Process parameter optimization control method and system for gas-heat synergistic driving injection molding

Abstract

The application provides a process parameter optimization control method and system for gas-heat synergistic driving injection molding, comprising formalizing a multi-objective optimization problem of gas-heat synergy based on product demand and equipment constraints; collecting gas-heat dynamic data in the injection molding process through a multi-modal sensor network, extracting physical-guided coupling characteristics from the gas-heat dynamic data; constructing a digital twin model of physical-data fusion, solving the digital twin model through multi-objective optimization, and obtaining an optimal process parameter set; deploying the optimal process parameter set to an injection molding machine, calculating actual coupling characteristic values by real-time monitoring of process parameters, evaluating the deviation of the actual coupling characteristic values from the predicted values of the digital twin model, triggering a feedback optimization mechanism to generate parameter correction amounts, and finally outputting the final process parameters of the next cycle. The application reduces the defect rate of injection molded products and production energy consumption, and improves the dimensional stability and molding quality consistency of products.

Description

Technical Field

[0001] This invention relates to the field of process parameter control technology, specifically to a method and system for optimizing and controlling process parameters in gas-heat co-driven injection molding. Background Technology

[0002] Injection molding, as one of the core technologies in polymer material processing, has been widely used in the mass production of complex structural parts in the automotive, electronics, and medical fields. With the continuous improvement of the demand for lightweighting and precision, gas-assisted injection molding (GAIM) technology has significantly reduced material consumption, clamping force, and molding internal stress by introducing high-pressure gas to assist melt filling in the hollow part, making it an important process direction for large thin-walled products and tubular parts.

[0003] In existing technologies, the optimization of injection molding process parameters mainly relies on trial-and-error methods based on empirical formulas or single-objective numerical simulation methods. The former involves repeated adjustments within the parameter range recommended in the process manual, which is not only time-consuming and costly, but also difficult to balance multiple objectives such as defect rate, energy consumption, and dimensional stability. The latter, while able to predict melt flow and temperature field distribution using finite element analysis, generally suffers from the following limitations: First, traditional numerical models often treat gas penetration and heat conduction as independent physical fields, lacking explicit characterization of the gas-heat dynamic coupling mechanism, resulting in insufficient prediction accuracy. Second, model calibration relies on offline experimental data and cannot respond in real time to production disturbances such as material batch fluctuations and mold wear. Third, optimization algorithms often employ weighted single-objective transformation or simple multi-objective evolution strategies without being guided by physical constraints, making it difficult to meet the requirements of online control in terms of solution quality and decision-making efficiency. In addition, existing control systems often use fixed threshold feedback, lacking the ability to predictively correct deviations in coupling characteristics. When the process state deviates from the design point, it is difficult to quickly generate the optimal parameter correction amount, resulting in insufficient product quality consistency. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method and system for optimizing and controlling process parameters in gas-heat co-driven injection molding, thereby solving the problems mentioned in the background section.

[0005] To achieve the above objectives, the present invention provides the following technical solution: 1. A method for optimizing and controlling process parameters in gas-heat synergistic injection molding, comprising the following steps:

[0006] S1: A formalized multi-objective optimization problem of gas-heat synergy based on product requirements and equipment constraints;

[0007] The multi-objective optimization problem includes defining an objective function, constraints, and a parameter space. The objective function includes defect rate, energy consumption, and dimensional stability. The constraints include equipment constraints, process constraints, and coupling constraints. The coupling constraint is that the ratio of gas penetration depth to thermal boundary layer thickness is within a preset threshold range.

[0008] S2: Collect gas-thermal dynamic data during the injection molding process through a multimodal sensor network, and extract the coupling characteristics of physical guidance from the gas-thermal dynamic data;

[0009] Among them, the coupling characteristic is the ratio of gas penetration depth to thermal boundary layer thickness;

[0010] S3: Construct a physical-data fusion digital twin model, perform multi-objective optimization on the digital twin model, and obtain the optimal set of process parameters;

[0011] S4: Deploy the optimal set of process parameters to the injection molding machine, monitor the process parameters in real time and calculate the actual coupling characteristic values, evaluate the deviation between them and the predicted values ​​of the digital twin model, trigger the feedback optimization mechanism to generate parameter correction amounts, and finally output the final process parameters for the next cycle.

[0012] As a further preferred embodiment, the specific steps of S2 include:

[0013] Based on the constraints defined by S1, a sensor network is deployed to collect production data of the injection molding process. Dynamic data related to the gas-heat interaction of the injection molding process is determined based on the production data to obtain process parameter data.

[0014] The coupling feature extraction guided by rational analysis is performed on the process parameter data;

[0015] The specific process of coupled feature extraction includes: based on the pressure sensor array data, using the analytical solution of the gas diffusion equation, the gas penetration depth is inferred from the arrival time of the gas front; the temperature gradient is extracted from the infrared thermal imager data to obtain the thermal boundary layer thickness;

[0016] The ratio of gas penetration depth to thermal boundary layer thickness is calculated in real time to obtain the coupling characteristics of the time series.

