A Data Analysis-Based Injection Molding Process Optimization Method and System
By deploying piezoelectric pressure sensors and thermocouple arrays on the surface of the mold cavity, the melt flow characteristics are calculated and a risk function is constructed. This solves the problem of scattered recording of pressure and temperature changes in the mold cavity in traditional injection molding processes, and realizes real-time optimization of injection molding process parameters and improved stability of product surface quality.
Patent Information
- Application Number
- CN202611067541.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-17
- Publication Date
- 2026-11-10
AI Technical Summary
In traditional injection molding process optimization methods, the correlation records of surface defects caused by changes in melt pressure and temperature in the mold cavity are scattered. Parameter corrections rely on preset process windows and operation inputs. It is difficult to form a continuous basis for the state transfer between the injection filling, holding pressure and cooling stages. The time when the detection record enters the adjustment stage is late. When batch switching or material state changes, the basis for process adjustment and the judgment of product surface quality are prone to lag.
By collecting transient pressure and temperature data of the mold cavity through piezoelectric pressure sensors and thermocouple arrays deployed in sensitive areas of the mold cavity surface, the melt flow characteristics such as holding pressure decay rate and cooling rate are calculated. A risk function is constructed and the process parameters are iteratively updated. Injection molding simulation is performed by combining material rheology and mold three-dimensional geometric model. The process parameters are iteratively corrected using the actual mold cavity pressure curve.
It enables real-time monitoring and analysis of pressure and temperature changes within the mold cavity, forming a closed-loop control system. This improves the accuracy of injection molding process parameters and the stability of product surface quality, while reducing the lag in process adjustments.
Smart Images

Figure CN123060891A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of injection molding control technology, and in particular to a method and system for optimizing injection molding processes based on data analysis. Background Technology
[0002] Injection molding control technology typically revolves around the acquisition, status judgment, command issuance, and recording and storage of parameters for the injection molding process, including injection molding machines, molds, temperature control equipment, hydraulic or servo drive mechanisms, part removal and inspection equipment, covering processes such as melt plasticizing, injection filling, holding pressure, cooling, and demolding. Pressure, temperature, displacement, speed, time, and product inspection records are transmitted between the equipment and management sides via sensors, controllers, industrial computers, or production management systems. Traditional injection molding process optimization methods and systems refer to the processing of data such as temperature, injection pressure, injection speed, holding pressure, cooling time, cycle time, and product dimensions, weight, and appearance defects during the injection molding process. Typically, the injection molding machine controller reads sensor and process formula records, compares the current batch parameters with preset process windows, historical batch records, or inspection results, and generates parameter correction values based on deviation ranges, alarm thresholds, quality judgment records, and operator input. These correction values are then written into the equipment control interface or formula storage area.
[0003] Traditional process adjustments often revolve around equipment settings, historical batch records, and inspection results. The correlation between changes in melt pressure and temperature within the mold cavity and surface defects is scattered. Parameter correction relies on preset process windows and operational inputs. The state transfer between injection molding, holding pressure, and cooling stages is difficult to establish a continuous basis. Inspection records are added to the adjustment process late. After parameters are written, there is a lack of closed-loop verification corresponding to the actual mold cavity pressure curve. When batches change or material states change, there is a tendency for a lag between the basis for process adjustments and the judgment of product surface quality. Summary of the Invention
[0004] The purpose of this invention is to overcome the shortcomings of the existing technology, and to this end, a data analysis-based injection molding process optimization method and system are proposed.
[0005] To achieve the above objectives, the present invention adopts the following technical solution: a data analysis-based injection molding process optimization method, comprising the following steps: S1: The transient pressure and temperature of the mold cavity are collected by piezoelectric pressure sensors and thermocouple arrays deployed in the sensitive area of the mold cavity surface, and the sampling time is aligned accordingly; S2: Based on the transient pressure and temperature of the mold cavity after the timing alignment, calculate the ratio of the transient pressure difference of the mold cavity at the start and end of the holding pressure stage to the corresponding time interval to obtain the holding pressure decay rate. Calculate the pressure rise rate of the transient pressure of the mold cavity at adjacent sampling times and the cooling rate of the transient temperature of the mold cavity at adjacent sampling times. Summarize the pressure rise rate, holding pressure decay rate, and cooling rate to form a melt flow feature set. S3: Calculate the deviation between each feature quantity in the melt flow feature set and the corresponding feature quantity threshold, and determine the corresponding surface defect type based on the feature quantity that shows the deviation according to the preset correspondence between feature quantity and surface defect type, so as to obtain surface defect information. S4: Using holding pressure, holding time, injection speed curve and mold surface temperature as independent variables, the surface defect information is called to construct a risk function. The gradient is calculated by perturbation and iteratively updated until convergence to obtain the process adjustment parameters. S5: Construct an injection molding simulation model based on material rheological parameters, thermodynamic parameters, and mold three-dimensional geometric model. Input the process adjustment parameters into the model to perform injection cycle simulation of filling, holding pressure, and cooling. Output the simulation pressure curve, melt front shear rate, and volume shrinkage rate. When the melt front shear rate and volume shrinkage rate are both within the preset qualified range, the process adjustment parameters are determined as the target process parameters. S6: Control the operation of the injection molding machine based on the target process parameters and collect the actual mold cavity pressure curve. Call the simulated pressure curve to calculate the root mean square error. When the root mean square error exceeds the preset deviation range, iteratively correct the target process parameters to obtain the corrected process parameters.
[0006] As a further aspect of the present invention, the process of S1 is specifically as follows: S101: Detect sensitive areas on the surface of the mold cavity, deploy piezoelectric pressure sensors in the sensitive areas on the surface of the mold cavity, collect the pressure signals output by the piezoelectric pressure sensors, read the values of the pressure signals at the sampling time to obtain the transient pressure of the mold cavity, deploy thermocouple arrays in the sensitive areas on the surface of the mold cavity, collect the temperature signals output by the thermocouple arrays, read the values of the temperature signals at the sampling time to obtain the transient temperature of the mold cavity; S102: Call the transient pressure and transient temperature of the mold cavity, extract the sampling time of each of the transient pressure and transient temperature of the mold cavity, and align the transient pressure and transient temperature of the mold cavity point by point according to the sampling time to obtain the time-aligned transient pressure and transient temperature of the mold cavity.
