Plastic mold injection molding system for injection molding part forming

Abstract

The plastic mold injection molding system comprises an analogue simulation module, an intelligent optimization algorithm module, a data acquisition module, a data synchronization mechanism, a feedback control module and a quality evaluation module, according to the system, simulation software is used for simulating the influence of different pressure maintaining parameters on warping, shrinkage, stress distribution and the like of a product, and online optimization and fine adjustment of the parameters in the pressure maintaining stage are achieved through reinforcement learning in combination with real-time collected data; a data synchronization mechanism ensures time consistency of simulation data and production data, and a feedback control module directly calls an injection molding machine interface based on an optimization result to realize continuous control of a multi-stage pressure maintaining curve; the quality evaluation module performs three-dimensional shape, size precision and defect detection on the molded part, and takes a result as a reward signal to update the optimization model; according to the invention, a'simulation-optimization-synchronization-feedback 'closed-loop structure is constructed, online simulation fine tuning is realized, and the molding quality and the production stability of the injection molded part are remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of injection molding systems, and specifically relates to a plastic mold injection system for molding injection molded parts. Background Technology

[0002] Injection molding is one of the most widely used processes in plastic product manufacturing. The process parameters during the holding pressure stage (such as holding pressure curve, holding time, mold cavity temperature distribution, and cooling rate) have a decisive impact on the dimensional accuracy, warpage, residual stress, and surface quality of the product. In existing technologies, the optimization of injection molding process parameters mainly relies on manual experience-based machine adjustment or offline simulation. Manual machine adjustment is inefficient and unstable, making it difficult to cope with batch variations in raw materials, changes in environmental temperature and humidity, and fluctuations in equipment performance. While offline simulation can predict the impact of different parameter combinations on product quality, the lack of real-time data linkage with the actual production process often makes it difficult to directly guide on-site production. Furthermore, traditional control systems mostly involve staged parameter adjustments, failing to achieve curve-level dynamic fine-tuning within a single holding pressure cycle, thus hindering further improvements in molding quality and consistency.

[0003] As shown in the existing injection molding system "as described in application number: CN202510158165.X, a control method and injection molding system for an injection molding system", the accuracy and response time of the pressure sensor in this patent directly affect the adjustment accuracy during the holding pressure stage. If the sensor accuracy is insufficient or the response is sluggish, it may lead to data delay or inaccuracy, thereby affecting the adjustment results. The injection molding process is highly dynamic, and real-time response is particularly important. The processing of neural network and sensor data has a certain delay, which can lead to unstable molding quality.

[0004] Therefore, there is an urgent need for an intelligent system that can synchronize simulation and actual data, optimize parameters in real time, and dynamically control the holding pressure curve during the injection molding process, so as to improve the stability of molding quality and the adaptability of production. Summary of the Invention

[0005] (a) Technical problems to be solved

[0006] The purpose of this invention is to overcome the problems of disconnect between offline simulation and actual production, lag in parameter adjustment, and lack of curve-level pressure holding control in the prior art. It provides a plastic mold injection system that can achieve real-time parameter fine-tuning, bidirectional linkage between simulation and actual production data, and quality feedback-driven optimization within a single production process, thereby significantly improving the molding quality of products, reducing the defect rate, and enhancing the stability and adaptability of production.

[0007] (II) Technical Solution

[0008] This invention is achieved through the following technical solution: This invention proposes a plastic mold injection system for injection molding parts, comprising:

[0009] The simulation module is based on the injection molding simulation software API and integrated with the control system. It can dynamically call the simulation program and automatically modify the simulation process parameters (including but not limited to the holding pressure curve, holding time, mold cavity temperature field, and cooling rate) to realize automatic simulation analysis under different parameter combinations and output quality prediction indicators such as warpage, shrinkage, stress distribution, and temperature field.

[0010] The intelligent optimization algorithm module communicates bidirectionally with the simulation module and control system. Based on simulation data, it constructs reinforcement learning models, deep neural networks, genetic algorithms, or fuzzy control models to train and obtain the mapping strategy of pressure holding parameters and quality indicators. During the production process, it combines real-time sensor data to perform state assessment and parameter fine-tuning, dynamically outputting the pressure curve and time adjustment scheme of the pressure holding stage, and realizing online closed-loop optimization.

