Injection molding quality optimization system and method based on data analysis

Abstract

The present invention relates to the technical field of injection molding quality optimization, and in particular to a system and method for injection molding quality optimization based on data analysis. The system includes an injection molding data acquisition module, a data processing module, a data analysis module, a quality optimization control module, and a feedback control module that are sequentially connected in communication. The injection molding data acquisition module collects process parameter data in real time through a sensor network; the data processing module performs data preprocessing and feature extraction using adversarial networks and topological data analysis; the data analysis module identifies potential quality influencing factors through fractal analysis of graph isomorphism; the quality optimization control module adjusts the process parameters based on analysis reports and optimization algorithms; and the feedback control module adjusts the equipment operating parameters in real time and stores the results. The present invention significantly improves the quality control accuracy of the injection molding process through multi-level data analysis and intelligent control, reduces human intervention, and improves production efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of injection molding quality optimization, and in particular to an injection molding quality optimization system and method based on data analysis. Background Art

[0002] Injection molding is an important production process in modern manufacturing. By injecting molten plastic into a mold and then cooling it to form it, various plastic products are finally made. Injection molding technology is widely used in the fields of automobiles, electronics, medical devices and daily necessities. However, due to the complexity of the injection molding process and the involvement of multiple process parameters, product quality is often affected by many factors. At present, there are still the following problems: the existing technology often relies on a single or limited number of sensors for data collection, resulting in insufficient data dimensions and difficulty in fully reflecting the key process parameters in the entire injection molding process. In addition, the real-time and accuracy of data collection are also insufficient, and the dynamic changes of process parameters cannot be reflected in time, thus affecting subsequent quality control; the identification of factors affecting quality in existing systems often relies on experience or simple statistical analysis, and lacks in-depth algorithm support. This method is prone to omission or misjudgment of key influencing factors, making it difficult to achieve accurate optimization of process parameters; traditional quality control systems often rely on manual adjustment of process parameters and lack intelligent optimization control methods. This method is not only inefficient, but also prone to unreasonable process parameter settings due to insufficient operator experience or misjudgment, thus affecting product consistency and stability. Summary of the invention

[0003] To solve the above problems, the present invention provides an injection molding quality optimization system and method based on data analysis, which solves the problem of how to enrich data collection dimensions, improve real-time and accuracy, and accurately identify and optimize key process parameters based on intelligent algorithms during the injection molding process, thereby improving the quality control accuracy of the injection molding process, reducing human intervention, and improving production efficiency.

[0004] To achieve the above object, the technical solution adopted by the present invention is:

[0005] In one aspect, a system for optimizing the quality of injection molding based on data analysis includes an injection molding data acquisition module, a data processing module, a data analysis module, a quality optimization control module, and a feedback control module that are sequentially communicatively connected;

[0006] The injection molding data acquisition module is used to collect process parameter data in the injection molding process in real time through a sensor network;

[0007] The data processing module is used to preprocess and enhance the process parameter data through an adversarial network, perform topological structure mining using topological data analysis, and generate a key feature data set;

[0008] The data analysis module is used to identify potential factors affecting injection molding quality based on key feature data sets and generate a process parameter data analysis report using a fractal analysis method of graph isomorphism;

[0009] The quality optimization control module is used to automatically adjust the injection molding process parameters based on the process parameter data analysis report and the key feature data set, and generate the optimal process parameter setting plan by combining historical data with the optimization algorithm;

[0010] The feedback control module is used to feed back the optimized process parameter setting plan to the injection molding equipment, and adjust the operating parameters of the equipment in real time in combination with the adaptive control algorithm, and store the results in the database.

[0011] Furthermore, the process parameter data includes mold temperature, injection pressure, cooling time, raw material characteristics and equipment status.

[0012] Furthermore, the operation process of the data processing module includes the following steps:

[0013] The sampling deviation of the process parameter data is corrected by multi-scale alignment technology, and the time series of the data is reconstructed and noise is filtered in combination with a bilinear interpolation algorithm;

[0014] Generate diverse and enhanced process parameter data using generative adversarial networks to simulate potential fluctuations under different process conditions;

[0015] Based on the enhanced process parameter data, combined with topological data analysis, the topological structure features in the process data are mined to generate a feature map reflecting the stability of the injection molding process, screen out the key features related to the injection molding quality, and generate a key feature data set; the key feature data set includes mold temperature gradient, injection pressure fluctuation, cooling time change rate, raw material rheological characteristic curve and equipment status parameter change trend.

[0016] Furthermore, the formula for topological data analysis is as follows:

[0017] ;

[0018] in, It represents the result of topological data analysis, which is used to identify the topological structure features in the injection molding process parameter data; n represents the number of dimensions of the process parameters; Indicates the importance of each process parameter to the overall topological characteristics; represents the persistent homology calculation result of the i-th dimension; represents the regularization coefficient, which is used to control the complexity of the overall model; m represents the number of topological features; Represents the weight coefficient of the jth topological feature; The homology group representing the jth topological feature is used to quantify the number of holes in the topological space and its impact on process stability.

