Process parameter optimization method, equipment and storage medium

Abstract

The present application discloses a process parameter optimization method, device and storage medium, which relate to the field of data processing technology. The process parameter optimization method includes: obtaining a model prompt word corresponding to the parameter information to be optimized, wherein the model prompt word is generated based on the structured parameter information corresponding to the parameter information to be optimized; inputting the model prompt word into several process parameter optimization models to obtain the process parameter optimization value output by each of the process parameter optimization models; and performing a preferential comparison of each of the process parameter optimization values ​​to determine the target parameter optimization information. The present application selects the best from the parameter optimization values ​​obtained from different large models to select the target parameters that meet the requirements, thereby optimizing the quality of production products, reducing dependence on the experience of process masters, and improving the accuracy of decision-making.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a process parameter optimization method, device and storage medium. Background Art

[0002] In the molding process of compression molding equipment, the traditional parameter adjustment method is highly dependent on the operator's experience. This method is not only inefficient, but also leads to inconsistent molding results due to differences in experience among different operators. The frequent trial and error process will consume a lot of time and materials. At the same time, it is difficult to cope with the optimization needs of complex process parameters, making it difficult to ensure the stability and consistency of product quality.

[0003] With the development of automated control, sensors, and computer-aided engineering technologies, the ability to monitor and simulate process parameters has been significantly improved. However, these technologies mainly provide auxiliary support and fail to fully utilize production data for intelligent optimization.

[0004] Therefore, with the advancement of Industry 4.0 and intelligent manufacturing, the requirements for efficient and stable production processes are increasing. Summary of the Invention

[0005] The main purpose of this application is to provide a process parameter optimization method, equipment and storage medium, aiming to solve the problem that traditional parameter adjustment methods are highly dependent on the operator's experience.

[0006] To achieve the above objectives, the present application proposes a process parameter optimization method, which includes:

[0007] Obtaining a model prompt word corresponding to the parameter information to be optimized, wherein the model prompt word is generated based on the structured parameter information corresponding to the parameter information to be optimized;

[0008] Inputting the model prompt words into a plurality of process parameter optimization models to obtain the process parameter optimization values ​​output by each of the process parameter optimization models;

[0009] The optimized values ​​of the process parameters are compared and selected to determine the target parameter optimization information.

[0010] In one embodiment, the training process of each process parameter optimization model includes:

[0011] Acquire several sets of model training data and determine model frameworks corresponding to several initial process parameter optimization models, wherein the model training data includes process parameter set values, product quality feedback, and mold parameters;

[0012] Performing data preprocessing, data dimensionality reduction, and feature extraction on each of the model training data to obtain several groups of model structured feature data;

[0013] The structured feature data of each model is input into a model framework corresponding to a number of initial process parameter optimization models for iterative training to obtain each process parameter optimization model.

[0014] In one embodiment, obtaining the model prompt word corresponding to the parameter information to be optimized includes:

[0015] Get the parameter information to be optimized;

[0016] Performing data preprocessing and importance analysis on the parameter information to be optimized to obtain structured parameter information;

[0017] Based on the structured parameter information, a model prompt word corresponding to the parameter information to be optimized is generated.

[0018] In one embodiment, the comparing the optimized values ​​of the process parameters to determine the target parameter optimization information includes:

[0019] Based on a preset integration strategy, the optimized values ​​of the process parameters are integrated to obtain a parameter integration ranking;

[0020] Determining whether the optimized values ​​of the process parameters meet the preset conflict range;

[0021] If so, target parameter optimization information is obtained based on the parameter integration ranking and expert scoring model.

[0022] In one embodiment, obtaining target parameter optimization information based on the parameter integration ranking and expert scoring model includes:

[0023] Inputting the optimized values ​​of each process parameter into the expert scoring model to obtain the expert score corresponding to each optimized value of the process parameter output by the expert scoring model;

[0024] Re-ranking the parameter integration ranking based on the expert scores corresponding to the optimized values ​​of each process parameter to obtain a comprehensive parameter ranking;

[0025] Based on the comprehensive ranking of the parameters, the target parameter optimization information is determined.