[0011] The data acquisition module includes sensors for pressure, temperature, and displacement, as well as an image acquisition unit, used to collect multimodal process data in real time during the molding process.

[0012] The data synchronization mechanism includes high-precision clock synchronization, timestamp marking, buffer scheduling, multimodal alignment, and data cleaning to ensure that simulation data and sensor data are strictly matched on the time axis, and to remove outliers and interpolate to complete them.

[0013] The feedback control module directly calls the injection molding machine's open interface to modify the control program of the holding pressure section in real time during production operation, thereby achieving curve-level continuous holding pressure control.

[0014] The quality assessment module includes a three-dimensional optical scanning device, a structural stress analysis module, and an image recognition-based defect detection module. It is used to collect quality indicators such as warpage, dimensional tolerance, residual stress, and surface defects, and feeds them back to the optimization module and simulation database as reward signals for reinforcement learning, so as to realize the linkage update between simulation and optimization.

[0015] The system constructs a closed-loop control structure of "simulation-optimization-synchronization-feedback", which can complete the entire chain of operation from data acquisition, strategy generation, real-time control of the holding pressure curve to quality feedback optimization within a production cycle, and realize online simulation fine-tuning.

[0016] (III) Beneficial Effects

[0017] By dynamically calling simulation software APIs and automatically modifying parameters, simulation results are no longer static references but can be updated in real time throughout the production cycle and drive process optimization. Through open interfaces, the injection molding machine's holding pressure section can be directly controlled, enabling continuous fine-tuning of multi-level holding pressure curves, rather than just adjusting a single pressure value or switching stages, significantly improving the accuracy of molding quality control. High-precision clock synchronization, timestamps, and buffer scheduling ensure accurate time matching between sensor data and simulation data, providing reliable input for the algorithm and fundamentally improving the stability of optimization decisions. Attached Figure Description

[0018] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0019] Figure 1 This is a system block diagram of the present invention.

[0020] Figure 2 This is a flowchart illustrating the workflow of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Example 1

[0023] Please refer to the accompanying drawings and background information. The technical problem to be solved in this embodiment is that, in the prior art, existing injection molding machines often rely on manual experience to set the parameters of each stage in multi-stage holding pressure, which cannot achieve precise control of the pressure curve and duration of the holding pressure stage during the injection molding process. Therefore, an injection molding holding pressure control system based on a simulation-intelligent optimization coupling mechanism is provided. The injection molding holding pressure control system includes:

[0024] The simulation module is used to perform virtual simulation of the holding pressure stage using injection molding simulation software. It simulates the impact of different combinations of holding pressure, holding time and other parameters on the quality of the product, including but not limited to simulation results such as warpage, shrinkage, stress distribution and temperature field changes.

[0025] In the initialization phase of the system of this invention, in order to improve the prediction accuracy and strategy guidance capability of the intelligent optimization algorithm module during the initial operation, the system first needs to run the simulation module to construct a sufficient number of high-quality sample datasets of the pressure holding process under the condition of no real injection molding production data or manufacturing demand targets; the main steps of this phase are as follows:

[0026] The intelligent optimization algorithm module communicates bidirectionally with the simulation module, performs training or iterative optimization based on simulation results and actual production data, and outputs an adjustment scheme for the pressure holding parameters. The intelligent optimization algorithm includes reinforcement learning, genetic algorithm, fuzzy logic control, etc.

[0027] The simulation module utilizes injection molding simulation software (such as Moldflow, Sigmasoft, etc.) to generate a large number of sample datasets based on typical process parameters; each dataset includes:

[0028] Input parameters (holding pressure, holding time, mold temperature, etc.);

[0029] Output indicators (shrinkage rate, warpage, stress distribution, temperature field distribution, etc.);

[0030] The data is uploaded to the intelligent optimization algorithm module via an API interface or middleware database, constructing a dataset in the following format:

[0031]

[0032] in For the combination of input parameters, The results are from the simulation.