[0019] Furthermore, the operation process of the data analysis module includes the following steps:

[0020] Extracting multidimensional features closely related to injection molding quality from the key feature data set, and generating a feature matrix in combination with dynamic changes of various process parameters;

[0021] Based on the feature matrix, a graph structure model is constructed, the complex correlation in the feature matrix is ​​analyzed by a fractal analysis method, potential process parameters affecting the molding quality are identified, and a graph isomorphism model is generated;

[0022] The fractal dimension is used to analyze the correlation between each node and the molding quality, and these factors are prioritized to generate a priority list of potential factors;

[0023] According to the priority list and in combination with the graph isomorphism analysis results, a process parameter data analysis report is generated.

[0024] Furthermore, the formula of the graph isomorphism model is as follows:

[0025] ;

[0026] in, The optimal matching result of the graph isomorphism model is used to identify the isomorphic relationship between the feature graph structures under different process parameter conditions; and They represent two feature graph structures generated under different injection molding conditions, and the feature graph structures are constructed based on the key feature data set; represents the weight coefficients of the i-th and j-th nodes, which are used to measure the relative importance of these process parameters in graph isomorphism analysis; Representation diagram and The distance difference between the i-th process parameter node and the j-th process parameter node in ; Represents the regularization coefficient, which is used to control the complexity of the model; It represents the overall difference between two graph structures and is calculated by the Euclidean distance of key feature datasets.

[0027] Furthermore, the operation process of the quality optimization control module includes the following steps:

[0028] Based on the process parameter data analysis report and combined with historical process data, a reinforcement learning algorithm is applied to generate a preliminary process parameter setting plan;

[0029] Use fuzzy logic multi-objective optimization algorithm to generate the optimal process parameter setting plan;

[0030] The final optimized process parameter setting plan is applied to the injection molding equipment, and the equipment operating parameters are monitored and adjusted in real time. At the same time, the adjusted parameters and optimization results are recorded in the database.

[0031] Furthermore, the feedback control module is specifically based on digital twin technology, and constructs a virtual simulation model through three-dimensional modeling technology, while simulating the operating state of the injection molding equipment in real time, and adaptively controls the actual injection molding equipment based on the simulation results.

[0032] On the other hand, a method for optimizing injection molding quality based on data analysis comprises the following steps:

[0033] Collect process parameter data of the injection molding process in real time through the sensor network;

[0034] Preprocessing and enhancing the process parameter data through an adversarial network, mining the topological structure using topological data analysis, and generating a key feature data set;

[0035] Based on the key feature data set, using the fractal analysis method of graph isomorphism, potential factors affecting the quality of injection molding are identified, and a process parameter data analysis report is generated;

[0036] Automatically adjust injection molding process parameters based on process parameter data analysis reports and key feature data sets, and generate the optimal process parameter setting solution by combining historical data with optimization algorithms;

[0037] The optimized process parameter setting scheme is fed back to the injection molding equipment, and the operating parameters of the equipment are adjusted in real time in combination with the adaptive control algorithm, and the results are stored in the database.

[0038] The beneficial effects of the present invention are:

[0039] In the present invention, the injection molding data acquisition module obtains the process parameter data in the injection molding process in real time through the sensor network, ensuring the integrity and accuracy of the data, and providing a reliable basis for subsequent data analysis and optimization. The data processing module uses an adversarial network to preprocess and enhance the collected process parameter data, and performs topological structure mining through topological data analysis, which can effectively extract key feature data sets from complex data, thereby improving the efficiency and accuracy of data analysis. The data analysis module is based on the fractal analysis method of graph isomorphism, which can deeply analyze the key feature data sets, accurately identify the potential factors affecting the quality of injection molding, and provide a scientific basis for optimization control. The quality optimization control module combines the process parameter data analysis report, the key feature data set and the historical data, generates the optimal process parameter setting scheme through the optimization algorithm, and can automatically adjust the injection molding process parameters, thereby effectively improving the consistency and stability of product quality. The feedback control module adjusts the operating parameters of the injection molding equipment in real time through the adaptive control algorithm, and stores the optimization results in the database, providing data support for subsequent process optimization and improvement, and further improving the intelligence level and self-optimization ability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a module schematic diagram of an injection molding quality optimization system based on data analysis of the present invention.

[0041] Figure 2 It is a flowchart of the operation process of the data analysis module provided by one embodiment of the present invention.

[0042] Figure 3 It is a flow chart of a method for optimizing injection molding quality based on data analysis according to the present invention. DETAILED DESCRIPTION

[0043] See also Figure 1-3 As shown, the present invention relates to an injection molding quality optimization system and method based on data analysis.