[0017] As a further preferred option, the specific process for constructing the physical-data fusion digital twin model includes:

[0018] Establish a gas penetration dynamics model to calculate the gas penetration depth and a thermal boundary layer heat transfer model to calculate the thermal boundary layer thickness;

[0019] ;

[0020] ;

[0021] in, The gas pressure is represented by t, and the time is represented by t. Indicates the solution permeability. Indicates porosity. Indicates gas viscosity. Indicates the gas source phase. Represents the Laplace operator. Indicates the density of the solution. Indicates the specific heat capacity of a solution. Indicates temperature. Indicates the thermal conductivity of the solution. Denotes the divergence operator, This indicates the gas convection heat transfer phase;

[0022] The ratio of gas penetration depth to thermal boundary layer thickness is embedded into the physical sub-model to obtain the predicted coupling characteristic value;

[0023] The data-driven sub-model uses a long short-term memory network to fit the nonlinear residual relationship between sensor data and coupling features;

[0024] The input features of the Long Short-Term Memory Network include raw sensor data, historical residual sequences, and material batch variation coefficients; the raw sensor data includes gas pressure, gas flow rate, and temperature gradient; the historical residual sequences are time-series data of the deviation between the predicted and actual values ​​of the physical sub-model.

[0025] A physics-data fusion mechanism is established to combine the theoretical predictions of the physical sub-model with the residual correction of the data-driven sub-model to obtain hybrid coupled feature predictions.

[0026] As a further preferred option, the physical-data fusion mechanism adopts an adaptive weighted fusion strategy:

[0027] ;

[0028] ;

[0029] in, This represents the predicted value of the hybrid coupling features. Indicates the weighting coefficient. This represents the prediction variance of the physical sub-model. Describe the variance of the data-driven sub-model.

[0030] As a further preferred option, the specific process for performing multi-objective optimization on the digital twin model to obtain the optimal set of process parameters includes:

[0031] The NSGA-III algorithm is used to iteratively evolve the formalized multi-objective optimization problem to obtain the Pareto front of the non-dominated solution set;

[0032] The TOPSIS method is used to select the solution with the highest relative closeness from the Pareto front as the optimal set of process parameters, and the predicted target values ​​of coupling features and their confidence intervals are output simultaneously.

[0033] As a further preferred embodiment, the process of configuring CIP system parameters based on a collaborative control strategy to execute cleaning operations, and dynamically adjusting the collaborative control strategy during the execution of cleaning operations, specifically includes:

[0034] Calculate the deviation between the actual coupling eigenvalues ​​and the coupling eigenvalues ​​predicted by the digital twin model, and simultaneously calculate the objective function deviation;

[0035] The deviation level is determined through a three-level threshold judgment mechanism:

[0036] If both the coupling eigenvalue deviation and the objective function deviation are less than the preset threshold lower limit, then the current process parameters are maintained.

[0037] If the deviation of the coupling feature value and / or the deviation of the objective function are greater than the lower limit of the preset threshold and less than the upper limit of the preset threshold, the secondary feedback mechanism is triggered.

[0038] If the deviation of the coupling feature value exceeds the preset threshold, or if a change in material batch or mold wear exceeds the threshold is detected, the digital twin model will be recalibrated.

[0039] When the secondary feedback optimization is triggered, a model predictive control framework is adopted to construct a locally linearized model at the current operating point and calculate the Jacobian matrix of the coupling feature to the process parameters.

[0040] ;

[0041] Where J represents the sensitivity matrix of the coupling feature to process parameters, and X represents the process parameters. This indicates the value to be taken at the X working point;

[0042] Solve the optimization problem within the predicted time domain period to obtain the optimal parameter correction amount;

[0043] The formula for solving optimization problems is:

[0044] ;

[0045] in, The value represents the length of the prediction time domain, and j represents the period index within the prediction time domain. Represents the target coupling eigenvalue. Let Q represent the weighted square norm, and let Q be the tracking accuracy weight matrix. Let R denote the weighted square norm, and R be the parameter adjustment cost weight matrix.

[0046] The optimal correction is superimposed on the original optimal parameter set to generate the final process parameters for the next cycle.

[0047] As a further preferred embodiment, a process parameter optimization and control system for gas-heat co-driven injection molding, used to implement the above-mentioned process parameter optimization and control method for gas-heat co-driven injection molding, includes:

[0048] The problem formalization module formalizes multi-objective optimization problems involving gas and heat synergy based on product requirements and equipment constraints.

[0049] The multi-objective optimization problem includes defining an objective function, constraints, and a parameter space. The objective function includes defect rate, energy consumption, and dimensional stability. The constraints include equipment constraints, process constraints, and coupling constraints. The coupling constraint is that the ratio of gas penetration depth to thermal boundary layer thickness is within a preset threshold range.

[0050] The computing module collects gas-thermal dynamic data during the injection molding process through a multimodal sensor network, and extracts the coupling features of physical guidance from the gas-thermal dynamic data.

[0051] Among them, the coupling characteristic is the ratio of gas penetration depth to thermal boundary layer thickness;

[0052] The optimization module calculates the theoretical liquid volume requirement based on the target flow rate and pipeline volume of the preset cleaning program, and compares it with real-time liquid volume data to obtain the adjustment amount.

[0053] The correction module is used to deploy the optimal set of process parameters to the injection molding machine. By monitoring the process parameters in real time and calculating the actual coupling characteristic values, it evaluates the deviation between the parameters and the predicted values ​​of the digital twin model, triggers the feedback optimization mechanism to generate parameter correction amounts, and finally outputs the final process parameters for the next cycle.