[0033] After receiving multi-dimensional sample data from the simulation module, the intelligent optimization algorithm module performs modeling based on reinforcement learning; its main process is as follows:

[0034] The optimization module uses the input parameters and corresponding simulation results of each round as a sample set to train a policy function:

[0035]

[0036] in Indicates the pressure holding parameter strategy. This is the reward function, used to measure molding quality targets (such as low shrinkage, high stress uniformity, etc.). The optimization module generates a new set of holding pressure strategy parameters. The strategy is then passed back to the simulation module via an interface or API. Upon receiving the strategy, the simulation module immediately performs a fast simulation calculation to obtain the new simulation results under that strategy. .

[0037] This simulation feedback serves as the next input to the optimization module, thus forming the following loop:

[0038]

[0039] This constructs an automated closed-loop process of "optimization-simulation-evaluation-re-optimization".

[0040] To adapt to environmental disturbances during production (such as changes in raw material batches and mold temperature fluctuations), the online simulation fine-tuning mechanism has an "online micro-simulation" function, the process of which is as follows:

[0041] After each product is formed, local simulation is triggered based on changes in real-time collected data; small batch simulation data is generated by limiting the range of variables (such as ±5% holding pressure);

[0042] If the deviation between the simulation feedback and the optimized strategy is too large, the system will activate the "rollback verification mechanism," reverting the strategy to the previous optimal state and adding a penalty term to re-correct the model.

[0043] The simulation module is equipped with a lightweight model cache or a fast prediction model (such as response surface modeling, surrogate model, etc.), which enables the estimated simulation output results to be obtained within 5 to 10 seconds after parameter input, thereby supporting real-time decision-making.

[0044] Furthermore, the system also incorporates an integration of genetic algorithms and fuzzy control mechanisms.

[0045] Genetic algorithm optimization path: In certain scenarios, the system can choose to perform global search optimization of simulation results through genetic algorithm; the initial pressure holding parameters are used as individuals in the population, and new parameter sets are generated through crossover and mutation operations. The fitness of these parameters is evaluated by the simulation module (e.g., minimization of deformation, uniform stress distribution, etc.), and finally the optimal individual is selected and sent back to the control system.

[0046] Fuzzy logic control integration path: To cope with complex nonlinear working conditions, the system can use a fuzzy controller to fuzzify the actual collected data; for example, linguistic rules such as "pressure rises too fast" and "mold cavity temperature is too high" are converted into a fuzzy rule base, and the holding time and pressure curve are dynamically corrected under the guidance of the inference engine, which has good robustness.

[0047] Based on the current sensor data acquisition status, the optimization module predicts the optimal combination of pressure holding parameters for the next cycle:

[0048] Multi-stage holding pressure curve:

[0049] Duration of each stage:

[0050] The control command is sent to the injection molding machine controller via OPC-UA or Euromap77 protocol to adjust the execution strategy during the holding pressure stage in real time.

[0051] The control system collects multimodal data (pressure, temperature, displacement, images, etc.) in real time and uploads them to the optimization module for state evaluation and strategy updates. In addition, the quality indicators after each molding cycle (such as warpage, dimensional deviation, and surface defects) serve as reward feedback to optimize model update parameters and improve prediction accuracy.

[0052] To ensure high time consistency, traceability, and low latency between simulation data, actual acquired data, and control feedback, the system is designed with the following mechanisms, which work together to support the real-time requirements of "online fine-tuning".

[0053] The system employs a precision time protocol for master-slave clock synchronization; the controller (master clock) broadcasts clock signals via the network, and each module (simulation host, optimization module, and sensors) performs periodic offset corrections; the synchronization formula is as follows:

[0054]

[0055] in: The time when the master clock sends a synchronization request;

[0056] : Receive time from the clock;

[0057] Send response time from the clock;

[0058] Master clock receive response time;

[0059] Offset Used to adjust the local system clock to ensure that the clock deviation is < 1ms, meeting the high consistency requirements of timing data in the injection molding process.

[0060] The system also employs a timestamp mechanism.

[0061] All data streams collected by sensors (pressure, temperature, image frames, etc.) and outputs from the simulation module are automatically timestamped with high precision. This constitutes a data record:

[0062]

[0063] For example: Pressure sensor data: Image frame number: Simulation output metrics: Using timestamps to unify the reference time base facilitates the alignment of multi-source data in the input of optimization algorithms.