[0044] Example 1

[0045] An injection molding quality optimization system based on data analysis, comprising an injection molding data acquisition module, a data processing module, a data analysis module, a quality optimization control module and a feedback control module which are sequentially communicatively connected;

[0046] The injection molding data acquisition module is used to collect process parameter data in the injection molding process in real time through a sensor network; the process parameter data includes mold temperature, injection pressure, cooling time, raw material characteristics and equipment status.

[0047] It should be noted that the injection molding data acquisition module monitors and collects process parameter data in the injection molding process in real time through a high-precision sensor network deployed at various key locations of the injection molding equipment. The details are as follows:

[0048] Sensor Array:

[0049] Mold temperature sensor: Multiple thermocouples and infrared temperature sensors are embedded inside and outside the mold to monitor the temperature distribution in different areas of the mold. Thermocouples can sense small differences in temperature changes, and infrared temperature sensors provide non-contact surface temperature measurement. The acquisition frequency of mold temperature data is set to 10 times per second to capture rapid temperature changes.

[0050] Injection pressure sensor: Piezoelectric sensors are installed at multiple key locations of the injection device (such as nozzles and mold inlets). These sensors can capture instantaneous changes in injection pressure at high frequency. The sensitivity of the piezoelectric sensor is set to 0.1MPa, and the data sampling frequency can reach 20kHz, ensuring that the complete pressure curve of each injection cycle is recorded.

[0051] Cooling time sensor: Use photoelectric sensors and time stamping devices to accurately measure cooling time. The photoelectric sensor is responsible for detecting the coolant flow rate and the cooling state changes of the mold during the cooling process, and the time stamping device records the start and end time of each cooling, with an accuracy of up to milliseconds.

[0052] Raw material characteristic sensor: Integrates online viscometer, density meter and composition analyzer. The online viscometer monitors the viscosity change of raw materials during heating in real time, the density meter measures the density of raw materials, and the composition analyzer analyzes the chemical composition of materials through near-infrared spectroscopy to ensure the consistency of each batch of raw materials.

[0053] Equipment status sensors: including vibration sensors, temperature sensors and position encoders. Vibration sensors are installed in key parts of the equipment, such as the injection unit and hydraulic system, to monitor mechanical vibrations during equipment operation and identify potential failure risks. Position encoders are used to monitor the position and motion trajectory of the injection screw and robotic arm to ensure accurate operation of the equipment.

[0054] All sensor data is transmitted to the central controller in real time via industrial Ethernet or wireless networks (such as Wi-Fi 6 or 5G). To ensure the stability of data transmission, the system is designed with redundant transmission paths. When one path fails, data can be seamlessly transmitted through the backup path. Data transmission uses data encryption technology (such as AES-256) to ensure data security during transmission.

[0055] The edge computing unit is located at the front end of the data acquisition module and is used to perform preliminary processing on the collected raw data. The edge computing unit is equipped with a high-performance embedded processor and FPGA accelerator, which can perform complex signal processing tasks such as data filtering, noise elimination, anomaly detection and preliminary data compression. The edge computing unit also integrates artificial intelligence algorithms to analyze sensor data in real time and identify potential quality issues, thereby reducing the bandwidth requirements for data transmission.

[0056] In order to prevent data loss, the injection molding data acquisition module also includes a local storage unit, which uses an SSD hard disk for temporary data storage. The time for data to be saved in the local storage unit can be adjusted as needed, usually set to 24 hours, so that it can be restored in case of accidents during data transmission. The local storage unit also supports regular automatic backup of data and can export data through USB or other interfaces.

[0057] The data processing module is used to preprocess and enhance the process parameter data through an adversarial network, perform topological structure mining using topological data analysis, and generate a key feature data set;

[0058] The operation process of the data processing module includes the following steps:

[0059] The sampling deviation of the process parameter data is corrected by multi-scale alignment technology, and the time series of the data is reconstructed and noise is filtered in combination with a bilinear interpolation algorithm;

[0060] Specifically, the data processing module first applies multi-scale alignment technology to align the sampling data of different frequencies by analyzing the time characteristics and spatial distribution of each process parameter data. This process uses an algorithm based on cross-correlation analysis to calculate the time offset between different sensors and perform alignment correction on the time axis. After completing the time alignment, the module reconstructs the data in time series through a bilinear interpolation algorithm. This algorithm fills the gaps between sampling points by performing bilinear interpolation in time and space to generate continuous and smooth time series data. This data reconstruction not only improves the resolution of the time series, but also effectively filters the random noise in the sampling process, ensuring the accuracy of subsequent analysis.