[0054] As a further preferred option, the specific process for constructing the physical-data fusion digital twin model includes:

[0055] Establish a gas penetration dynamics model to calculate the gas penetration depth and a thermal boundary layer heat transfer model to calculate the thermal boundary layer thickness;

[0056] ;

[0057] ;

[0058] in, The gas pressure is represented by t, and the time is represented by t. Indicates the solution permeability. Indicates porosity. Indicates gas viscosity. Indicates the gas source phase. Represents the Laplace operator. Indicates the density of the solution. Indicates the specific heat capacity of a solution. Indicates temperature. Indicates the thermal conductivity of the solution. Denotes the divergence operator, This indicates the gas convection heat transfer phase;

[0059] The ratio of gas penetration depth to thermal boundary layer thickness is embedded into the physical sub-model to obtain the predicted coupling characteristic value;

[0060] The data-driven sub-model uses a long short-term memory network to fit the nonlinear residual relationship between sensor data and coupling features;

[0061] The input features of the Long Short-Term Memory Network include raw sensor data, historical residual sequences, and material batch variation coefficients; the raw sensor data includes gas pressure, gas flow rate, and temperature gradient; the historical residual sequences are time-series data of the deviation between the predicted and actual values ​​of the physical sub-model.

[0062] A physics-data fusion mechanism is established to combine the theoretical predictions of the physics sub-model with the residual correction of the data-driven sub-model to obtain hybrid coupling feature predictions.

[0063] The physical-data fusion mechanism employs an adaptive weighted fusion strategy:

[0064] ;

[0065] ;

[0066] in, This represents the predicted value of the hybrid coupling features. Indicates the weighting coefficient. This represents the prediction variance of the physical sub-model. Describe the variance of the data-driven sub-model.

[0067] As a further preferred option, the specific process for performing multi-objective optimization on the digital twin model to obtain the optimal set of process parameters includes:

[0068] The NSGA-III algorithm is used to iteratively evolve the formalized multi-objective optimization problem to obtain the Pareto front of the non-dominated solution set;

[0069] The TOPSIS method is used to select the solution with the highest relative closeness from the Pareto front as the optimal set of process parameters, and the predicted target values ​​of coupling features and their confidence intervals are output simultaneously.

[0070] As a further preferred embodiment, the process of configuring CIP system parameters based on a collaborative control strategy to execute cleaning operations, and dynamically adjusting the collaborative control strategy during the execution of cleaning operations, specifically includes:

[0071] Calculate the deviation between the actual coupling eigenvalues ​​and the coupling eigenvalues ​​predicted by the digital twin model, and simultaneously calculate the objective function deviation;

[0072] The deviation level is determined through a three-level threshold judgment mechanism:

[0073] If both the coupling eigenvalue deviation and the objective function deviation are less than the preset threshold lower limit, then the current process parameters are maintained.

[0074] If the deviation of the coupling feature value and / or the deviation of the objective function are greater than the lower limit of the preset threshold and less than the upper limit of the preset threshold, the secondary feedback mechanism is triggered.

[0075] If the deviation of the coupling feature value exceeds the preset threshold, or if a change in material batch or mold wear exceeds the threshold is detected, the digital twin model will be recalibrated.

[0076] When the secondary feedback optimization is triggered, a model predictive control framework is adopted to construct a locally linearized model at the current operating point and calculate the Jacobian matrix of the coupling feature to the process parameters.

[0077] ;

[0078] Where J represents the sensitivity matrix of the coupling feature to process parameters, and X represents the process parameters. This indicates the value to be taken at the X working point;

[0079] Solve the optimization problem within the predicted time domain period to obtain the optimal parameter correction amount;

[0080] The formula for solving optimization problems is:

[0081] ;

[0082] in, The value represents the length of the prediction time domain, and j represents the period index within the prediction time domain. Represents the target coupling eigenvalue. Let Q represent the weighted square norm, and let Q be the tracking accuracy weight matrix. Let R denote the weighted square norm, and R be the parameter adjustment cost weight matrix.

[0083] The optimal correction is superimposed on the original optimal parameter set to generate the final process parameters for the next cycle.

[0084] This invention provides a method and system for optimizing and controlling process parameters in gas-thermal co-driven injection molding, which has the following advantages: It collects gas-thermal dynamic data during the injection molding process through a multimodal sensor network, determines process parameters related to the gas-thermal co-driven injection molding process control based on the gas-thermal dynamic data, including gas pressure, melt temperature, mold temperature, and injection rate; it determines a multi-objective optimization problem description for gas-thermal co-driven injection based on the process parameters, establishes a physical-data fusion digital twin model based on the multi-objective optimization problem description, and uses the NSGA-III algorithm to optimize the multi-objective process parameters of the digital twin model to obtain the Pareto optimal solution set; it designs a TOPSIS decision strategy based on the Pareto optimal solution set, calculates the coupling characteristic deviation based on real-time monitoring data of the injection molding process, and designs a model predictive control feedback optimization mechanism based on the deviation; it dynamically corrects the process parameters of the current cycle based on the TOPSIS decision strategy and the feedback optimization mechanism, and outputs the final process parameters for the next cycle. This application combines the requirements of injection molded products with equipment constraints, analyzes the key contradictions in the optimization and control of the gas-heat synergistic injection molding process, establishes a coupled characteristic model of gas penetration depth and thermal boundary layer thickness, and achieves intelligent optimization of process parameters through a physical-data fusion digital twin model and multi-objective optimization algorithm. This reduces the defect rate and production energy consumption of injection molded products, improves the dimensional stability and molding quality consistency of products, and enhances the adaptive ability of the production process to disturbances such as material variation and mold wear. Attached Figure Description

[0085] Figure 1 This is a flowchart of the process parameter optimization and control method for gas-heat synergistic injection molding according to the present invention;

[0086] Figure 2 This is a block diagram of the process parameter optimization control system for gas-heat co-driven injection molding according to the present invention. Detailed Implementation

[0087] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0088] The following disclosure provides many different embodiments or examples for implementing various structures of the invention. To simplify the disclosure, specific examples of components and arrangements are described below. These are merely examples and are not intended to limit the invention. Furthermore, reference numerals and / or letters may be repeated in different examples; such repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed. In addition, examples of various specific processes and materials are provided in this invention, but those skilled in the art will recognize the application of other processes and / or the use of other materials.