[0007] As a further aspect of the present invention, the process of S2 is specifically as follows: S201: Call the transient pressure of the mold cavity in the time-aligned transient pressure and transient temperature of the mold cavity, subtract the transient pressure of the mold cavity at adjacent sampling times to obtain the pressure difference, calculate the ratio of the pressure difference to the sampling time interval to obtain the pressure rise rate, subtract the transient pressure of the mold cavity at the start and end times of the pressure holding stage to obtain the pressure difference, calculate the ratio of the pressure difference to the corresponding time interval to obtain the pressure holding attenuation rate; S202: Call the transient temperature of the mold cavity from the time-aligned transient pressure and transient temperature of the mold cavity, subtract the transient temperatures of the mold cavity at adjacent sampling times to obtain the temperature difference, calculate the ratio of the temperature difference to the sampling time interval to obtain the cooling rate, and summarize the pressure rise rate, the holding pressure decay rate and the cooling rate to form a melt flow feature set.
[0008] As a further aspect of the present invention, the process of S3 is specifically as follows: S301: Obtain material properties and mold structure, and set a value range for each feature quantity in the melt flow feature set based on the material properties and mold structure to obtain the feature quantity threshold. S302: Compare each feature quantity in the melt flow feature set with the corresponding value range in the feature quantity threshold, calculate the deviation of each feature quantity from the corresponding value range, and obtain the feature quantity deviation set; S303: Obtain the preset correspondence between feature quantities and surface defect types, and obtain surface defect information for the surface defect types in the feature quantity mapping correspondence where deviations occur in the feature quantity deviation set.
[0009] As a further aspect of the present invention, the process of S4 is specifically as follows: S401: Using holding pressure, holding time, injection speed curve and mold surface temperature as independent variables, and the defect risk corresponding to the surface defect information as the dependent variable, establish a mapping relationship between the independent variables and the defect risk to obtain the risk function; S402: Invoke the risk function, apply a preset small-range perturbation to each independent variable, calculate the difference between the risk function value after applying the perturbation and the risk function value before applying the perturbation, obtain the change in the risk function, calculate the ratio of the change to the preset small-range perturbation to approximate the partial derivative of each independent variable, summarize to form the gradient, call the preset learning rate, subtract the product of the learning rate and the corresponding gradient for the current value of each independent variable to obtain the updated value, repeat the calculation of gradient and updated value until the risk function converges, and determine the updated value of each independent variable when the risk function converges as the process adjustment parameter.
[0010] As a further aspect of the present invention, the process of repeatedly calculating the gradient and updating the value until the risk function converges specifically comprises: Obtain the upper and lower limits of each pre-configured independent variable, and perform a comparison operation between each updated value and the corresponding upper and lower limits. When the updated value exceeds the upper limit, the updated value is set to the upper limit; when the updated value is lower than the lower limit, the updated value is set to the lower limit, thus obtaining the constrained updated value. The gradient of the risk function is recalculated based on the updated value after constraints. The gradient and the updated value after constraints are repeatedly calculated until the risk function converges. The updated value after constraints at the time of convergence is determined as the process adjustment parameter.
[0011] As a further aspect of the present invention, the process of S5 is specifically as follows: S501: Obtain material rheological parameters, thermodynamic parameters, and mold three-dimensional geometric model; establish a calculation model of injection molding cycle by associating material rheological parameters, thermodynamic parameters, and mold three-dimensional geometric model to obtain injection molding simulation model; S502: Input the process adjustment parameters into the injection molding simulation model, perform injection cycle simulation of filling, holding pressure and cooling, and output the simulation pressure curve, melt front shear rate and volume shrinkage rate; S503: Compare the melt front shear rate and the volume shrinkage rate with the preset acceptable range. If both the melt front shear rate and the volume shrinkage rate are within the preset acceptable range, determine the process adjustment parameters as the target process parameters.
[0012] As a further aspect of the present invention, the process of S6 is specifically as follows: S601: Control the operation of the injection molding machine according to the target process parameters, monitor the pressure change in the mold cavity, collect the actual mold cavity pressure during the operation of the injection molding machine at the sampling time, and obtain the actual mold cavity pressure curve; S602: Align the actual cavity pressure curve with the simulated pressure curve point by point at the same sampling time. Subtract the actual cavity pressure from the corresponding pressure of the simulated pressure curve at each sampling time to obtain the pressure difference. Take the square root of the average of all pressure differences to obtain the root mean square error. S603: Compare the root mean square error with a preset deviation range. When the root mean square error exceeds the preset deviation range, iteratively correct the target process parameters to obtain corrected process parameters.
[0013] As a further aspect of the present invention, the process of iteratively correcting the target process parameters to obtain the corrected process parameters specifically involves: The surface image of the product molded by the injection molding machine is acquired. For the surface image, the pixel area of the surface defect region and the total pixel area of the product surface are identified respectively. The ratio of the pixel area of the surface defect region to the total pixel area of the product surface is calculated to obtain the surface defect rate. The root mean square error is compared with a preset deviation range, and the surface defect rate is compared with a preset defect rate threshold. When the root mean square error exceeds the preset deviation range or the surface defect rate exceeds the preset defect rate threshold, the target process parameters are iteratively corrected to obtain the corrected process parameters.
[0014] A data analysis-based injection molding process optimization system, wherein the data analysis-based injection molding process optimization system is used to execute the above-described data analysis-based injection molding process optimization method, the system comprising: Multi-source sensing module: Collects transient pressure and temperature of the mold cavity through piezoelectric pressure sensors and thermocouple arrays deployed in sensitive areas on the surface of the mold cavity, and performs time-series alignment according to the sampling time; Feature extraction module: Based on the time-aligned transient pressure and temperature of the mold cavity, calculate the ratio of the transient pressure difference at the start and end of the holding pressure stage to the corresponding time interval to obtain the holding pressure decay rate, calculate the pressure rise rate of the transient pressure of the mold cavity at adjacent sampling times, calculate the cooling rate of the transient temperature of the mold cavity at adjacent sampling times, and summarize the pressure rise rate, holding pressure decay rate and cooling rate to form a melt flow feature set; Defect identification module: Calculates the deviation between each feature quantity in the melt flow feature set and the corresponding feature quantity threshold, and determines the corresponding surface defect type based on the feature quantity that shows deviation according to the preset correspondence between feature quantity and surface defect type, thereby obtaining surface defect information; The optimization solution module uses holding pressure, holding time, injection speed curve and mold surface temperature as independent variables, calls the surface defect information to construct a risk function, calculates the gradient through perturbation and iteratively updates it until convergence, and obtains the process adjustment parameters. Simulation verification module: Based on material rheological parameters, thermodynamic parameters and mold three-dimensional geometric model, an injection molding simulation model is constructed. The process adjustment parameters are input into the model to perform injection cycle simulation of filling, holding pressure and cooling. The simulated pressure curve, melt front shear rate and volume shrinkage rate are output. When the melt front shear rate and volume shrinkage rate are both within the preset qualified range, the process adjustment parameters are determined as the target process parameters. Online correction module: Based on the target process parameters, the injection molding machine is controlled to operate and the actual mold cavity pressure curve is collected. The root mean square error is calculated by calling the simulated pressure curve. When the root mean square error exceeds the preset deviation range, the target process parameters are iteratively corrected to obtain the corrected process parameters.