[0061] Generate diverse and enhanced process parameter data using generative adversarial networks to simulate potential fluctuations under different process conditions;

[0062] Specifically, after data preprocessing is completed, the module uses a generative adversarial network (GAN) to enhance the process parameter data. GAN consists of a generator and a discriminator. The generator attempts to generate simulated data close to the real data distribution, while the discriminator distinguishes between real data and generated data. Through adversarial training, the generator is gradually optimized and eventually generates an enhanced data set with diversity and authenticity. GAN is trained to simulate possible process parameter fluctuations under different process conditions, such as changes in material fluidity at different temperatures, the impact of equipment aging on the pressure curve, etc. By generating diverse process parameter data, the system can test various extreme process conditions in a virtual environment and identify potential problems that may affect the quality of injection molding. These enhanced data not only increase the richness of the data set, but also provide a more comprehensive sample for subsequent topological structure analysis.

[0063] Based on the enhanced process parameter data, combined with topological data analysis, the topological structure features in the process data are mined to generate a feature map reflecting the stability of the injection molding process, screen out the key features related to the injection molding quality, and generate a key feature data set; the key feature data set includes mold temperature gradient, injection pressure fluctuation, cooling time change rate, raw material rheological characteristic curve and equipment status parameter change trend.

[0064] Specifically, in topological data analysis (TDA), persistent homology is a tool for studying shape features in multidimensional data. By analyzing the topological features of data at different scales, such as the appearance and disappearance of connected components, loops, and holes, the deep structure in the process parameter data can be revealed. For example, persistent homology can help identify how process parameters behave as stable or unstable structures in topological space under different temperature and pressure conditions. Using the results of TDA analysis, the system generates a feature map that reflects the stability of the process. The map shows the topological structure of each process parameter and the evolution of these structures under different process conditions. For example, the distribution of mold temperature may show different topological shapes in different cooling time periods, and the persistence of these shapes can be used to evaluate the stability of the cooling process.

[0065] Through the feature map, the system can screen out key features that are directly related to the quality of injection molding. These features include but are not limited to mold temperature gradients, injection pressure fluctuations, cooling time change rates, raw material rheological characteristic curves, and equipment status parameter change trends. The screening criteria are the persistence of these features under different process conditions and the degree of impact on quality. In addition to analyzing each feature individually, the system also evaluates the interactions between these features. For example, the relationship between mold temperature and injection pressure may have a significant impact on the dimensional accuracy of the final product. The system identifies these interactions through topological analysis and incorporates them into the key feature data set.

[0066] The key feature dataset generated in the end is a high-dimensional dataset that contains detailed information of all the selected key features, including their values, topological relationships, change trends, etc. These data not only provide important inputs for subsequent analysis modules, but can also be used for real-time monitoring and quality assessment during the production process. The key feature dataset will serve as the basis for subsequent analysis and optimization control.

[0067] Furthermore, the formula for topological data analysis is as follows:

[0068] ;

[0069] in, It represents the result of topological data analysis, which is used to identify the topological structure features in the injection molding process parameter data; n represents the number of dimensions of the process parameters; Indicates the importance of each process parameter to the overall topological characteristics; represents the persistent homology calculation result of the i-th dimension; represents the regularization coefficient, which is used to control the complexity of the overall model; m represents the number of topological features; Represents the weight coefficient of the jth topological feature; The homology group representing the jth topological feature is used to quantify the number of holes in the topological space and its impact on process stability.

[0070] The data analysis module is used to identify potential factors affecting injection molding quality based on key feature data sets and generate a process parameter data analysis report using a fractal analysis method of graph isomorphism;

[0071] The operation process of the data analysis module includes the following steps:

[0072] Extracting multidimensional features closely related to injection molding quality from the key feature data set, and generating a feature matrix in combination with dynamic changes of various process parameters;

[0073] Specifically, multidimensional features closely related to the quality of injection molding are extracted from the key feature data set. These features include mold temperature gradient, injection pressure fluctuation, cooling time change rate, raw material rheological characteristic curve and equipment state parameter change trend. Each feature represents a dimension, and the data analysis module obtains the dynamic changes of these features at different time points and process conditions through time series analysis and statistical analysis. The extracted features are not only analyzed individually, but also the dynamic interactions between them are considered. For example, there may be a complex mutual influence relationship between injection pressure fluctuation and mold temperature gradient. The system quantifies these relationships through methods such as multivariate regression analysis and mutual information analysis.

[0074] The rows of the feature matrix represent different time points or process conditions, and the columns represent individual features. Each element in the matrix represents the value of a feature or its rate of change under specific conditions. The construction of the feature matrix uses standardization and normalization to eliminate the differences between different feature dimensions and improve the comparability of the analysis results. In order to capture the dynamic changes of process parameters, the feature matrix contains not only the current eigenvalues, but also the second-order and third-order derivative information such as the rate of change of the eigenvalues ​​and acceleration. This extended matrix structure provides a rich data foundation for the subsequent graph structure model construction and complex correlation analysis.