[0089] like Figure 1 As shown, this embodiment of the invention provides a method for optimizing and controlling process parameters in gas-heat co-driven injection molding, comprising the following steps:

[0090] S1: A formalized multi-objective optimization problem of gas-heat synergy based on product requirements and equipment constraints;

[0091] Among them, multi-objective optimization problems include defining the objective function, constraints, and parameter space;

[0092] The objective function includes defect rate, energy consumption and dimensional stability, and the constraints include equipment constraints, process constraints and coupling constraints. The coupling constraint is that the ratio of gas penetration depth to thermal boundary layer thickness is within a preset threshold range.

[0093] Specifically, by acquiring product target parameter data and equipment physical constraints, and formally defining the objective function, constraints, and parameter space under gas-heat synergistic drive, a formalized multi-objective optimization problem is obtained, the calculation formula of which is:

[0094] ;

[0095] in, Indicates the defect rate. Indicates energy consumption. denoted by dimensional stability, and T denotes transpose.

[0096] Specifically, acquiring target parameter data for the product includes product geometry, material properties, quality requirements, and historical production data; acquiring physical limitations of the equipment involves extracting constraints from the injection molding machine specifications, such as maximum gas pressure, melt temperature range, upper limit of mold cooling rate, and gas supply system capacity. These collected data determine the parameter space boundaries and constraints.

[0097] Specifically, in the formal definition of the objective function, the defect rate is quantified based on product requirements, energy consumption is the total energy consumption per cycle, and dimensional stability is specifically the standard deviation of the product's critical dimensions; in the formal definition of the constraints, equipment constraints originate from the physical limitations of the equipment, process constraints originate from the injection molding mechanism, and coupling constraints are defined as the ratio of gas penetration depth to thermal boundary layer thickness being within a preset threshold range; the parameter space is formally defined as a vector of process parameters. , which indicates Gas pressure, Indicates the melt temperature. Indicates the mold temperature. Indicates the injection rate.

[0098] It should be noted that the coupling constraint originates from the gas-heat interaction mechanism. If the coupling constraint is much smaller than the preset threshold, insufficient gas diffusion will lead to uneven filling. If the coupling constraint is much larger than the preset threshold, the thermal boundary layer will be too thin, causing melt degradation. The coupling constraint is usually set to 1. The parameter range of the parameter space is jointly limited by equipment constraints and process constraints.

[0099] S2: Collect gas-thermal dynamic data during the injection molding process through a multimodal sensor network, and extract the coupling characteristics of physical guidance from the gas-thermal dynamic data;

[0100] The coupling characteristic is the ratio of gas penetration depth to thermal boundary layer thickness.

[0101] Specifically, based on the constraints defined by S1, a sensor network is deployed to collect production data of the injection molding process. Based on the production data, dynamic data related to the gas-heat interaction of the injection molding process is determined to obtain process parameter data.

[0102] The coupling feature extraction guided by rational analysis is performed on the process parameter data;

[0103] The multimodal sensor network includes pressure sensors and flow meters on the gas side, and thermocouple arrays and infrared thermal imagers on the thermal side; coupled feature extraction uses the gas diffusion equation to invert the gas penetration depth and calculates the thermal boundary layer thickness using the temperature gradient.

[0104] The specific process of coupled feature extraction includes: based on the pressure sensor array data, using the analytical solution of the gas diffusion equation, the gas penetration depth is inferred from the arrival time of the gas front; the temperature gradient is extracted from the infrared thermal imager data to obtain the thermal boundary layer thickness;

[0105] The ratio of gas penetration depth to thermal boundary layer thickness is calculated in real time to obtain the coupling characteristics of the time series.

[0106] S3: Construct a physical-data fusion digital twin model, perform multi-objective optimization on the digital twin model, and obtain the optimal set of process parameters;

[0107] The model includes a physical sub-model based on fluid dynamics and heat transfer equations, a data-driven sub-model that fits the relationship between sensor data and coupling characteristics, and a fusion mechanism that combines the predictions of the physical sub-model with the residual correction of the data-driven sub-model.

[0108] Specifically, the process of building a physical-data fusion digital twin model includes:

[0109] Establish a gas penetration dynamics model to calculate the gas penetration depth and a thermal boundary layer heat transfer model to calculate the thermal boundary layer thickness;

[0110] ;

[0111] ;

[0112] in, The gas pressure is represented by t, and the time is represented by t. Indicates the solution permeability. Indicates porosity. Indicates gas viscosity. Indicates the gas source phase. Represents the Laplace operator. Indicates the density of the solution. Indicates the specific heat capacity of a solution. Indicates temperature. Indicates the thermal conductivity of the solution. Denotes the divergence operator, This indicates the gas convection heat transfer phase.

[0113] Understandably, the gas penetration dynamics model and the thermal boundary layer heat transfer model together constitute the physical sub-model.

[0114] The ratio of gas penetration depth to thermal boundary layer thickness is embedded into the physical sub-model to obtain the predicted coupling characteristic value;

[0115] The data-driven sub-model uses a long short-term memory network to fit the nonlinear residual relationship between sensor data and coupling features;

[0116] The input features of the Long Short-Term Memory Network include raw sensor data, historical residual sequences, and material batch variation coefficients. The raw sensor data includes gas pressure, gas flow rate, and temperature gradient. The historical residual sequences are time-series data of the deviation between the predicted and actual values ​​of the physical sub-model.