[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, transient pressure and transient temperature of the mold cavity are obtained in the sensitive area on the surface of the mold cavity by using a piezoelectric pressure sensor and a thermocouple array, and are aligned according to the sampling time to make the pressure rise rate, holding pressure decay rate and cooling rate consistent. Each feature quantity is compared with its corresponding threshold item by item to form surface defect information, so that the adjustment of the independent variable of the risk function has a clear correspondence with the defect type; The process adjustment parameters are first output by the injection molding simulation model as simulated pressure curves, melt front shear rate and volume shrinkage rate, and then iterative correction is triggered by the root mean square error of the actual mold cavity pressure curve, so that parameter generation, simulation verification and equipment operation feedback form a closed control basis. Attached Figure Description
[0016] Figure 1 This is a schematic diagram of the injection molding process optimization method based on data analysis according to the present invention; Figure 2 This is a schematic diagram of the data analysis-based injection molding process optimization workflow of the present invention; Figure 3 This is a schematic diagram illustrating the gradient calculation principle of the risk function perturbation in this invention; Figure 4 This is a schematic diagram illustrating the principle of the actual mold cavity pressure curve and the simulated pressure curve of the present invention; Figure 5 This is a schematic diagram of the data analysis-based injection molding process optimization system of the present invention. Detailed Implementation
[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.
[0018] This embodiment provides a data analysis-based injection molding process optimization method. The method revolves around a closed-loop data flow between mold cavity pressure, temperature, defect risk, molding simulation, and online correction. The sensitive area on the mold cavity surface refers to the location where changes in melt flow, pressure transmission, and cooling heat transfer are concentrated within the mold cavity. This location is jointly determined by the mold cavity structure, melt flow path, and areas prone to product surface defects, and is used to collect pressure and temperature data. The transient pressure in the mold cavity is time-series data converted from the pressure signal output by the piezoelectric pressure sensor at the sampling moment. The transient temperature in the mold cavity is time-series data converted from the temperature signal output by the thermocouple array at the sampling moment. Both types of data use the sampling moment as a common index for subsequent processing.
[0019] Please see Figure 1 and Figure 2 S1: Transient pressure and temperature in the mold cavity are collected using piezoelectric pressure sensors and thermocouple arrays deployed on the sensitive areas of the mold cavity surface, and time-aligned according to the sampling time. The technical purpose of this step is to provide a comparable time reference for pressure and temperature changes within the same molding cycle. This is because filling, holding pressure, and cooling during injection molding occur continuously. The pressure signal reflects the state of the melt's action on the cavity wall, and the temperature signal reflects the heat exchange state between the melt and the mold. Only when they correspond at the same sampling time can subsequent rate, deviation, and risk assessments have the same physical time meaning. The original signals are first retained at their respective sampling times, and then the correspondence between pressure points and temperature points is established according to the sampling times to form the time-aligned transient pressure and temperature of the mold cavity. This result serves as the input for extracting melt flow characteristics.
[0020] S101: Detects sensitive areas on the mold cavity surface. Piezoelectric pressure sensors are deployed in these areas to collect pressure signals. The pressure signal values are read at the sampling times to obtain the transient pressure of the mold cavity. A thermocouple array is also deployed in the sensitive areas to collect temperature signals. The temperature signal values are read at the sampling times to obtain the transient temperature of the mold cavity. The detection of sensitive areas confirms that the pressure and temperature acquisition locations correspond to changes in the melt state within the mold cavity. The pressure signal is output by the piezoelectric pressure sensor as the pressure state within the cavity changes, and the temperature signal is output by the thermocouple array as the heat transfer state on the cavity surface changes. The reading process follows the sampling time sequence, organizing the pressure and temperature signals within consecutive molding cycles into data sequences with time-series indexes. This allows subsequent processing to trace the specific sampling time within the same molding cycle for each pressure and temperature value. If pressure or temperature signals are missing during acquisition, the missing location is marked and the alignment and feature calculation are paused to prevent the missing signal from being mistakenly identified as a valid formed state. If the sampling time formats are inconsistent, the sampling time is converted into a unified time index expression before reading continues to ensure that the pressure sequence and temperature sequence can enter the same alignment process.
[0021] S102: The transient pressure and temperature of the mold cavity are retrieved. The sampling times for each transient pressure and temperature are extracted, and the transient pressure and temperature are aligned point-by-point according to their sampling times to obtain the time-aligned transient pressure and temperature. The point-by-point alignment logic uses the sampling time as a common index, placing the pressure and temperature values at the same sampling time into the same data record, ensuring that each record simultaneously contains the pressure state, temperature state, and corresponding sampling time. If there are sampling times in the pressure sequence that cannot be matched with the temperature sequence, or vice versa, the unmatched records are first marked as unusable for paired calculations, while the matched records are retained for subsequent feature extraction. The time-aligned result retains the original order of pressure and temperature changes and establishes a synchronous relationship between pressure and temperature, thus allowing it to be directly called by S2 for calculating the pressure rise rate, holding pressure decay rate, and cooling rate.
[0022] Please see Figure 1 and Figure 2 S2: Based on the time-aligned transient pressure and temperature of the mold cavity, calculate the ratio of the transient pressure difference at the start and end of the holding pressure stage to the corresponding time interval to obtain the holding pressure decay rate. Calculate the pressure rise rate of the transient pressure at adjacent sampling times, and calculate the cooling rate of the transient temperature at adjacent sampling times. Summarize the pressure rise rate, holding pressure decay rate, and cooling rate to form a melt flow feature set. The technical purpose of this step is to convert continuous pressure and temperature time-series data into feature quantities that can describe the melt flow and cooling state. The pressure rise rate is used to describe the rate of increase in cavity pressure between adjacent sampling times, the holding pressure decay rate is used to describe the trend of pressure maintenance and release during the holding pressure stage, and the cooling rate is used to describe the rate of decrease in cavity temperature between adjacent sampling times. The melt flow feature set is a data set formed by summing the pressure rise rate, holding pressure decay rate, and cooling rate. Each feature quantity retains the source sampling time or source stage, so that subsequent deviation judgments can be traced back to the specific pressure change, temperature change, or holding pressure change process.