[0064] To address time alignment issues caused by varying sampling frequencies from different data sources (e.g., 1kHz for pressure data and 30fps for image data) or communication latency, the system incorporates a multi-channel buffer queue within its intelligent optimization module, including:

[0065] Image cache queue

[0066] Sensor buffer queue

[0067] Each model input undergoes timestamp matching via the scheduling module to find the closest time point to the target time. Data combination:

[0068]

[0069] This ensures that the model input data is accurately aligned in the time dimension, reducing error propagation.

[0070] The system uses a sliding window algorithm and interpolation compensation mechanism to clean and correct the input data; the processing flow includes:

[0071] Anomaly rejection: for pressure signals Perform sliding variance monitoring:

[0072]

[0073] like If an abnormal jump is detected, it will be filtered or discarded.

[0074] For gaps caused by dropped frames or lost packets, linear or spline interpolation is used to fill them in:

[0075]

[0076] If an image frame is missing, it is filled by nearest neighbor copying or simulated prediction of the frame to ensure that the input structure is consistent.

[0077] To avoid production interruptions caused by single points of failure, the system is designed with a fault-tolerant rollback mechanism. The specific process is as follows: real-time monitoring of the simulation module and communication interface response status; if three consecutive simulation failures or communication timeouts exceed 500ms, an alarm is triggered; the system automatically rolls back to the previous frame's stable parameter group. Continue the current process; this mechanism ensures that even if the algorithm module or simulation module malfunctions temporarily, the current cycle pressure holding process can still operate stably.

[0078] The feedback control module is the intermediate communication bridge connecting the "intelligent optimization algorithm module" and the "injection molding machine execution control layer" in the system. It can receive the pressure holding stage control parameters (such as pressure holding curve and pressure holding time) output from the optimization algorithm module; parse and encapsulate these control strategies into control command format; and call the injection molding machine interface API to inject the control strategy into the injection molding control system in real time.

[0079] The optimized module output format is as follows:

[0080] Holding pressure vector:

[0081] Time segment vector:

[0082] This indicates that the pressure holding stage is divided into... Each sub-stage has a holding pressure of [values ​​to be filled in]. The time is * Furthermore, the feedback control module can encapsulate it into a control instruction format;

[0083] Traditional pressure holding control uses 2-3 levels of pressure control; the system of this invention allows for any number of segments and continuous smooth interpolation.

[0084] The holding pressure can be modeled as a function. Generated by network fitting or curve interpolation (such as B-splines):

[0085]

[0086] in As basis functions, This represents the pressure value corresponding to the control point.

[0087] After receiving the data, the injection molding machine can execute the pressure curve point by point using a high-frequency controller (above 1kHz).

[0088] After each round of modeling is completed, a reward-feedback-driven adaptive optimization of the model will be performed, specifically as follows:

[0089] After each round of molding is completed, the quality assessment module will collect the quality indicators of the product, including:

[0090] Warpage (calculated from 3D scanning modeling)

[0091] Dimensional deviations (obtained by measuring instruments or visual measurement)

[0092] Surface defect rate (detected by image recognition algorithm)

[0093] Stress distribution uniformity (analyzed by photoelasticity or X-ray diffraction) These quality indicators are converted into a reward function. ,For example:

[0094]

[0095] Weight It is set by process engineers based on quality priorities.

[0096] Based on the reward signal, the optimization module updates the strategy parameters using the strategy gradient method (or other optimization algorithms), enabling the model to gradually learn how to adjust the pressure holding strategy under different working conditions to improve molding quality.