[0117] And the formula for calculating the residual is: ,in Represents the residual. Represents the actual coupling eigenvalues. This represents the coupling characteristic value predicted by the physical sub-model.

[0118] A physics-data fusion mechanism is established to combine the theoretical predictions of the physics sub-model with the residual correction of the data-driven sub-model to obtain hybrid coupling feature predictions.

[0119] Specifically, the physical-data fusion mechanism adopts an adaptive weighted fusion strategy:

[0120] ;

[0121] ;

[0122] in, This represents the predicted value of the hybrid coupling features. Indicates the weighting coefficient. This represents the prediction variance of the physical sub-model. Describe the variance of the data-driven sub-model.

[0123] In this embodiment, the specific process for performing multi-objective optimization on the digital twin model to obtain the optimal set of process parameters includes:

[0124] Specifically, the NSGA-III algorithm is used to iteratively evolve the formalized multi-objective optimization problem to obtain the Pareto front of the non-dominated solution set.

[0125] The TOPSIS method is used to select the solution with the highest relative closeness from the Pareto front as the optimal set of process parameters, and the predicted target values ​​of coupling features and their confidence intervals are output simultaneously.

[0126] It should be noted that each solution in the non-dominated solution set corresponds to a different gas-heat synergistic trade-off strategy.

[0127] S4: Deploy the optimal set of process parameters to the injection molding machine, monitor the process parameters in real time and calculate the actual coupling characteristic values, evaluate the deviation between them and the predicted values ​​of the digital twin model, trigger the feedback optimization mechanism to generate parameter correction amounts, and finally output the final process parameters for the next cycle.

[0128] Specifically, the deviation between the actual coupling feature value and the coupling feature value predicted by the digital twin model is calculated, and the deviation of the objective function is calculated simultaneously.

[0129] The deviation level is determined through a three-level threshold judgment mechanism:

[0130] If both the coupling eigenvalue deviation and the objective function deviation are less than the preset threshold lower limit, then the current process parameters are maintained.

[0131] If the deviation of the coupling feature value and / or the deviation of the objective function are greater than the lower limit of the preset threshold and less than the upper limit of the preset threshold, the secondary feedback mechanism is triggered.

[0132] If the deviation of the coupling feature value exceeds the preset threshold, or if a change in material batch or mold wear exceeds the threshold is detected, the digital twin model will be recalibrated.

[0133] It should be noted that the calibration of the digital twin model is based on existing technology and will not be described in detail here.

[0134] When the secondary feedback optimization is triggered, a model predictive control framework is adopted to construct a locally linearized model at the current operating point and calculate the Jacobian matrix of the coupling feature to the process parameters.

[0135] ;

[0136] Where J represents the sensitivity matrix of the coupling feature to process parameters, and X represents the process parameters. This indicates the value to be taken at the X working point;

[0137] Solve the optimization problem within the predicted time domain period to obtain the optimal parameter correction amount;

[0138] The formula for solving optimization problems is:

[0139] ;

[0140] in, The value represents the length of the prediction time domain, and j represents the period index within the prediction time domain. Represents the target coupling eigenvalue. Let Q represent the weighted square norm, and let Q be the tracking accuracy weight matrix. Let R denote the weighted square norm, and R be the parameter adjustment cost weight matrix.

[0141] It should be noted that the calculation formula for solving optimization problems includes equipment constraints and coupling constraints.

[0142] The optimal correction is superimposed on the original optimal parameter set to generate the final process parameters for the next cycle.

[0143] This application discloses a process parameter optimization and control method for gas-thermal co-driven injection molding. The method collects gas-thermal dynamic data during the injection molding process using a multimodal sensor network. Based on this data, it determines the process parameters related to the gas-thermal co-driven injection molding process control, including gas pressure, melt temperature, mold temperature, and injection rate. A multi-objective optimization problem description for gas-thermal co-driven molding is then established based on these parameters. A physical-data fusion digital twin model is built, and the NSGA-III algorithm is used to optimize the multi-objective process parameters of the digital twin model, obtaining the Pareto optimal solution set. A TOPSIS decision strategy is designed based on the Pareto optimal solution set. The coupling characteristic deviation is calculated based on real-time monitoring data of the injection molding process, and a model predictive control feedback optimization mechanism is designed based on this deviation. Finally, the process parameters for the current cycle are dynamically corrected based on the TOPSIS decision strategy and the feedback optimization mechanism, and the final process parameters for the next cycle are output. This application combines the requirements of injection molded products with equipment constraints, analyzes the key contradictions in the optimization and control of the gas-heat synergistic injection molding process, establishes a coupled characteristic model of gas penetration depth and thermal boundary layer thickness, and achieves intelligent optimization of process parameters through a physical-data fusion digital twin model and multi-objective optimization algorithm. This reduces the defect rate and production energy consumption of injection molded products, improves the dimensional stability and molding quality consistency of products, and enhances the adaptive ability of the production process to disturbances such as material variation and mold wear.

[0144] like Figure 2 As shown, this embodiment also provides a process parameter optimization and control system for gas-heat co-driven injection molding, used to implement the above-mentioned process parameter optimization and control method for gas-heat co-driven injection molding, including:

[0145] The problem formalization module formalizes multi-objective optimization problems involving gas and heat synergy based on product requirements and equipment constraints.

[0146] The multi-objective optimization problem includes defining the objective function, constraints, and parameter space. The objective function includes defect rate, energy consumption, and dimensional stability. The constraints include equipment constraints, process constraints, and coupling constraints. The coupling constraint is that the ratio of gas penetration depth to thermal boundary layer thickness is within a preset threshold range.