[0023] S201: The transient pressure and temperature of the mold cavity after timing alignment are retrieved. The transient pressure at adjacent sampling times are subtracted to obtain the pressure difference. The ratio of this pressure difference to the sampling time interval is calculated to obtain the pressure rise rate. Similarly, the transient pressure at the start and end of the holding pressure phase is subtracted to obtain the pressure difference. The ratio of this pressure difference to the corresponding time interval is calculated to obtain the holding pressure decay rate. The pressure rise rate is calculated by first reading the pressure sequence in the order of sampling times, then comparing the pressure changes at adjacent sampling times, and correlating this pressure change with the time interval between the two sampling times to obtain the rate of pressure change over time. The holding pressure decay rate is calculated by first locating the start and end times of the holding pressure phase in the pressure sequence, then reading the pressure states at these two times, and correlating the pressure difference within the holding pressure phase with the corresponding time interval to obtain the degree of pressure decay during the holding pressure phase. If the start and end times of the holding pressure phase cannot be located in the pressure sequence, stop generating the holding pressure decay rate and mark the holding pressure phase data as unusable. If the order of adjacent sampling times is abnormal, rearrange them according to a unified sampling time order before performing interpolation. The pressure rise rate and holding pressure decay rate are then entered into the melt flow feature set, which, together with the cooling rate, provides the pressure-side information required for defect judgment.
[0024] S202: The transient temperature of the mold cavity is retrieved from the time-aligned transient pressure and temperature data. The transient temperatures at adjacent sampling times are subtracted to obtain the temperature difference. The ratio of this temperature difference to the sampling time interval is calculated to obtain the cooling rate. The pressure rise rate, holding pressure decay rate, and cooling rate are then combined to form the melt flow feature set. The cooling rate is formed based on the sampling time sequence of the temperature sequence. Temperature changes at adjacent sampling times are compared, and the temperature changes are correlated with the time interval between adjacent sampling times to determine how quickly the mold cavity temperature decreases as the molding process progresses. The cooling rate differs from the pressure rise rate and holding pressure decay rate; the former comes from the temperature signal, while the latter two come from the pressure signal. However, all three are based on time-aligned data, thus describing the linkage between melt filling, holding pressure transfer, and cooling solidification within the same molding cycle. If there are unreadable sampling times in the temperature sequence, the corresponding positions are excluded from adjacent difference processing, and the available temperature records are used to form the cooling rate. If the available temperature records are insufficient to form adjacent relationships, the generation of the cooling rate is stopped, and the melt flow feature set is marked as incomplete. The resulting melt flow feature set is received by S3 and used to determine the deviation of each feature from the feature threshold.
[0025] Please see Figure 1 and Figure 2S3: Calculate the deviation between each feature quantity in the melt flow feature set and its corresponding feature quantity threshold. Based on the preset correspondence between feature quantities and surface defect types, determine the corresponding surface defect type according to the feature quantity with deviation, thus obtaining surface defect information. The technical purpose of this step is to convert the pressure, holding pressure, and cooling changes in the melt flow feature set into judgment information related to product surface defects. The feature quantity threshold is a pre-formed judgment condition for each feature quantity, derived from the constraints of material properties and mold structure on melt flow, pressure holding, and cooling state. The surface defect information is a record of defect types obtained by mapping the feature quantities with deviations, carrying the feature quantity with deviation, the corresponding surface defect type, and the calling relationship between the defect type and the subsequent risk function. Through the mapping of each deviation and correspondence, S4 can directly construct a risk function with defect risk as the dependent variable, instead of reinterpreting the original sensor data in the optimization solution stage.
[0026] S301: Obtain material properties and mold structure, and set value ranges for each characteristic quantity in the melt flow characteristic set based on the material properties and mold structure to obtain characteristic quantity thresholds. Material properties are used to describe the flow response of the melt under pressure and temperature changes, and mold structure is used to describe the limitations of cavity shape, flow path, and heat transfer boundary on pressure and temperature changes. The process of setting value ranges is based on the physical meaning of each characteristic quantity, and acceptable judgment boundaries are established for pressure rise rate, holding pressure decay rate, and cooling rate, so that each characteristic quantity has a corresponding characteristic quantity threshold. In this method, the characteristic quantity threshold exists in the form of preset judgment conditions and does not need to be given specific values; its function is to provide a reference object for the same characteristic quantity in subsequent comparison calculations. If the material properties or mold structure cannot be obtained, the characteristic quantity threshold setting is stopped and the threshold source is marked as incomplete to avoid using judgment conditions without a source in subsequent deviation calculations.
[0027] S302: For each feature quantity in the melt flow feature set, a comparison operation is performed with the corresponding value range in the feature quantity threshold. The deviation of each feature quantity from the corresponding value range is calculated to obtain the feature quantity deviation set. The comparison operation matches each feature quantity by name. The pressure rise rate is compared only with the value range corresponding to the pressure rise rate, the holding pressure decay rate is compared only with the value range corresponding to the holding pressure decay rate, and the cooling rate is compared only with the value range corresponding to the cooling rate. The deviation represents the deviation state of the feature quantity from the corresponding value range. When the feature quantity is within the corresponding value range, the item is recorded as having no deviation; when the feature quantity falls outside the corresponding value range, the item is recorded as having a deviation, and the deviation direction and source feature quantity are retained. The feature quantity deviation set is a collection formed by summarizing all deviation records. It includes both the feature quantities with deviations and the judgment results of those without deviations, so that S303 can perform defect type mapping only for the feature quantities with deviations. If a feature quantity lacks a corresponding value range, it is first marked as unable to be judged and is not used as the basis for deviation.
[0028] S303: Obtain the preset correspondence between feature quantities and surface defect types. For feature quantities with deviations in the feature quantity deviation set, map the surface defect types in the correspondence to obtain surface defect information. The preset correspondence between feature quantities and surface defect types is a rule that establishes an association between different feature quantity deviation states and identifiable product surface defect types. When invoked, the name of the feature quantity with deviation and the deviation state are used as the query basis to find the surface defect type from the correspondence. If the pressure rise rate deviation, holding pressure decay rate deviation, or cooling rate deviation points to different surface defect types, the surface defect information retains their respective mapping results. If multiple deviations point to the same surface defect type, the surface defect information associates the same defect type with multiple source deviations, so that the subsequent risk function can simultaneously receive the impact of multiple feature quantities on defect risk. If the feature quantity with deviation cannot match the surface defect type in the correspondence, the deviation is first marked as an unmatched defect source and is not used as the basis for defect risk construction. The unmatched state is reserved for subsequent manual rule maintenance or process record review.