[0026] In one embodiment, the training process of the expert scoring model includes:

[0027] Acquire several sets of expert parameter adjustment knowledge information and several sets of parameter adjustment optimization values, and convert each set of expert parameter adjustment knowledge information into expert knowledge rules;

[0028] Based on the expert knowledge rules, an expert knowledge base is established, and the expert knowledge base is injected into the initial expert scoring model;

[0029] The optimized values ​​of the parameter adjustments are input into the initial expert scoring model for iterative training to obtain the expert scoring model.

[0030] In one embodiment, after comparing the optimized values ​​of the process parameters and determining the target parameter optimization information, the method further includes:

[0031] Performing visualization processing on the target parameter optimization information to obtain visualization parameter information;

[0032] Pushing the visualization parameter information to a target user for viewing by the target user, and obtaining user feedback information corresponding to the visualization parameter information;

[0033] Based on the user feedback information, the process parameter optimization process is optimized.

[0034] In one embodiment, before obtaining the model prompt word corresponding to the parameter information to be optimized, the method further includes:

[0035] Acquiring parameter information to be optimized, and inputting the parameter information to be optimized into a process parameter defect recognition model to obtain a process parameter defect type output by the process parameter defect recognition model;

[0036] Based on the process parameter defect type, determining a parameter optimization path corresponding to the parameter information to be optimized by a preset greedy algorithm;

[0037] If the parameter optimization path is a plurality of large model prediction strategies, the model prompt words corresponding to the parameter information to be optimized are obtained.

[0038] In one embodiment, before inputting the model prompt words into a plurality of process parameter optimization models to obtain the process parameter optimization values ​​output by each of the process parameter optimization models, the method further includes:

[0039] If the process parameter optimization model is an external large model, obtaining the external model interface corresponding to the process parameter optimization model;

[0040] According to the external model interface, the external large model is called to execute the step of inputting the model prompt words into several process parameter optimization models through the external large model to obtain the process parameter optimization values ​​output by each process parameter optimization model.

[0041] In addition, to achieve the above objectives, the present application also proposes a process parameter optimization device, which includes:

[0042] An acquisition module is used to obtain the model prompt words corresponding to the parameter information to be optimized;

[0043] An output module, used for inputting the model prompt words into a plurality of process parameter optimization models to obtain the process parameter optimization values ​​output by each of the process parameter optimization models;

[0044] The optimization module is used to compare the optimized values ​​of each process parameter and determine the target parameter optimization information.

[0045] In addition, to achieve the above-mentioned purpose, the present application also proposes a process parameter optimization device, which includes: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the computer program is configured to implement the steps of the process parameter optimization method as described above.

[0046] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium and stores a computer program. When the computer program is executed by a processor, the steps of the process parameter optimization method described above are implemented.

[0047] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the process parameter optimization method as described above.

[0048] The present application provides a process parameter optimization method, device and storage medium. The process parameter optimization method obtains model prompt words corresponding to the parameter information to be optimized, wherein the model prompt words are generated based on the structured parameter information corresponding to the parameter information to be optimized, and then the model prompt words are input into several process parameter optimization models to obtain the process parameter optimization values ​​output by each of the process parameter optimization models, and then the process parameter optimization values ​​are selectively compared to determine the target parameter optimization information, thereby optimizing the quality of production products, reducing dependence on the experience of process masters, and improving the accuracy of decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0050] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0051] Figure 1 A schematic diagram of a process flow diagram provided for Example 1 of the process parameter optimization method of this application;

[0052] Figure 2 Schematic diagram of the system architecture provided for the process parameter optimization method of this application;

[0053] Figure 3 A schematic diagram of the process principle for large-scale model prediction provided for the process parameter optimization method of this application;

[0054] Figure 4 A schematic diagram of the principle of selecting algorithm parameters for multiple large models provided for the process parameter optimization method of this application;

[0055] Figure 5 A brief example diagram of the overall process flow of process parameter optimization provided for the process parameter optimization method of this application;

[0056] Figure 6 This is a schematic diagram of the module structure of the process parameter optimization device according to an embodiment of the present application;

[0057] Figure 7 Schematic diagram of the equipment structure of the hardware operating environment involved in the process parameter optimization method in the embodiment of the present application.