[0147] The computing module collects gas-thermal dynamic data during the injection molding process through a multimodal sensor network, and extracts the coupling features of physical guidance from the gas-thermal dynamic data.

[0148] Among them, the coupling characteristic is the ratio of gas penetration depth to thermal boundary layer thickness;

[0149] The optimization module calculates the theoretical liquid volume requirement based on the target flow rate and pipeline volume of the preset cleaning program, and compares it with real-time liquid volume data to obtain the adjustment amount.

[0150] The correction module is used to deploy the optimal set of process parameters to the injection molding machine. By monitoring the process parameters in real time and calculating the actual coupling characteristic values, it evaluates the deviation between the parameters and the predicted values ​​of the digital twin model, triggers the feedback optimization mechanism to generate parameter correction amounts, and finally outputs the final process parameters for the next cycle.

[0151] The specific process for constructing a physical-data fusion digital twin model includes:

[0152] Establish a gas penetration dynamics model to calculate the gas penetration depth and a thermal boundary layer heat transfer model to calculate the thermal boundary layer thickness;

[0153] ;

[0154] ;

[0155] in, The gas pressure is represented by t, and the time is represented by t. Indicates the solution permeability. Indicates porosity. Indicates gas viscosity. Indicates the gas source phase. Represents the Laplace operator. Indicates the density of the solution. Indicates the specific heat capacity of a solution. Indicates temperature. Indicates the thermal conductivity of the solution. Denotes the divergence operator, This indicates the gas convection heat transfer phase;

[0156] The ratio of gas penetration depth to thermal boundary layer thickness is embedded into the physical sub-model to obtain the predicted coupling characteristic value;

[0157] The data-driven sub-model uses a long short-term memory network to fit the nonlinear residual relationship between sensor data and coupling features;

[0158] The input features of the Long Short-Term Memory Network include raw sensor data, historical residual sequences, and material batch variation coefficients; the raw sensor data includes gas pressure, gas flow rate, and temperature gradient; the historical residual sequences are time-series data of the deviation between the predicted and actual values ​​of the physical sub-model.

[0159] A physics-data fusion mechanism is established to combine the theoretical predictions of the physics sub-model with the residual correction of the data-driven sub-model to obtain hybrid coupling feature predictions.

[0160] The physical-data fusion mechanism employs an adaptive weighted fusion strategy:

[0161] ;

[0162] ;

[0163] in, This represents the predicted value of the hybrid coupling features. Indicates the weighting coefficient. This represents the prediction variance of the physical sub-model. Describe the variance of the data-driven sub-model.

[0164] In this embodiment, the specific process for performing multi-objective optimization on the digital twin model to obtain the optimal set of process parameters includes:

[0165] The NSGA-III algorithm is used to iteratively evolve the formalized multi-objective optimization problem to obtain the Pareto front of the non-dominated solution set;

[0166] The TOPSIS method is used to select the solution with the highest relative closeness from the Pareto front as the optimal set of process parameters, and the predicted target values ​​of coupling features and their confidence intervals are output simultaneously.

[0167] The process of configuring CIP system parameters based on a collaborative control strategy to execute cleaning operations, and dynamically adjusting the collaborative control strategy during the execution of cleaning operations, specifically includes:

[0168] Calculate the deviation between the actual coupling eigenvalues ​​and the coupling eigenvalues ​​predicted by the digital twin model, and simultaneously calculate the objective function deviation;

[0169] The deviation level is determined through a three-level threshold judgment mechanism:

[0170] If both the coupling eigenvalue deviation and the objective function deviation are less than the preset threshold lower limit, then the current process parameters are maintained.

[0171] If the deviation of the coupling feature value and / or the deviation of the objective function are greater than the lower limit of the preset threshold and less than the upper limit of the preset threshold, the secondary feedback mechanism is triggered.

[0172] If the deviation of the coupling feature value exceeds the preset threshold, or if a change in material batch or mold wear exceeds the threshold is detected, the digital twin model will be recalibrated.

[0173] When the secondary feedback optimization is triggered, a model predictive control framework is adopted to construct a locally linearized model at the current operating point and calculate the Jacobian matrix of the coupling feature to the process parameters.

[0174] ;

[0175] Where J represents the sensitivity matrix of the coupling feature to process parameters, and X represents the process parameters. This indicates the value to be taken at the X working point;

[0176] Solve the optimization problem within the predicted time domain period to obtain the optimal parameter correction amount;

[0177] The formula for solving optimization problems is:

[0178] ;

[0179] in, The value represents the length of the prediction time domain, and j represents the period index within the prediction time domain. Represents the target coupling eigenvalue. Let Q represent the weighted square norm, and let Q be the tracking accuracy weight matrix. Let R denote the weighted square norm, and R be the parameter adjustment cost weight matrix.

[0180] The optimal correction is superimposed on the original optimal parameter set to generate the final process parameters for the next cycle.