[0029] Please see Figure 1 , Figure 2 and Figure 3S4: Using holding pressure, holding time, injection speed curve, and mold surface temperature as independent variables, a risk function is constructed by invoking surface defect information. The gradient is calculated through perturbation and iteratively updated until convergence to obtain the process adjustment parameters. The technical purpose of this step is to transform surface defect information into a solution process for adjustable process parameters. Holding pressure, holding time, injection speed curve, and mold surface temperature are process variables that can affect pressure transmission within the cavity, melt filling speed, and cooling state. Defect risk is the evaluation object of surface defect information in the optimization solution, used to illustrate the risk correlation between the current combination of process variables and the surface defect type. The risk function is the mapping relationship between independent variables and defect risk; the input is the current combination of process variables, and the output is the risk state corresponding to the surface defect information. The perturbation gradient calculation does not rely on analytical formulas but instead applies a preset small-range perturbation amount around each independent variable, compares the changes in the risk function value before and after the perturbation, and determines the sensitive direction of risk to changes in that independent variable. The iterative update process adjusts the independent variables along the direction of risk reduction until the risk function reaches a convergent state, and then the values of the independent variables at the convergence point are determined as process adjustment parameters.
[0030] S401: Establish a mapping relationship between the independent variables and the defect risk using holding pressure, holding time, injection speed curve, and mold surface temperature as independent variables, and the defect risk corresponding to surface defect information as the dependent variable, to obtain the risk function. In this step, the independent variables exist as adjustable items of the current injection molding process. Holding pressure affects cavity pressure compensation during the holding stage, holding time affects the continuous state of pressure holding, injection speed curve affects the melt front advancement process, and mold surface temperature affects the cooling and solidification process. Surface defect information provides the source of the dependent variable for the risk function; the types of surface defects and their corresponding deviations jointly determine the evaluation direction of the defect risk. When establishing the mapping relationship, each set of independent variable values is mapped to the defect risk pointed to by the surface defect information, so that the risk function value can be obtained again after each change of the independent variables. If the surface defect information is empty, the risk function does not use defect deviation as the basis for updating and maintains the current process variable combination for simulation verification; if the surface defect information contains unmatched defect sources, the risk function only calls the matched defect types to avoid unmatched deviations entering the optimization solution and causing erroneous updates.
[0031] S402: Invoke the risk function, apply a preset small-range perturbation to each independent variable, calculate the difference between the risk function value after the perturbation and the risk function value before the perturbation, obtain the change in the risk function, calculate the ratio of the change to the preset small-range perturbation to approximate the partial derivative of each independent variable, summarize to form the gradient, call the preset learning rate, subtract the product of the learning rate and the corresponding gradient for the current value of each independent variable to obtain the updated value, repeat the calculation of gradient and updated value until the risk function converges, and determine the updated value of each independent variable when the risk function converges as the process adjustment parameter. The perturbation is performed item by item for each independent variable. When a perturbation is applied to a certain independent variable, the other independent variables remain unchanged, so that the change in the risk function can correspond to that independent variable. The partial derivative is used to represent the direction and strength of the response of the defect risk to the change of a single independent variable. The partial derivatives of multiple independent variables are summarized to form the gradient, which indicates the most direct adjustment direction of the risk change under the current combination of independent variables. The learning rate is a pre-configured update magnitude control condition used to convert gradients into updates for each independent variable. Subtraction shifts the independent variables in the direction that decreases the risk function, and then the risk function is called again with the updated value, repeating the perturbation, gradient formation, and update process. Convergence indicates that after continuous updates, the risk function no longer produces changes that satisfy the update conditions. The update value at convergence has an adjustment result that matches the defect risk and is therefore determined as the process adjustment parameter. If the risk function value cannot be returned, the current independent variable is retained and the iteration is marked as failed; if the independent variable update exceeds the available boundary during iteration, value constraints are applied before continuing the calculation.
[0032] The system obtains the pre-configured upper and lower limits of each independent variable, and compares each updated value with its corresponding upper and lower limits. The upper and lower limits represent the boundary conditions acceptable to the injection molding process for each independent variable, derived from its adjustable range configuration within the current process. The comparison is performed item by item to ensure that the holding pressure, holding time, injection speed curve, and mold surface temperature do not deviate from their respective boundaries. This process aligns the optimization direction of the risk function with the actual executable process adjustment range, preventing mathematical updates from entering states unacceptable to subsequent simulations or injection molding machine operation.
[0033] When the updated value exceeds the upper limit, it is set to the upper limit; when the updated value falls below the lower limit, it is set to the lower limit, resulting in a constrained updated value. The constrained updated value retains the direction of gradient update for defect risk adjustment while pulling out-of-bounds results back within pre-configured usable boundaries. If the updated value lies between the upper and lower limits, it is directly used as the constrained updated value. The constrained result is then re-inputted into the risk function calculation, ensuring that subsequent iterations always remain within the executable process range.
[0034] The gradient of the risk function is recalculated based on the updated values after constraints. This process of calculating the gradient and updating values after constraints is repeated until the risk function converges. The updated values after constraints at the point of convergence are then used as the process adjustment parameters. The purpose of recalculating the gradient is to reflect the true direction of risk change of the constrained combination of independent variables, because after boundary constraints, the input received by the risk function is different from the original updated values. During repeated iterations, the updated values after constraints are confirmed in each round before the perturbation gradient calculation and update are re-executed, ensuring that the process adjustment parameters are guided by the direction of defect risk reduction and satisfy the boundary values of the independent variables. The process adjustment parameters formed at convergence are received by S5 and used for verification of the filling, holding pressure, and cooling cycles of the injection molding simulation model.
[0035] Please see Figure 1 and Figure 2 S5: Based on material rheological parameters, thermodynamic parameters, and the 3D geometric model of the mold, an injection molding simulation model is constructed. Process adjustment parameters are input into the model to simulate the injection cycle of filling, holding pressure, and cooling. The simulated pressure curve, melt front shear rate, and volumetric shrinkage rate are output. When both the melt front shear rate and volumetric shrinkage rate are within the preset acceptable range, the process adjustment parameters are determined as the target process parameters. The technical purpose of this step is to verify the process adjustment parameters obtained through the risk function within the molding physical process. Material rheological parameters describe the melt response under flow and shear conditions; thermodynamic parameters describe the heat exchange and cooling state between the melt and the mold; and the 3D geometric model of the mold provides the cavity space, flow path, and boundary morphology. After associating the above objects, the injection molding simulation model can deduce pressure, shear, and shrinkage changes according to the injection cycle sequence of filling, holding pressure, and cooling. The simulated pressure curve is used for subsequent comparison with the actual mold cavity pressure curve, and the melt front shear rate and volumetric shrinkage rate are used to determine whether the process adjustment parameters meet the preset acceptable range in the molding simulation.