[0075] Based on the feature matrix, a graph structure model is constructed, the complex correlation in the feature matrix is ​​analyzed by a fractal analysis method, potential process parameters affecting the molding quality are identified, and a graph isomorphism model is generated;

[0076] Specifically, the graph structure model is constructed as follows: In the graph structure model, each node represents a process parameter feature. The weight of the node is represented by the value of the feature at a specific time point or process condition. The system determines the connection relationship (i.e., edge) between nodes by analyzing the feature matrix, and these edges reflect the correlation between different process parameter features. The weight of the edge reflects the strength of the association between two nodes (features). The calculation of the weight is based on metrics such as the correlation coefficient, mutual information, or Euclidean distance of the eigenvalues. For example, if the correlation between the mold temperature gradient and the injection pressure fluctuation is high, the system will assign a larger weight to the edge between them. In order to reduce the complexity of the graph structure, the system will sparse the weights of the edges and only retain the most significant correlations. This optimization process is achieved through sparse regularization (such as L1 regularization) to ensure that the graph structure model is both concise and representative.

[0077] The fractal analysis method is constructed as follows:

[0078] Fractal dimension calculation: Fractal dimension is a key indicator for analyzing complex correlations in graph structures. The system evaluates the degree of correlation between the node (feature) and the overall molding quality by calculating the fractal dimension of each node. The higher the fractal dimension, the greater the impact of the feature on quality at different scales.

[0079] Multi-scale fractal analysis: The system uses multi-scale fractal analysis to evaluate the performance of features at different time scales. This method helps to identify key process parameters that have a more lasting impact on quality over a long time span. For example, the fractal dimension of the mold temperature gradient may show a more significant impact over a longer time window.

[0080] The fractal dimension is used to analyze the correlation between each node and the molding quality, and these factors are prioritized to generate a priority list of potential factors;

[0081] It should be noted that, based on the graph structure model, the system uses the graph isomorphism algorithm to identify graph structures that perform similarly under different process conditions. Graph isomorphism recognition is to classify similar graph structures into the same category by calculating the isomorphic mapping of the graph, thereby identifying the combination of process features that perform consistently under different conditions. The system further performs stability analysis on the isomorphic models to evaluate the persistence of these isomorphic structures under different process conditions. If an isomorphic model appears repeatedly in multiple molding processes, it means that it may be a key factor affecting the molding quality. Based on the fractal dimension, the system sorts the influence of each node (feature) and generates a priority list of potential factors. The higher the priority, the greater the influence of the feature on the molding quality. The priority list is not only based on the fractal dimension, but also combines the stability analysis results of the graph isomorphism model. If a feature appears in multiple isomorphic models and its fractal dimension is high, the system will further increase the priority of the feature.

[0082] According to the priority list and in combination with the graph isomorphism analysis results, a process parameter data analysis report is generated.

[0083] Furthermore, the formula of the graph isomorphism model is as follows:

[0084] ;

[0085] in, The optimal matching result of the graph isomorphism model is used to identify the isomorphic relationship between the feature graph structures under different process parameter conditions; and They represent two feature graph structures generated under different injection molding conditions, and the feature graph structures are constructed based on the key feature data set; represents the weight coefficients of the i-th and j-th nodes, which are used to measure the relative importance of these process parameters in graph isomorphism analysis; Representation diagram and The distance difference between the i-th process parameter node and the j-th process parameter node in ; Represents the regularization coefficient, which is used to control the complexity of the model; It represents the overall difference between two graph structures and is calculated by the Euclidean distance of key feature datasets.

[0086] The quality optimization control module is used to automatically adjust the injection molding process parameters based on the process parameter data analysis report and the key feature data set, and generate the optimal process parameter setting plan by combining historical data with the optimization algorithm;

[0087] The operation process of the quality optimization control module includes the following steps:

[0088] Based on the process parameter data analysis report and combined with historical process data, a reinforcement learning algorithm is applied to generate a preliminary process parameter setting plan;

[0089] Specifically, the Deep Reinforcement Learning (DRL) algorithm is used, the core of which is to optimize decision-making through continuous iteration of tentative operations and environmental feedback. Specifically, the algorithm regards the process parameter setting as a continuous action space, and evaluates the impact of different parameter combinations on quality through interaction with the virtual simulation environment. The system assigns a reward function to each process parameter setting scheme, and the reward value is based on a comprehensive evaluation of actual quality inspection results, production efficiency, energy consumption and other indicators. The reinforcement learning algorithm gradually adjusts the parameter setting by maximizing the cumulative reward to generate a preliminary process parameter setting scheme.