[0097] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A plastic mold injection system for molding injection molded parts, characterized in that: Includes the following modules: The simulation module is used to perform virtual simulation of the holding pressure stage using injection molding simulation software, simulating the effects of different holding pressures and holding times on product quality, including warpage, shrinkage, stress distribution, and temperature field changes. The intelligent optimization algorithm module communicates bidirectionally with the simulation module and the control system. It is used to construct reinforcement learning models, deep neural networks, genetic algorithms or fuzzy control models based on the simulation data, train them to obtain the mapping strategy between pressure holding parameters and quality indicators, and fine-tune the parameters in real-time production by combining the current sensor data. It dynamically outputs the pressure curve and pressure holding time adjustment scheme of the pressure holding stage to achieve closed-loop optimization. The data acquisition module, including pressure sensors, temperature sensors, and image acquisition units, is used to acquire process data in real time during the injection molding process. A data synchronization mechanism is used to ensure the consistency between simulation data and actual sensor data. The data synchronization mechanism includes: A clock synchronization mechanism is used to calibrate the simulation system, sensor devices, and control system with a unified time reference, ensuring that all module data are collected and interacted based on a unified time axis. A timestamp mechanism is used to add high-precision timestamps to all sensor-collected data and simulation data. The buffer mechanism sets up a data cache in the intelligent optimization algorithm module to temporarily store and schedule data streams with different frequencies and delays, ensuring that the data input to the algorithm module is time-aligned. The data alignment and cleaning module is used to perform time axis alignment, outlier removal, interpolation, and missing value completion on sensor data and simulation data to ensure the accuracy of the data input into the intelligent optimization module. The feedback control module is used to dynamically control the injection molding equipment to perform corresponding operations based on the pressure holding curve and time adjustment scheme output in real time by the optimization algorithm, and to feed the adjusted execution parameters back to the simulation module in real time to trigger incremental simulation updates. The quality assessment module is used to collect and evaluate the quality indicators of each batch of molded products, including three-dimensional morphology, dimensional tolerances, surface defects, and stress distribution, and uses them as reinforcement learning feedback signals to update and optimize the model strategy.

2. The plastic mold injection system for injection molding of injection molded parts according to claim 1, characterized in that: The reinforcement learning model used in the intelligent optimization algorithm module adopts the policy gradient method, and the policy network is constructed using the product shrinkage rate, warpage deformation, and surface defect rate as reward functions.

3. The plastic mold injection system for injection molding of injection molded parts according to claim 1, characterized in that: The simulation module is integrated with the control system based on the simulation software API interface. It can dynamically call the simulation program and automatically modify the simulation process parameters, including the holding pressure curve, holding time, mold cavity temperature field, and cooling rate, to achieve automatic simulation analysis under different parameter combinations.

4. The plastic mold injection system for injection molding of injection molded parts according to claim 1, characterized in that: The clock synchronization mechanism adopts a high-precision clock synchronization protocol to perform end-to-end synchronization of the simulation server, sensor acquisition unit, controller and image processing unit, with the maximum time error controlled within 1 millisecond.

5. The plastic mold injection system for injection molding of injection molded parts according to claim 1, characterized in that: The buffer mechanism includes an image data buffer queue and a sensor data buffer queue, and is equipped with a time alignment scheduler. It performs multimodal alignment processing on data from different sources according to timestamps, so that the state vector input to the optimization module accurately matches the real working conditions.

6. The plastic mold injection system for injection molding of injection molded parts according to claim 1, characterized in that: The feedback control module directly calls the control program of the injection molding machine's open interface to adjust the holding pressure section in real time based on the results of the optimization algorithm, and performs continuous curve-level control of the multi-stage holding pressure process.

7. The plastic mold injection system for injection molding of injection molded parts according to claim 1, characterized in that: The quality assessment module includes: A three-dimensional optical scanning device for acquiring morphological data of molded plastic products; a structural stress analysis module for analyzing the influence of residual stress during the holding pressure stage on the mechanical properties of the products; and a defect identification submodule for detecting surface defects, streaks, and bubbles based on image recognition algorithms.

8. A method for optimizing and controlling pressure holding during injection molding, characterized in that, The method is based on a plastic mold injection system for injection molding of injection molded parts as described in claims 1-7. The method includes the following steps: S1 Initialization Phase: Run the simulation module to generate a simulation sample dataset of the pressure holding process based on typical process settings; S2 Optimization Training Phase: Input simulation data to train the intelligent optimization algorithm module and obtain the initial policy network; S3 Data Synchronization Phase: Multimodal sensor data is collected in real time during the production process and aligned using clock synchronization, timestamp mechanism and buffer scheduling; S4 Fine-tuning decision stage: Input the current state vector into the optimization model, output the fine-tuning strategy of holding pressure and holding time, and send it to the injection molding machine controller to adjust the execution process in real time; S5 Feedback Learning Phase: Collect quality feedback of the products in this cycle as a reward signal to update the optimization model, and feed the adjusted strategy parameters back to the simulation module to update the simulation database, realizing the linkage closed loop of simulation and optimization.

Images