in three dimensions, including the cavity, runner system and cooling channel. The temperature, pressure and shear stress distribution data inside the mold are obtained by simulation analysis of the injection molding process. The adaptive slow feature analysis method can extract key slow-varying features from complex time-varying data and effectively capture the essential dynamic characteristics of the injection molding process. The sparse autoencoder model processes the injection molding time-varying features, realizes the effective compression and reconstruction of high-dimensional features, and retains the key information of the original data. By calculating the temperature uniformity and cooling efficiency indicators, the cooling channel is parameterized and optimized. Combined with the particle swarm algorithm, it can efficiently search for the optimal solution in the complex solution space. By performing frequency domain analysis on the temperature field, the energy concentration index is introduced. The multi-objective evaluation function is used to comprehensively consider the temperature uniformity and cooling efficiency, which can better balance multiple performance indicators in the optimization process, and realize the intelligent design and optimization of the injection mold cooling system.
[0147] The present application also provides an injection molding device based on a plastic mold, wherein the injection molding device based on a plastic mold includes a memory and a processor, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the processor executes the steps of the injection molding method based on the plastic mold in the above-mentioned embodiments.
[0148] The present application also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein instructions are stored in the computer, and when the instructions are executed on a computer, the computer executes the steps of the plastic mold-based injection molding method.
[0149] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.
[0150] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.
[0151] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.
Claims
1. An injection molding method based on a plastic mold, characterized in that: The method comprises: Perform 3D modeling on the plastic mold to obtain a 3D model of the mold including the cavity, runner system and cooling channel; Performing injection molding process simulation analysis on the three-dimensional model of the mold to obtain temperature distribution data, pressure distribution data, and shear stress distribution data inside the mold; The temperature distribution data, the pressure distribution data and the shear stress distribution data are subjected to an adaptive slow feature analysis to obtain a time-varying feature set for injection molding; specifically, the method comprises: adaptively sampling the temperature distribution data, the pressure distribution data and the shear stress distribution data to obtain a multi-scale time series; constructing a sliding time window according to the multi-scale time series, and generating a data subset sequence based on the sliding time window; recursively calculating the data subset sequence to obtain a time-varying covariance matrix estimate, and performing generalized eigenvalue decomposition on the time-varying covariance matrix estimate to obtain an adaptive eigenvector and an eigenvalue pair; performing nonlinear activation on the adaptive eigenvector to obtain a nonlinear slow feature map, and constructing a time-varying covariance matrix estimate according to the nonlinear slow feature map. time consistency constraint, obtain a slow feature optimization objective function; perform random gradient descent on the slow feature optimization objective function to obtain an optimal slow feature extractor, and perform feature extraction on the temperature distribution data, the pressure distribution data and the shear stress distribution data according to the optimal slow feature extractor to obtain a temperature slow feature sequence, a pressure slow feature sequence and a shear stress slow feature sequence; perform adaptive forgetting factor mechanism processing on the temperature slow feature sequence, the pressure slow feature sequence and the shear stress slow feature sequence to obtain temperature time-varying features, pressure time-varying features and shear stress time-varying features; perform feature fusion on the temperature time-varying features, the pressure time-varying features and the shear stress time-varying features to obtain an injection molding time-varying feature set; The injection molding time-varying feature set is input into a preset sparse autoencoder model to calculate temperature uniformity and cooling efficiency, and obtain temperature uniformity and cooling efficiency indicators; specifically including: inputting the injection molding time-varying feature set into a preset sparse autoencoder model, and normalizing the injection molding time-varying feature set to obtain a normalized input vector; multiplying the normalized input vector with the first layer weight matrix and adding a first bias to obtain the linear output of the first hidden layer; performing a ReLU activation function operation on the linear output of the first hidden layer to obtain a first nonlinear feature representation of the first hidden layer; multiplying the first nonlinear feature representation with the second layer weight matrix and adding a second bias to obtain the linear output of the second hidden layer; performing a Sigmoid activation function operation on the linear output of the second hidden layer to obtain a second nonlinear feature representation of the second hidden layer; calculating the KL divergence based on the second nonlinear feature representation to obtain a sparsity penalty term, and multiplying the second nonlinear feature representation with the decoder weight matrix and adding a third bias to obtain. to the reconstructed output vector; feature analysis is performed on the reconstructed output vector to obtain a temperature field matrix and a corresponding time vector; the mean temperature and standard deviation of each moment are calculated according to the temperature field matrix to obtain a temperature statistical sequence, and the temperature statistical sequence is time-weighted averaged to obtain a comprehensive temperature uniformity index; the temperature gradient of each spatial point is calculated according to the temperature field matrix and the time vector to obtain a cooling rate field, and the cooling rate field is spatially averaged to obtain an average cooling rate curve; the characteristic time of the cooling process is calculated according to the average cooling rate curve to obtain a cooling efficiency index; Fourier transform is performed on the temperature field matrix to obtain a temperature field spectrum distribution, and energy concentration is calculated according to the temperature field spectrum distribution to obtain a temperature field frequency domain index; the temperature uniformity index, the cooling efficiency index and the temperature field frequency domain index are normalized to obtain a standardized index set; a multi-objective evaluation function is constructed according to the standardized index set, and temperature uniformity and cooling efficiency indexes are calculated according to the multi-objective evaluation function; According to the temperature uniformity and cooling efficiency indicators, parameter modeling and parameter optimization are performed on the cooling channel to obtain cooling channel optimization parameters.