[0090] Use fuzzy logic multi-objective optimization algorithm to generate the optimal process parameter setting plan;

[0091] It should be noted that the fuzzy logic multi-objective optimization algorithm is used to deal with multiple conflicting objectives, such as product quality, production speed, energy consumption, etc. The algorithm uses fuzzy set theory to fuzzify the evaluation criteria of these objectives, so that it can generate optimization solutions under uncertainty and fuzziness. The module first defines an objective function for each optimization objective (such as product accuracy, material utilization, energy consumption), and these functions are weighted according to actual production needs to form a comprehensive objective function. The fuzzy logic system converts the degree of satisfaction of each objective into fuzzy membership, and determines the weight and priority of each objective through fuzzy reasoning and defuzzification. The fuzzy logic multi-objective optimization algorithm generates a Pareto optimal solution set through iterative calculation, which represents the best compromise between different objectives. The module will select an optimal process parameter setting solution from the Pareto solution set to ensure the best comprehensive effect of each objective.

[0092] The final optimized process parameter setting plan is applied to the injection molding equipment, and the equipment operating parameters are monitored and adjusted in real time. At the same time, the adjusted parameters and optimization results are recorded in the database.

[0093] Specifically, the optimized process parameter setting scheme is automatically applied to the injection molding equipment to adjust the equipment's operating parameters (such as injection speed, temperature, pressure curve, etc.) in real time. During the process parameter application process, the module continuously monitors the equipment's operating status and product quality data. Through embedded sensors and Industrial Internet of Things (IIoT) technology, the system collects equipment operation data in real time and compares it with the optimization scheme. If operational deviations or environmental changes are found, the system will immediately adjust the parameters and optimize the equipment through adaptive control algorithms to ensure the stability of the production process. The results of each process parameter adjustment and the corresponding production data will be recorded in the database. These data are not only used for real-time monitoring, but also input into reinforcement learning and optimization algorithms as new historical data, so that the system can gradually optimize and improve the quality and effect of process parameter settings in future production.

[0094] Furthermore, the formula of the fuzzy logic multi-objective optimization algorithm is as follows:

[0095]

[0096] in, It represents the optimal process parameter setting scheme, which is used to optimize the key process parameters in the injection molding process to improve product quality; S represents the process parameter set, including mold temperature, injection pressure, cooling time, raw material rheological properties and equipment status, etc.; represents the i-th objective function, which measures the contribution of a certain process parameter in the multi-objective optimization, such as the uniformity of mold temperature, the stability of injection pressure, the rationality of cooling time, the matching of raw material rheological properties, and the stability of equipment status; represents the weight coefficient associated with the i-th objective function, which is used to balance the relative importance of different optimization objectives, which may include shortening the molding cycle, reducing the scrap rate, improving the consistency of the product, etc.; m represents the number of objective functions, which may include optimization objectives of multiple process parameters; represents the regularization term, which is used to control the complexity of the model; Represents the regularization coefficient, which adjusts the influence of the regularization term on the overall optimization goal and balances the relationship between complexity and optimization effect.

[0097] The feedback control module is used to feed back the optimized process parameter setting scheme to the injection molding equipment, and adjust the operating parameters of the equipment in real time in combination with the adaptive control algorithm, and store the results in the database. The feedback control module is specifically based on digital twin technology, and constructs a virtual simulation model through three-dimensional modeling technology, while simulating the operating state of the injection molding equipment in real time, and adaptively controls the actual injection molding equipment based on the simulation results.

[0098] It should be noted that the construction of the digital twin model is as follows: When the system is initialized, the injection molding equipment is modeled in detail using 3D modeling technology. The model includes the geometric structure, material properties, dynamic behavior, and operation logic of the equipment. The model is calibrated with the historical data of the actual equipment to give it dynamic response characteristics similar to those of the real equipment. As the system runs, the model will continuously obtain real-time data (such as temperature, pressure, speed, etc.) from the actual equipment and dynamically adjust its own parameters to maintain consistency with the equipment status. This dynamic update mechanism ensures the accuracy and real-time performance of the model.

[0099] Specifically, the module acquires the equipment's operating parameters and production environment data in real time through a high-frequency data acquisition system. In order to reduce the impact of network latency and data transmission, the feedback control module uses edge computing technology to perform preliminary data processing and analysis locally. Real-time data is integrated into the digital twin model, which analyzes the current operating status through efficient algorithms and predicts possible equipment deviations or process problems. For example, potential mold blockage or material fluidity problems can be predicted through abnormal changes in the pressure curve.

[0100] The adaptive control algorithm is based on a combination of Model Predictive Control (MPC) and Adaptive Fuzzy Control. MPC is used to predict the future behavior of the equipment under different control parameters, while adaptive fuzzy control can handle the nonlinear characteristics of the system, especially when dealing with changes in the characteristics of different batches of materials. During the injection molding process, the adaptive control algorithm adjusts the key operating parameters of the equipment, such as injection speed, melt temperature, cooling time, etc., in real time based on the prediction results of the digital twin model. This adjustment is made gradually to avoid negative effects on product quality. For example, if it is found that the injection speed is too fast and the internal pressure of the mold increases, the system will reduce the injection speed in the next cycle and extend the cooling time to ensure the stable quality of the final product.