[0181] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for optimizing and controlling process parameters in gas-heat co-driven injection molding, characterized in that, Includes the following steps: S1: A formalized multi-objective optimization problem of gas-heat synergy based on product requirements and equipment constraints; The multi-objective optimization problem includes defining an objective function, constraints, and a parameter space. The objective function includes defect rate, energy consumption, and dimensional stability. The constraints include equipment constraints, process constraints, and coupling constraints. The coupling constraint is that the ratio of gas penetration depth to thermal boundary layer thickness is within a preset threshold range. S2: Collect gas-thermal dynamic data during the injection molding process through a multimodal sensor network, and extract the physical guidance coupling characteristics from the gas-thermal dynamic data; Among them, the coupling characteristic is the ratio of gas penetration depth to thermal boundary layer thickness; S3: Construct a physical-data fusion digital twin model, perform multi-objective optimization on the digital twin model, and obtain the optimal set of process parameters; S4: Deploy the optimal set of process parameters to the injection molding machine, monitor the process parameters in real time and calculate the actual coupling characteristic values, evaluate the deviation between them and the predicted values ​​of the digital twin model, trigger the feedback optimization mechanism to generate parameter correction, and finally output the final process parameters for the next cycle. The specific process for constructing a physical-data fusion digital twin model includes: The ratio of gas penetration depth to thermal boundary layer thickness is embedded into the physical sub-model to obtain the predicted coupling characteristic value; The data-driven sub-model uses a long short-term memory network to fit the nonlinear residual relationship between sensor data and coupling features; The input features of the Long Short-Term Memory Network include raw sensor data, historical residual sequences, and material batch variation coefficients; the raw sensor data includes gas pressure, gas flow rate, and temperature gradient; the historical residual sequences are time-series data of the deviation between the predicted and actual values ​​of the physical sub-model. A physics-data fusion mechanism is established to combine the theoretical predictions of the physical sub-model with the residual correction of the data-driven sub-model to obtain hybrid coupled feature predictions.

2. The method for optimizing and controlling process parameters in gas-heat co-driven injection molding according to claim 1, characterized in that, The specific steps of S2 include: Based on the constraints defined by S1, a sensor network is deployed to collect production data of the injection molding process. Dynamic data related to the gas-heat interaction of the injection molding process is determined based on the production data to obtain process parameter data. Physically guided coupling feature extraction is performed on process parameter data; The specific process of coupled feature extraction includes: based on the pressure sensor array data, using the analytical solution of the gas diffusion equation, the gas penetration depth is inferred from the arrival time of the gas front; the temperature gradient is extracted from the infrared thermal imager data to obtain the thermal boundary layer thickness; The ratio of gas penetration depth to thermal boundary layer thickness is calculated in real time to obtain the coupling characteristics of the time series.

3. The method for optimizing and controlling process parameters in gas-heat co-driven injection molding according to claim 1, characterized in that, The specific process for constructing a physical-data fusion digital twin model also includes: Establish a gas penetration dynamics model to calculate the gas penetration depth and a thermal boundary layer heat transfer model to calculate the thermal boundary layer thickness; ； ； in, The gas pressure is represented by t, and the time is represented by t. Indicates the solution permeability. Indicates porosity. Indicates gas viscosity. Indicates the gas source phase. Represents the Laplace operator. Indicates the density of the solution. Indicates the specific heat capacity of a solution. Indicates temperature. Indicates the thermal conductivity of the solution. Denotes the divergence operator, This indicates the gas convection heat transfer phase.

4. The method for optimizing and controlling process parameters in gas-heat co-driven injection molding according to claim 3, characterized in that, The physical-data fusion mechanism employs an adaptive weighted fusion strategy: ； ； in, This represents the predicted value of the hybrid coupling features. This represents the coupling eigenvalues ​​predicted by the physical sub-model. Indicates the weighting coefficient. Represents the residual. This represents the prediction variance of the physical sub-model. Describe the variance of the data-driven sub-model.

5. The method for optimizing and controlling process parameters in gas-heat co-driven injection molding according to claim 1, characterized in that, The specific process for solving the digital twin model through multi-objective optimization to obtain the optimal set of process parameters includes: The NSGA-III algorithm is used to iteratively evolve the formalized multi-objective optimization problem to obtain the Pareto front of the non-dominated solution set; The TOPSIS method is used to select the solution with the highest relative closeness from the Pareto front as the optimal set of process parameters, and the predicted target values ​​of coupling features and their confidence intervals are output simultaneously.

6. The method for optimizing and controlling process parameters in gas-heat co-driven injection molding according to claim 1, characterized in that, The process of evaluating the deviation from the digital twin model's prediction, triggering a feedback optimization mechanism to generate parameter corrections, and finally outputting the final process parameters for the next cycle specifically includes: Calculate the deviation between the actual coupling feature values ​​and the coupling feature values ​​predicted by the digital twin model, and simultaneously calculate the objective function deviation; The deviation level is determined through a three-level threshold judgment mechanism: If both the coupling eigenvalue deviation and the objective function deviation are less than the preset threshold lower limit, then the current process parameters are maintained. If the deviation of the coupling feature value and / or the deviation of the objective function are greater than the lower limit of the preset threshold and less than the upper limit of the preset threshold, the secondary feedback mechanism is triggered. If the deviation of the coupling feature value is greater than the upper limit of the preset threshold, or if a change in material batch or mold wear exceeds the threshold is detected, the digital twin model will be recalibrated. When the secondary feedback optimization is triggered, a model predictive control framework is adopted to construct a locally linearized model at the current operating point and calculate the Jacobian matrix of the coupling feature to the process parameters. ； Where J represents the sensitivity matrix of the coupling feature to process parameters. X represents the coupling eigenvalues ​​predicted by the physical sub-model, and X represents the process parameters. This indicates the value to be taken at the X working point; Solve the optimization problem within the predicted time domain period to obtain the optimal parameter correction amount; The formula for solving optimization problems is: ； in, The value represents the length of the prediction time domain, and j represents the period index within the prediction time domain. Represents the target coupling eigenvalue. This represents the predicted value of the hybrid coupling features. Let Q represent the weighted square norm, and let Q be the tracking accuracy weight matrix. Let R represent the weighted square norm, and R be the parameter adjustment cost weight matrix. The optimal correction is superimposed on the original optimal parameter set to generate the final process parameters for the next cycle.