[0036] S501: Acquire material rheological parameters, thermodynamic parameters, and the mold's 3D geometric model. Establish a computational model for the injection molding cycle by associating these parameters, resulting in an injection molding simulation model. Material rheological parameters are derived from the flow performance description of the current molding material; thermodynamic parameters are derived from the heat exchange performance description of the current molding material and the mold; and the mold's 3D geometric model is derived from the spatial description of the product cavity, gating path, and mold structure. The associative process unifies the material's flow response under pressure and shear, the heat exchange response between the material and the mold, and the spatial flow path of the melt within the cavity into the computational model of the injection molding cycle. This computational model organizes inputs and outputs according to the process sequence of filling, holding pressure, and cooling, allowing process adjustment parameters to be used as boundary and control conditions in the simulation process. If material rheological parameters, thermodynamic parameters, or the mold's 3D geometric model are missing, stop building the injection molding simulation model and mark the source of the missing parameters to prevent subsequent simulation outputs from losing their physical source.
[0037] S502: Input the process adjustment parameters into the injection molding simulation model, execute the injection cycle simulation of filling, holding pressure, and cooling, and output the simulated pressure curve, melt front shear rate, and volume shrinkage rate. The process adjustment parameters serve as the process input for the simulation, affecting melt propagation during the filling stage, pressure transmission during the holding pressure stage, and volume change during the cooling stage. The filling simulation first deduces the melt front propagation process based on the injection speed curve and the three-dimensional geometry model of the mold. The holding pressure simulation then deduces the cavity pressure maintenance and transmission process based on the holding pressure and holding time. The cooling simulation then deduces the temperature drop and shrinkage process based on the mold surface temperature and thermodynamic parameters. The simulated pressure curve records the cavity pressure change with the sampling time during the molding cycle, the melt front shear rate records the shear change of the melt front during the flow process, and the volume shrinkage rate records the volume change caused by cooling and solidification. All of the above outputs are fed into S503, where the simulated pressure curve is also called by S6 to calculate the root mean square error with the actual cavity pressure curve.
[0038] S503: A comparison calculation is performed between the melt front shear rate and volumetric shrinkage rate and a preset acceptable range. If both the melt front shear rate and volumetric shrinkage rate are within the preset acceptable range, the process adjustment parameter is determined as the target process parameter. The preset acceptable range is a pre-configured simulation judgment condition for the melt front shear rate and volumetric shrinkage rate, used to determine whether the simulation output matches the molding quality requirements. The comparison calculation is performed separately for the melt front shear rate and volumetric shrinkage rate. Only when both simulation outputs are within their respective preset acceptable ranges are the process adjustment parameter determined as the target process parameter. If either simulation output is not within the preset acceptable range, the process adjustment parameter does not enter the target process parameter confirmation process and returns to the process adjustment process. Based on the unmet melt front shear rate or volumetric shrinkage rate, the upper or lower limit of the corresponding independent variable is tightened. Under the updated value constraints, the perturbation gradient calculation and iterative update are re-executed to continue forming new verifiable parameters. After the target process parameter is formed, it is received by S6 and used to control the operation of the injection molding machine and collect the actual mold cavity pressure curve.
[0039] Please see Figure 1 , Figure 2 and Figure 4 S6: Based on the target process parameters, control the injection molding machine to operate and collect the actual cavity pressure curve. Calculate the root mean square error (RMSE) using the simulated pressure curve. When the RMS error exceeds the preset deviation range, iteratively correct the target process parameters to obtain the corrected process parameters. The technical purpose of this step is to put the simulated and verified target process parameters into the actual injection molding operation for closed-loop verification. The actual cavity pressure curve is the cavity pressure time series collected by the injection molding machine running according to the target process parameters. The simulated pressure curve is the simulation result output from S5. The RMS error describes the overall deviation between the actual cavity pressure curve and the simulated pressure curve. The preset deviation range is a pre-configured condition used to determine whether the simulated pressure and the actual pressure are consistent. When the RMS error exceeds the preset deviation range, it indicates that the pressure response of the target process parameters in actual operation deviates from the simulation result, requiring iterative correction of the target process parameters. The corrected process parameters, as the output after closed-loop adjustment, are used for subsequent injection molding operations or further verification.
[0040] S601: Control the injection molding machine operation according to the target process parameters, monitor the pressure changes in the mold cavity, and collect the actual mold cavity pressure during the injection molding machine operation at the sampling time to obtain the actual mold cavity pressure curve. In this step, the target process parameters are converted into control inputs for the injection molding machine operation, enabling the injection molding machine to execute the molding process according to the holding pressure, holding time, injection speed curve, and mold surface temperature corresponding to the target process parameters. The actual mold cavity pressure acquisition still uses the sampling time as an index, recording the state of pressure change in the cavity as the injection molding machine operates, forming the actual mold cavity pressure curve. If the injection molding machine operation command transmission fails, the pressure curve acquisition is stopped first, and the target process parameters are retained; if the pressure acquisition interface is abnormal, the actual mold cavity pressure curve is marked as unusable and not included in the root mean square error calculation. After the actual mold cavity pressure curve is formed, it is received by S602 and used for point-by-point alignment with the simulated pressure curve.
[0041] S602: Align the actual mold cavity pressure curve with the simulated pressure curve point by point according to the same sampling time. For each sampling time, subtract the corresponding pressure of the actual mold cavity pressure from the corresponding pressure of the simulated pressure curve to obtain the pressure difference. Take the square root of the square of all pressure differences to obtain the root mean square error. Point-by-point alignment still uses the sampling time as a common index. First, put the pressure values in the actual mold cavity pressure curve and the pressure values in the simulated pressure curve at the same sampling time into the same comparison record, and then generate a pressure difference for each comparison record. The square, average, and square root processing are used to combine the pressure differences of multiple sampling times into a whole deviation measure, so that both positive and negative deviations can contribute to the degree of deviation. After the root mean square error is formed, it retains a comprehensive description of all pressure differences and can be individually judged against the preset deviation range. If there are sampling times that cannot be matched between the actual mold cavity pressure curve and the simulated pressure curve, first remove the records that cannot be compared in pairs and mark the data alignment status; if there are not enough comparable records to form the root mean square error, stop the deviation judgment and return to the pressure curve acquisition stage to verify the data source.