[0101] After each parameter adjustment, the system will immediately detect the effect of the adjustment and input feedback data (such as molding quality test results, equipment status, etc.) back to the digital twin model. Through this closed-loop control, the system can self-correct and gradually optimize process parameters. All adjusted parameters and operation results will be recorded in detail in the database to form a complete data link. This data is not only used for current production optimization, but also provides a reference for future production batches. By analyzing these historical data, the system can identify long-term trends and potential problems and make preventive adjustments in advance.

[0102] In actual operation, the digital twin model will keep in real-time synchronization with the actual equipment. Any changes in the equipment status will be reflected in the model immediately, and simulated prediction and adjustment will be performed through the adaptive control algorithm. This synchronization ensures the timeliness and effectiveness of feedback control. In addition to automatic control, the system can also generate operation optimization suggestions for operators to refer to. For example, if the system detects that the adjustment range of a certain parameter exceeds the safe range, the system will automatically generate an alarm prompt and suggest how to adjust other parameters to maintain production stability.

[0103] Example 2

[0104] A method for optimizing injection molding quality based on data analysis comprises the following steps:

[0105] Collect process parameter data of the injection molding process in real time through the sensor network;

[0106] Preprocessing and enhancing the process parameter data through an adversarial network, mining the topological structure using topological data analysis, and generating a key feature data set;

[0107] Based on the key feature data set, using the fractal analysis method of graph isomorphism, potential factors affecting the quality of injection molding are identified, and a process parameter data analysis report is generated;

[0108] Automatically adjust injection molding process parameters based on process parameter data analysis reports and key feature data sets, and generate the optimal process parameter setting solution by combining historical data with optimization algorithms;

[0109] The optimized process parameter setting scheme is fed back to the injection molding equipment, and the operating parameters of the equipment are adjusted in real time in combination with the adaptive control algorithm, and the results are stored in the database.

[0110] In this embodiment, a method for optimizing injection molding quality based on data analysis is applied to a system for optimizing injection molding quality based on data analysis described in Example 1, which will not be described in detail here.

[0111] In summary, the present invention realizes multi-dimensional real-time acquisition of process parameters by deploying a high-precision sensor network at key positions of the injection molding equipment, and performs preliminary processing of the data through the edge computing unit to ensure the real-time, accuracy and security of the data. In particular, the application of multi-scale alignment technology and bilinear interpolation algorithm effectively corrects the sampling bias and noise and improves the reliability of the data. The process parameter data is simulated and enhanced by generating adversarial networks to generate a variety of process data sets, so that the system can test and identify potential quality problems under different process conditions. This enhanced data set not only enriches the data samples, but also provides a more comprehensive basis for topological data analysis.

[0112] The present invention utilizes topological data analysis technology, and the system can mine topological features that are closely related to the quality of injection molding from complex process data, and further identify potential process factors that affect quality through fractal analysis methods. This multi-dimensional, multi-level data analysis method ensures the accurate identification of key features and provides a scientific basis for subsequent optimization control. Based on the generated process parameter data analysis report, the system can automatically adjust the injection molding process parameters, and generate the optimal process parameter setting plan through reinforcement learning algorithm and fuzzy logic multi-objective optimization algorithm. Combined with the adaptive control algorithm, the system can adjust the operating parameters of the equipment in real time to ensure the stability of the injection molding process and the consistency of product quality.

[0113] The virtual simulation model constructed by the digital twin technology of the present invention is synchronized with the actual equipment in real time, so that the system can dynamically simulate and predict the operating status of the equipment, thereby achieving more accurate feedback control and real-time optimization. This closed-loop control mode not only improves the adaptive ability of the system, but also reduces product defects caused by process fluctuations or equipment abnormalities. Through the automated optimization and real-time control of the system, the various process parameters in the injection molding production process can be optimally configured, thereby shortening the production cycle, improving product quality consistency, and reducing the scrap rate. In addition, the optimized process parameter setting scheme also effectively reduces energy consumption and further reduces production costs.

[0114] The above implementation modes are merely descriptions of the preferred implementation modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary engineering and technical personnel in the field shall fall within the protection scope determined by the claims of the present invention.