7. A process parameter optimization and control system for gas-heat co-driven injection molding, used to implement the process parameter optimization and control method for gas-heat co-driven injection molding as described in any one of claims 1-6, characterized in that, include: The problem formalization module formalizes multi-objective optimization problems involving gas and heat synergy based on product requirements and equipment constraints. The multi-objective optimization problem includes defining an objective function, constraints, and a parameter space. The objective function includes defect rate, energy consumption, and dimensional stability. The constraints include equipment constraints, process constraints, and coupling constraints. The coupling constraint is that the ratio of gas penetration depth to thermal boundary layer thickness is within a preset threshold range. The computing module collects gas-thermal dynamic data during the injection molding process through a multimodal sensor network and extracts the physical guidance coupling characteristics from the gas-thermal dynamic data. Among them, the coupling characteristic is the ratio of gas penetration depth to thermal boundary layer thickness; The optimization module is used to construct a physical-data fusion digital twin model, perform multi-objective optimization on the digital twin model, and obtain the optimal set of process parameters. The specific process for constructing a physical-data fusion digital twin model includes: The ratio of gas penetration depth to thermal boundary layer thickness is embedded into the physical sub-model to obtain the predicted coupling characteristic value; The data-driven sub-model uses a long short-term memory network to fit the nonlinear residual relationship between sensor data and coupling features; The input features of the Long Short-Term Memory Network include raw sensor data, historical residual sequences, and material batch variation coefficients; the raw sensor data includes gas pressure, gas flow rate, and temperature gradient; the historical residual sequences are time-series data of the deviation between the predicted and actual values ​​of the physical sub-model. A physics-data fusion mechanism is established to combine the theoretical predictions of the physics sub-model with the residual correction of the data-driven sub-model to obtain hybrid coupling feature predictions. The correction module is used to deploy the optimal set of process parameters to the injection molding machine. By monitoring the process parameters in real time and calculating the actual coupling characteristic values, it evaluates the deviation between the parameters and the predicted values ​​of the digital twin model, triggers the feedback optimization mechanism to generate parameter correction amounts, and finally outputs the final process parameters for the next cycle.

8. A process parameter optimization and control system for gas-heat co-driven injection molding according to claim 7, characterized in that, The specific process for constructing a physical-data fusion digital twin model also includes: Establish a gas penetration dynamics model to calculate the gas penetration depth and a thermal boundary layer heat transfer model to calculate the thermal boundary layer thickness; ； ； in, The gas pressure is represented by t, and the time is represented by t. Indicates the solution permeability. Indicates porosity. Indicates gas viscosity. Indicates the gas source phase. Represents the Laplace operator. Indicates the density of the solution. Indicates the specific heat capacity of a solution. Indicates temperature. Indicates the thermal conductivity of the solution. Denotes the divergence operator, Indicates the gas convection heat transfer phase; The physical-data fusion mechanism employs an adaptive weighted fusion strategy: ； ； in, This represents the predicted value of the hybrid coupling features. This represents the coupling eigenvalues ​​predicted by the physical sub-model. Indicates the weighting coefficient. Represents the residual. This represents the prediction variance of the physical sub-model. Describe the variance of the data-driven sub-model.

9. A process parameter optimization and control system for gas-heat co-driven injection molding according to claim 7, characterized in that, The specific process for solving the digital twin model through multi-objective optimization to obtain the optimal set of process parameters includes: The NSGA-III algorithm is used to iteratively evolve the formalized multi-objective optimization problem to obtain the Pareto front of the non-dominated solution set; The TOPSIS method is used to select the solution with the highest relative closeness from the Pareto front as the optimal set of process parameters, and the predicted target values ​​of coupling features and their confidence intervals are output simultaneously.

10. A process parameter optimization and control system for gas-heat co-driven injection molding according to claim 7, characterized in that, The process of evaluating the deviation from the digital twin model's prediction, triggering a feedback optimization mechanism to generate parameter corrections, and finally outputting the final process parameters for the next cycle specifically includes: Calculate the deviation between the actual coupling feature values ​​and the coupling feature values ​​predicted by the digital twin model, and simultaneously calculate the objective function deviation; The deviation level is determined through a three-level threshold judgment mechanism: If both the coupling eigenvalue deviation and the objective function deviation are less than the preset threshold lower limit, then the current process parameters are maintained. If the deviation of the coupling feature value and / or the deviation of the objective function are greater than the lower limit of the preset threshold and less than the upper limit of the preset threshold, the secondary feedback mechanism is triggered. If the deviation of the coupling feature value is greater than the upper limit of the preset threshold, or if a change in material batch or mold wear exceeds the threshold is detected, the digital twin model will be recalibrated. When the secondary feedback optimization is triggered, a model predictive control framework is adopted to construct a locally linearized model at the current operating point and calculate the Jacobian matrix of the coupling feature to the process parameters. ； Where J represents the sensitivity matrix of the coupling feature to process parameters. X represents the coupling eigenvalues ​​predicted by the physical sub-model, and X represents the process parameters. This indicates the value to be taken at the X working point; Solve the optimization problem within the predicted time domain period to obtain the optimal parameter correction amount; The formula for solving optimization problems is: ； in, The value represents the length of the prediction time domain, and j represents the period index within the prediction time domain. Represents the target coupling eigenvalue. This represents the predicted value of the hybrid coupling features. Let Q represent the weighted square norm, and let Q be the tracking accuracy weight matrix. Let R represent the weighted square norm, and R be the parameter adjustment cost weight matrix. The optimal correction is superimposed on the original optimal parameter set to generate the final process parameters for the next cycle.

Images