2. The plastic mold-based injection molding method according to claim 1, characterized in that: The three-dimensional modeling of the plastic mold is performed to obtain a three-dimensional model of the mold including a cavity, a runner system and a cooling channel, including: Modeling the outer contour of the plastic mold to obtain a mold shell geometry model, and reverse engineering modeling the mold cavity of the plastic mold according to the product design drawing to obtain a mold cavity geometry model; Performing parting surface analysis on the cavity geometry model to obtain mold parting surface data, and performing parting design on the plastic mold according to the mold parting surface data to obtain initial geometry models of the movable mold and the fixed mold; Performing Boolean operations on the initial geometric models of the movable mold and the fixed mold to obtain a cavity structure model, and designing a runner system according to the cavity structure model to obtain a runner system model including a runner, a runner and a gate; Performing flow analysis on the runner system model to obtain melt flow path data, and designing an initial layout of the cooling channel according to the melt flow path data to obtain an initial model of the cooling channel; Processing the cooling channel initial model to obtain an adjustable cooling channel parameter set; The cavity structure model, the runner system model and the cooling channel initial model are integrated to obtain a three-dimensional mold model including the cavity, the runner system and the cooling channel.
3. The plastic mold-based injection molding method according to claim 1, characterized in that: The injection molding process simulation analysis is performed on the three-dimensional model of the mold to obtain temperature distribution data, pressure distribution data and shear stress distribution data inside the mold, including: Performing finite element mesh division on the three-dimensional model of the mold to obtain a mold mesh model, and assigning material properties to the mold mesh model according to the physical property parameters of the injection molding material to obtain a simulation model with material characteristics; Setting injection molding process parameters and boundary conditions for the simulation model to obtain an initial simulation condition set, and performing a filling phase analysis on the simulation model according to the initial simulation condition set to obtain melt flow front line data; Performing a time series analysis on the melt flow front line data to obtain a filling time distribution diagram, and performing a pressure holding phase analysis on the simulation model according to the filling time distribution diagram to obtain initial pressure field data inside the mold; Performing gradient calculation on the initial pressure field data to obtain pressure distribution data, and performing cooling phase analysis on the simulation model according to the pressure distribution data to obtain a numerical solution to the heat conduction equation; The numerical solution of the heat conduction equation is interpolated to obtain temperature distribution data, and the simulation model is subjected to rheological analysis based on the temperature distribution data and the pressure distribution data to obtain shear stress distribution data.
4. The plastic mold-based injection molding method according to claim 1, characterized in that: The step of performing parameter modeling and parameter optimization on the cooling channel according to the temperature uniformity and cooling efficiency index to obtain the cooling channel optimization parameters includes: Performing parameter modeling on the cooling channel to obtain a parameter vector describing the shape of the cooling channel; Constructing an objective function according to the parameter vector, wherein the objective function is used to comprehensively consider the optimization objectives of temperature uniformity and cooling efficiency; Applying geometric constraints and process constraints to the objective function to obtain a constrained optimization problem expression; Creating an initialized particle swarm according to the constrained optimization problem expression, and performing fitness evaluation on each particle in the initialized particle swarm to obtain an initial fitness value set; Update the individual optimal position and the global optimal position of the initialized particle swarm according to the initial fitness value set to obtain an updated particle swarm; updating the velocity and position of the updated particle swarm to obtain a new generation of particle swarm; The optimization iteration process is judged according to the preset termination condition to obtain a Boolean value for determining whether to continue the iteration; A conditional judgment is performed on the Boolean value. If the Boolean value is false, an optimal particle is obtained, and a cooling channel optimization parameter of the cooling channel is obtained according to the decoding of the optimal particle.