Claims

1. An injection molding quality optimization system based on data analysis, characterized in that: It includes an injection molding data acquisition module, a data processing module, a data analysis module, a quality optimization control module and a feedback control module which are sequentially connected in communication; The injection molding data acquisition module is used to collect process parameter data in the injection molding process in real time through a sensor network; The data processing module is used to preprocess and enhance the process parameter data through an adversarial network, perform topological structure mining using topological data analysis, and generate a key feature data set; The data analysis module is used to identify potential factors affecting injection molding quality based on key feature data sets and generate a process parameter data analysis report using a fractal analysis method of graph isomorphism; The quality optimization control module is used to automatically adjust the injection molding process parameters based on the process parameter data analysis report and the key feature data set, and generate the optimal process parameter setting plan by combining historical data with the optimization algorithm; The feedback control module is used to feed back the optimized process parameter setting scheme to the injection molding equipment, and adjust the operating parameters of the equipment in real time in combination with the adaptive control algorithm, and store the results in the database; The operation process of the data processing module includes the following steps: The sampling deviation of the process parameter data is corrected by multi-scale alignment technology, and the time series of the data is reconstructed and noise is filtered in combination with a bilinear interpolation algorithm; Generate diverse and enhanced process parameter data using generative adversarial networks to simulate potential fluctuations under different process conditions; Based on the enhanced process parameter data, combined with topological data analysis, the topological structure features in the process data are mined to generate a feature map reflecting the stability of the injection molding process, screen out key features related to the injection molding quality, and generate a key feature data set; the key feature data set includes mold temperature gradient, injection pressure fluctuation, cooling time change rate, raw material rheological characteristic curve and equipment status parameter change trend; The formula for topological data analysis is as follows: ； in, It represents the result of topological data analysis, which is used to identify the topological structure features in the injection molding process parameter data; n represents the number of dimensions of the process parameters; Indicates the importance of each process parameter to the overall topological characteristics; represents the persistent homology calculation result of the i-th dimension; represents the regularization coefficient, which is used to control the complexity of the overall model; m represents the number of topological features; Represents the weight coefficient of the jth topological feature; The homology group representing the jth topological feature is used to quantify the number of holes in the topological space and its impact on process stability; The operation process of the data analysis module includes the following steps: Extracting multidimensional features closely related to injection molding quality from the key feature data set, and generating a feature matrix in combination with dynamic changes of various process parameters; Based on the feature matrix, a graph structure model is constructed, the complex correlation in the feature matrix is ​​analyzed by a fractal analysis method, potential process parameters affecting the molding quality are identified, and a graph isomorphism model is generated; The fractal dimension is used to analyze the correlation between each node and the molding quality, and these factors are prioritized to generate a priority list of potential factors; Generate a process parameter data analysis report based on the priority list and in combination with the graph isomorphism analysis results; The formula of the graph isomorphism model is as follows: ； in, The optimal matching result of the graph isomorphism model is used to identify the isomorphic relationship between the feature graph structures under different process parameter conditions; and They represent two feature graph structures generated under different injection molding conditions, and the feature graph structures are constructed based on the key feature data set; represents the weight coefficients of the i-th and j-th nodes, which are used to measure the relative importance of these process parameters in graph isomorphism analysis; Representation diagram and The distance difference between the i-th process parameter node and the j-th process parameter node in ; Represents the regularization coefficient, which is used to control the complexity of the model; It represents the overall difference between two graph structures and is calculated by the Euclidean distance of key feature datasets.

2. The injection molding quality optimization system based on data analysis according to claim 1, characterized in that: The process parameter data include mold temperature, injection pressure, cooling time, raw material characteristics and equipment status.

3. The injection molding quality optimization system based on data analysis according to claim 1, characterized in that: The operation process of the quality optimization control module includes the following steps: Based on the process parameter data analysis report and combined with historical process data, a reinforcement learning algorithm is applied to generate a preliminary process parameter setting plan; Use fuzzy logic multi-objective optimization algorithm to generate the optimal process parameter setting plan; The final optimized process parameter setting plan is applied to the injection molding equipment, and the equipment operating parameters are monitored and adjusted in real time. At the same time, the adjusted parameters and optimization results are recorded in the database.

4. The injection molding quality optimization system based on data analysis according to claim 1, characterized in that: The feedback control module is specifically based on digital twin technology, which constructs a virtual simulation model through three-dimensional modeling technology, simulates the operating status of the injection molding equipment in real time, and adaptively controls the actual injection molding equipment based on the simulation results.

5. A method for optimizing injection molding quality based on data analysis, characterized in that: The method is applied to an injection molding quality optimization system based on data analysis as described in any one of claims 1 to 4, comprising the following steps: Collect process parameter data of the injection molding process in real time through the sensor network; Preprocessing and enhancing the process parameter data through an adversarial network, mining the topological structure using topological data analysis, and generating a key feature data set; Based on the key feature data set, using the fractal analysis method of graph isomorphism, potential factors affecting the quality of injection molding are identified, and a process parameter data analysis report is generated; Automatically adjust injection molding process parameters based on process parameter data analysis reports and key feature data sets, and generate the optimal process parameter setting solution by combining historical data with optimization algorithms; The optimized process parameter setting scheme is fed back to the injection molding equipment, and the operating parameters of the equipment are adjusted in real time in combination with the adaptive control algorithm, and the results are stored in the database.

Images