[0058] The purpose, features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0059] It should be understood that the specific embodiments described herein are merely used to explain the technical solutions of the present application and are not intended to limit the present application.

[0060] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0061] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, mobile phone, etc., or an electronic device capable of implementing the above functions, a big data service platform, a process parameter optimization system, etc. The following uses the process parameter optimization system as an example to illustrate this embodiment and the following embodiments.

[0062] Based on this, the present invention provides a method for optimizing process parameters. Figure 1 , Figure 1 A flow chart illustrating the first embodiment of the process parameter optimization method of this application.

[0063] In this embodiment, the process parameter optimization method includes steps S11 to S13:

[0064] Step S11, obtaining a model prompt word corresponding to the parameter information to be optimized, wherein the model prompt word is generated based on the structured parameter information corresponding to the parameter information to be optimized;

[0065] It should be noted that the parameter information to be optimized refers to the relevant information of the process parameters that need to be adjusted and optimized in the production process such as injection molding, die casting, rubber molding, etc., including parameter setting values, etc., and is not limited here.

[0066] It should be further explained that the model prompt words refer to prompt text or data structures generated based on the parameter information to be optimized, which are used to guide the process parameter optimization model to generate parameter optimization values. By obtaining accurate model prompt words, clear and accurate input is provided for subsequent process parameter optimization, ensuring that the optimization model can understand and generate parameter optimization values ​​for specific problems, thereby improving the pertinence and effectiveness of the optimization.

[0067] Specifically, parameter information to be optimized is obtained, and then data preprocessing and importance analysis are performed on the parameter information to be optimized to obtain structured parameter information, and then model prompt words corresponding to the parameter information to be optimized are generated based on the structured parameter information.

[0068] Step S12, inputting the model prompt words into a plurality of process parameter optimization models to obtain the process parameter optimization values ​​output by each of the process parameter optimization models;

[0069] It should be noted that the process parameter optimization model refers to a model built based on a specific algorithm or rule, which is used to generate the optimized value of the process parameter according to the input model prompt word. The model may include models built based on different technologies such as machine learning, deep learning, and expert system, so as to output optimization suggestions for specific problems by analyzing and processing the input data. Figure 2 , Figure 2 Schematic diagram of the system architecture provided for the process parameter optimization method of this application.

[0070] This image shows the system architecture of an intelligent manufacturing system, which is divided into four layers: infrastructure layer, model layer, interaction layer, and application layer. Each layer has its own specific functions and components, which together form a complete intelligent manufacturing solution.

[0071] The infrastructure layer provides the necessary hardware resources, such as servers and workstations, to run the software and algorithms of the intelligent manufacturing system. It also collects and stores various data from the production line, including machine operation data, product quality data, production efficiency data, etc. It also collects and stores industry knowledge, which contains the expertise and experience of specific industries to guide the decision-making and optimization of the intelligent manufacturing system. The regression model in the model layer is used to predict continuous variables, such as product size and weight; the classification model is used to classify data into different categories, such as product qualification and defect type; the data analysis model is used to analyze and interpret patterns and trends in the data; the control model is used to control the operation of the system, such as the automatic adjustment and optimization of the machine; the general model is applicable to a variety of scenarios and problems, and provides a general solution; the reasoning model is used for logical reasoning and decision-making, such as fault diagnosis and parameter optimization.

[0072] Secondly, the network system in the interaction layer is used to connect different devices and systems to achieve data transmission and sharing; the intelligent system integrates artificial intelligence technology to handle complex tasks and make decisions; the physical system is used for actual production lines and machines to execute production tasks. The application layer is applied to compression molding equipment, including: intelligent injection molding machines for injection molding of plastic products, with intelligent control and optimization capabilities; intelligent die-casting machines for die-casting of metal parts, capable of intelligently adjusting parameters to improve quality and efficiency; and intelligent rubber machines for the production of rubber products, with intelligent monitoring and adjustment capabilities. There are no restrictions on the specific type of compression molding equipment here, and configuration can be based on actual conditions.