5. An injection molding device based on a plastic mold, characterized in that: Used to perform the injection molding method based on a plastic mold as described in any one of claims 1 to 4, the device comprises: A modeling module is used to perform three-dimensional modeling of the plastic mold to obtain a three-dimensional model of the mold including the cavity, runner system and cooling channel; A simulation module is used to simulate and analyze the injection molding process of the three-dimensional model of the mold to obtain temperature distribution data, pressure distribution data and shear stress distribution data inside the mold; The analysis module is used to perform adaptive slow feature analysis on the temperature distribution data, the pressure distribution data and the shear stress distribution data to obtain a time-varying feature set of injection molding; specifically comprising: adaptively sampling the temperature distribution data, the pressure distribution data and the shear stress distribution data to obtain a multi-scale time series; constructing a sliding time window according to the multi-scale time series, and generating a data subset sequence based on the sliding time window; recursively calculating the data subset sequence to obtain a time-varying covariance matrix estimate, and performing generalized eigenvalue decomposition on the time-varying covariance matrix estimate to obtain an adaptive eigenvector and an eigenvalue pair; performing nonlinear activation on the adaptive eigenvector to obtain a nonlinear slow feature map, and generating an adaptive eigenvector according to the nonlinear slow feature map. Injection constructs a time consistency constraint to obtain a slow feature optimization objective function; performs random gradient descent on the slow feature optimization objective function to obtain an optimal slow feature extractor, and performs feature extraction on the temperature distribution data, the pressure distribution data, and the shear stress distribution data according to the optimal slow feature extractor to obtain a temperature slow feature sequence, a pressure slow feature sequence, and a shear stress slow feature sequence; performs adaptive forgetting factor mechanism processing on the temperature slow feature sequence, the pressure slow feature sequence, and the shear stress slow feature sequence to obtain temperature time-varying features, pressure time-varying features, and shear stress time-varying features; performs feature fusion on the temperature time-varying features, the pressure time-varying features, and the shear stress time-varying features to obtain an injection molding time-varying feature set; A calculation module is used to input the injection molding time-varying feature set into a preset sparse autoencoder model to calculate temperature uniformity and cooling efficiency, and obtain temperature uniformity and cooling efficiency indicators; specifically comprising: inputting the injection molding time-varying feature set into a preset sparse autoencoder model, and normalizing the injection molding time-varying feature set to obtain a normalized input vector; multiplying the normalized input vector with the first layer weight matrix and adding a first bias to obtain the linear output of the first hidden layer; performing a ReLU activation function operation on the linear output of the first hidden layer to obtain a first nonlinear feature representation of the first hidden layer; multiplying the first nonlinear feature representation with the second layer weight matrix and adding a second bias to obtain the linear output of the second hidden layer; performing a Sigmoid activation function operation on the linear output of the second hidden layer to obtain a second nonlinear feature representation of the second hidden layer; calculating the KL divergence according to the second nonlinear feature representation to obtain a sparsity penalty term, and multiplying the second nonlinear feature representation with the decoder weight matrix and adding a third Bias, obtain a reconstructed output vector; perform feature analysis on the reconstructed output vector to obtain a temperature field matrix and a corresponding time vector; calculate the temperature mean and standard deviation at each moment according to the temperature field matrix to obtain a temperature statistical sequence, and perform time-weighted averaging on the temperature statistical sequence to obtain a comprehensive temperature uniformity index; calculate the temperature gradient of each spatial point according to the temperature field matrix and the time vector to obtain a cooling rate field, and perform spatial averaging on the cooling rate field to obtain an average cooling rate curve; calculate the characteristic time of the cooling process according to the average cooling rate curve to obtain a cooling efficiency index; perform Fourier transform on the temperature field matrix to obtain a temperature field spectrum distribution, and calculate the energy concentration according to the temperature field spectrum distribution to obtain a temperature field frequency domain index; normalize the temperature uniformity index, the cooling efficiency index and the temperature field frequency domain index to obtain a standardized index set; construct a multi-objective evaluation function according to the standardized index set, and calculate the temperature uniformity and cooling efficiency indexes according to the multi-objective evaluation function; The optimization module is used to perform parameter modeling and parameter optimization on the cooling channel according to the temperature uniformity and cooling efficiency indicators to obtain the cooling channel optimization parameters.
6. An injection molding device based on a plastic mold, characterized in that: The plastic mold-based injection molding device includes: a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instructions in the memory to enable the plastic mold-based injection molding equipment to execute the plastic mold-based injection molding method according to any one of claims 1-4.
7. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the plastic mold-based injection molding method as described in any one of claims 1 to 4 is implemented.
Citation Information
Patent Citations
Temperature control system and injection molding machine
CN114311576A
Light-weight injection mold manufacturing method, light-weight injection mold and injection product
CN115816717A