[0042] S603: This function compares the root mean square error (RMSE) with a preset deviation range. When the RMS error exceeds the preset deviation range, iterative correction is performed on the target process parameters to obtain corrected process parameters. The preset deviation range is the criterion for allowing deviation between the simulated pressure curve and the actual mold cavity pressure curve, and it is directly compared with the RMS error during operation. When the RMS error is within the preset deviation range, the target process parameters remain consistent with the actual operating pressure response, and the target process parameters can be used as the current operating output. When the RMS error exceeds the preset deviation range, the iterative correction process adjusts the target process parameters according to the pressure deviation status and re-enters the operation for data acquisition or simulation verification, ensuring that the correction direction maintains a closed loop with the actual pressure response. If the correction process cannot complete parameter writing, the available parameters from the previous round are retained, and the correction failure status is marked to prevent invalid parameters from entering the operation.
[0043] The system acquires surface images of the product molded by the injection molding machine. For each surface image, it identifies the pixel area of the defective region and the total pixel area of the product surface. The ratio of the pixel area of the defective region to the total pixel area is calculated to obtain the surface defect rate. The system compares the root mean square error (RMS) with a preset deviation range and the surface defect rate with a preset defect rate threshold. If the RMS exceeds the preset deviation range or the surface defect rate exceeds the preset defect rate threshold, iterative corrections are made to the target process parameters to obtain corrected process parameters. The surface images are acquired from the product surface after molding by the injection molding machine. The pixel area of the defective region is the area record obtained after identifying the defective region in the image. The total pixel area of the product surface is the overall area record corresponding to the product surface area in the same surface image. The surface defect rate correlates the defective region area with the overall product surface area to supplement the judgment of the product surface state beyond pressure curve deviation. The preset defect rate threshold is a pre-configured judgment condition for the surface defect rate, which is compared with the surface defect rate when called. The root mean square error (RMSE) and surface defect rate constitute a dual correction trigger condition. Whenever the actual pressure response deviates from the simulated pressure, or the product surface defect rate exceeds the preset defect rate threshold, the target process parameters continue to be iteratively corrected. If surface image acquisition fails, the correction trigger is still executed based on the RMSE; if the surface defect area cannot be identified, the image recognition result is first marked as unusable, and the surface defect rate is not used as the basis for correction triggering.
[0044] Please see Figure 5 A data analysis-based injection molding process optimization system is proposed, wherein the multi-source sensing module corresponds to the acquisition of transient pressure and temperature in the mold cavity in S1 and the time-series alignment according to the sampling time; the feature extraction module corresponds to the extraction of pressure rise rate, holding pressure decay rate and cooling rate and the formation of melt flow feature set in S2; the defect identification module corresponds to the formation of feature quantity deviation set and the acquisition of surface defect information in S3; the optimization solution module corresponds to the construction of risk function, gradient calculation of perturbation and iterative update to convergence in S4; the simulation verification module corresponds to the construction of injection molding simulation model, injection cycle simulation and determination of target process parameters in S5; and the online correction module corresponds to the acquisition of actual mold cavity pressure curve, calculation of root mean square error and correction of process parameters in S6.
[0045] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.
Claims
1. A data analysis-based injection molding process optimization method, characterized in that, Includes the following steps: S1: The transient pressure and temperature of the mold cavity are collected by piezoelectric pressure sensors and thermocouple arrays deployed in the sensitive area of the mold cavity surface, and the sampling time is aligned accordingly; S2: Based on the transient pressure and temperature of the mold cavity after the timing alignment, calculate the ratio of the transient pressure difference of the mold cavity at the start and end of the holding pressure stage to the corresponding time interval to obtain the holding pressure decay rate. Calculate the pressure rise rate of the transient pressure of the mold cavity at adjacent sampling times and the cooling rate of the transient temperature of the mold cavity at adjacent sampling times. Summarize the pressure rise rate, holding pressure decay rate, and cooling rate to form a melt flow feature set. S3: Calculate the deviation between each feature quantity in the melt flow feature set and the corresponding feature quantity threshold, and determine the corresponding surface defect type based on the feature quantity that shows the deviation according to the preset correspondence between feature quantity and surface defect type, so as to obtain surface defect information. S4: Using holding pressure, holding time, injection speed curve and mold surface temperature as independent variables, the surface defect information is called to construct a risk function. The gradient is calculated by perturbation and iteratively updated until convergence to obtain the process adjustment parameters. S5: Construct an injection molding simulation model based on material rheological parameters, thermodynamic parameters, and mold three-dimensional geometric model. Input the process adjustment parameters into the model to perform injection cycle simulation of filling, holding pressure, and cooling. Output the simulation pressure curve, melt front shear rate, and volume shrinkage rate. When the melt front shear rate and volume shrinkage rate are both within the preset qualified range, the process adjustment parameters are determined as the target process parameters. S6: Control the operation of the injection molding machine based on the target process parameters and collect the actual mold cavity pressure curve. Call the simulated pressure curve to calculate the root mean square error. When the root mean square error exceeds the preset deviation range, iteratively correct the target process parameters to obtain the corrected process parameters.
2. The injection molding process optimization method based on data analysis according to claim 1, characterized in that, The process of S1 is as follows: S101: Detect sensitive areas on the surface of the mold cavity, deploy piezoelectric pressure sensors in the sensitive areas on the surface of the mold cavity, collect the pressure signals output by the piezoelectric pressure sensors, read the values of the pressure signals at the sampling time to obtain the transient pressure of the mold cavity, deploy thermocouple arrays in the sensitive areas on the surface of the mold cavity, collect the temperature signals output by the thermocouple arrays, read the values of the temperature signals at the sampling time to obtain the transient temperature of the mold cavity; S102: Call the transient pressure and transient temperature of the mold cavity, extract the sampling time of each of the transient pressure and transient temperature of the mold cavity, and align the transient pressure and transient temperature of the mold cavity point by point according to the sampling time to obtain the time-aligned transient pressure and transient temperature of the mold cavity.
3. The injection molding process optimization method based on data analysis according to claim 1, characterized in that, The process of S2 is as follows: S201: Call the transient pressure of the mold cavity in the time-aligned transient pressure and transient temperature of the mold cavity, subtract the transient pressure of the mold cavity at adjacent sampling times to obtain the pressure difference, calculate the ratio of the pressure difference to the sampling time interval to obtain the pressure rise rate, subtract the transient pressure of the mold cavity at the start and end times of the pressure holding stage to obtain the pressure difference, calculate the ratio of the pressure difference to the corresponding time interval to obtain the pressure holding attenuation rate; S202: Call the transient temperature of the mold cavity from the time-aligned transient pressure and transient temperature of the mold cavity, subtract the transient temperatures of the mold cavity at adjacent sampling times to obtain the temperature difference, calculate the ratio of the temperature difference to the sampling time interval to obtain the cooling rate, and summarize the pressure rise rate, the holding pressure decay rate and the cooling rate to form a melt flow feature set.
4. The injection molding process optimization method based on data analysis according to claim 1, characterized in that, The process of S3 is as follows: S301: Obtain material properties and mold structure, and set a value range for each feature quantity in the melt flow feature set based on the material properties and mold structure to obtain the feature quantity threshold. S302: Compare each feature quantity in the melt flow feature set with the corresponding value range in the feature quantity threshold, calculate the deviation of each feature quantity from the corresponding value range, and obtain the feature quantity deviation set; S303: Obtain the preset correspondence between feature quantities and surface defect types, and obtain surface defect information for the surface defect types in the feature quantity mapping correspondence where deviations occur in the feature quantity deviation set.
5. The injection molding process optimization method based on data analysis according to claim 1, characterized in that, The process of S4 is as follows: S401: Using holding pressure, holding time, injection speed curve and mold surface temperature as independent variables, and the defect risk corresponding to the surface defect information as the dependent variable, establish a mapping relationship between the independent variables and the defect risk to obtain the risk function; S402: Invoke the risk function, apply a preset small-range perturbation to each independent variable, calculate the difference between the risk function value after applying the perturbation and the risk function value before applying the perturbation, obtain the change in the risk function, calculate the ratio of the change to the preset small-range perturbation to approximate the partial derivative of each independent variable, summarize to form the gradient, call the preset learning rate, subtract the product of the learning rate and the corresponding gradient for the current value of each independent variable to obtain the updated value, repeat the calculation of gradient and updated value until the risk function converges, and determine the updated value of each independent variable when the risk function converges as the process adjustment parameter.
6. The injection molding process optimization method based on data analysis according to claim 5, characterized in that, The process of repeatedly calculating the gradient and updating the value until the risk function converges is specifically as follows: Obtain the upper and lower limits of each pre-configured independent variable, and perform a comparison operation between each updated value and the corresponding upper and lower limits. When the updated value exceeds the upper limit, the updated value is set to the upper limit; when the updated value is lower than the lower limit, the updated value is set to the lower limit, thus obtaining the constrained updated value. The gradient of the risk function is recalculated based on the updated value after constraints. The gradient and the updated value after constraints are repeatedly calculated until the risk function converges. The updated value after constraints at the time of convergence is determined as the process adjustment parameter.
7. The injection molding process optimization method based on data analysis according to claim 1, characterized in that, The process of S5 is as follows: S501: Obtain material rheological parameters, thermodynamic parameters, and mold three-dimensional geometric model; establish a calculation model of injection molding cycle by associating material rheological parameters, thermodynamic parameters, and mold three-dimensional geometric model to obtain injection molding simulation model; S502: Input the process adjustment parameters into the injection molding simulation model, perform injection cycle simulation of filling, holding pressure and cooling, and output the simulation pressure curve, melt front shear rate and volume shrinkage rate; S503: Compare the melt front shear rate and the volume shrinkage rate with the preset acceptable range. If both the melt front shear rate and the volume shrinkage rate are within the preset acceptable range, determine the process adjustment parameters as the target process parameters.
8. The injection molding process optimization method based on data analysis according to claim 1, characterized in that, The process of S6 is as follows: S601: Control the operation of the injection molding machine according to the target process parameters, monitor the pressure change in the mold cavity, collect the actual mold cavity pressure during the operation of the injection molding machine at the sampling time, and obtain the actual mold cavity pressure curve; S602: Align the actual cavity pressure curve with the simulated pressure curve point by point at the same sampling time. Subtract the actual cavity pressure from the corresponding pressure of the simulated pressure curve at each sampling time to obtain the pressure difference. Take the square root of the average of all pressure differences to obtain the root mean square error. S603: Compare the root mean square error with a preset deviation range. When the root mean square error exceeds the preset deviation range, iteratively correct the target process parameters to obtain corrected process parameters.
9. The injection molding process optimization method based on data analysis according to claim 8, characterized in that, The process of iteratively correcting the target process parameters to obtain the corrected process parameters is specifically as follows: The surface image of the product molded by the injection molding machine is acquired. For the surface image, the pixel area of the surface defect region and the total pixel area of the product surface are identified respectively. The ratio of the pixel area of the surface defect region to the total pixel area of the product surface is calculated to obtain the surface defect rate. The root mean square error is compared with a preset deviation range, and the surface defect rate is compared with a preset defect rate threshold. When the root mean square error exceeds the preset deviation range or the surface defect rate exceeds the preset defect rate threshold, the target process parameters are iteratively corrected to obtain the corrected process parameters.
10. A data analysis-based injection molding process optimization system, characterized in that, The system is used to implement the injection molding process optimization method based on data analysis as described in any one of claims 1-9, and the system comprises: Multi-source sensing module: Collects transient pressure and temperature of the mold cavity through piezoelectric pressure sensors and thermocouple arrays deployed in sensitive areas on the surface of the mold cavity, and performs time-series alignment according to the sampling time; Feature extraction module: Based on the time-aligned transient pressure and temperature of the mold cavity, calculate the ratio of the transient pressure difference at the start and end of the holding pressure stage to the corresponding time interval to obtain the holding pressure decay rate, calculate the pressure rise rate of the transient pressure of the mold cavity at adjacent sampling times, calculate the cooling rate of the transient temperature of the mold cavity at adjacent sampling times, and summarize the pressure rise rate, holding pressure decay rate and cooling rate to form a melt flow feature set; Defect identification module: Calculates the deviation between each feature quantity in the melt flow feature set and the corresponding feature quantity threshold, and determines the corresponding surface defect type based on the feature quantity that shows deviation according to the preset correspondence between feature quantity and surface defect type, thereby obtaining surface defect information; The optimization solution module uses holding pressure, holding time, injection speed curve and mold surface temperature as independent variables, calls the surface defect information to construct a risk function, calculates the gradient through perturbation and iteratively updates it until convergence, and obtains the process adjustment parameters. Simulation verification module: Based on material rheological parameters, thermodynamic parameters and mold three-dimensional geometric model, an injection molding simulation model is constructed. The process adjustment parameters are input into the model to perform injection cycle simulation of filling, holding pressure and cooling. The simulation pressure curve, melt front shear rate and volume shrinkage rate are output. When the melt front shear rate and volume shrinkage rate are both within the preset qualified range, the process adjustment parameters are determined as the target process parameters. Online correction module: Based on the target process parameters, the injection molding machine is controlled to operate and the actual mold cavity pressure curve is collected. The root mean square error is calculated by calling the simulated pressure curve. When the root mean square error exceeds the preset deviation range, the target process parameters are iteratively corrected to obtain the corrected process parameters.