[0073] The overall architecture diagram illustrates how intelligent manufacturing systems, through collaborative work at different levels, achieve intelligent integration throughout the entire process, from data collection and analysis to decision-making and execution. This architecture enables manufacturers to improve production efficiency, reduce costs, enhance product quality, and achieve more flexible production management.

[0074] It should be further explained that the process parameter optimization values ​​refer to the process parameter optimization values ​​independently output by process parameter optimization models of multiple different model frameworks. In addition, they can also include structured outputs of optimization parameter recommendations and adjustment reasons, recommended adjusted process parameter values, and logical explanations behind these adjustments to improve the interpretability of the optimization.

[0075] Specifically, the model prompt words are input into several process parameter optimization models to obtain the process parameter optimization values ​​output by each process parameter optimization model, so as to utilize the advantages of different models and generate parameter optimization values ​​for process parameters from different angles and methods, thereby providing diversified choices for subsequent parameter optimization value selection and improving the accuracy and reliability of the optimization results. Figure 3 .

[0076] In this embodiment, the system will simultaneously input the model prompt words into multiple pre-trained process parameter optimization models, so that each process parameter optimization model independently generates its own process parameter optimization value based on the input prompt words, providing rich candidate solutions for subsequent parameter optimization value selection.

[0077] Step S13, comparing the optimized values ​​of the process parameters to determine target parameter optimization information;

[0078] It should be noted that the preferential comparison refers to evaluating and comparing multiple candidate process parameter optimization values ​​according to certain standards and methods, so as to screen out the best parameter optimization value.

[0079] It should be further explained that the target parameter optimization information refers to the final process parameter setting values ​​obtained after optimization processing. These parameter values ​​are considered to be the most suitable parameter settings for current production needs after comprehensively considering various factors such as optimization goals and actual production conditions.

[0080] In addition, in the process of preferential comparison, a variety of methods can be used to evaluate the pros and cons of parameter optimization values. For example, a large model can be used to predict the evaluation index prediction value generated after the process parameter optimization value is actually put into production, that is, the prediction value of multiple evaluation indicators such as the degree of product quality improvement, the extent of production cost reduction, and the difficulty of parameter adjustment obtained after the actual injection molding process is optimized according to the process parameter optimization value is estimated, wherein the evaluation index can be set with reference to the user optimization intention of the target user, and a corresponding weight is assigned to each indicator, and then the evaluation index prediction value corresponding to each process parameter optimization value is weighted and scored according to these weights, and finally the parameter value with the highest score is selected as the target parameter optimization information. In a possible embodiment, the system will dynamically adjust the weight of the evaluation index in combination with the data feedback in actual production to better reflect the importance of different optimization goals in actual production, thereby improving the accuracy and practicality of preferential comparison.

[0081] Specifically, based on the preset integration strategy, the optimized values ​​of each process parameter are integrated to obtain the parameter integration ranking, and then it is determined whether the optimized values ​​of each process parameter meet the preset conflict range. If so, the target parameter optimization information is obtained based on the parameter integration ranking and the expert scoring model. Figure 4 .

[0082] Additionally, during the parameter optimization process, the optimized parameter values ​​can be fine-tuned based on actual production conditions. For example, the adjustment range of certain parameters can be revised based on the results of actual trial production, or the optimized parameter values ​​can be temporarily adjusted based on abnormal conditions during production.

[0083] Furthermore, a feedback mechanism is established through the system to collect data and information from the production process in real time, and dynamically adjust the parameter optimization values ​​according to this feedback information, thereby ensuring the stability and effectiveness of the optimization process and ultimately achieving the optimization goal of the production process.

[0084] This embodiment obtains model prompt words corresponding to the parameter information to be optimized, wherein the model prompt words are generated based on the structured parameter information corresponding to the parameter information to be optimized, and then inputs the model prompt words into several process parameter optimization models to obtain the process parameter optimization values ​​output by each of the process parameter optimization models, thereby performing preferential comparison of the process parameter optimization values, determining the target parameter optimization information, and then optimizing the quality of production products, reducing dependence on the experience of process masters, and improving the accuracy of decision-making.

[0085] In a feasible implementation manner, the training process of each process parameter optimization model includes:

[0086] Step S21, obtaining several sets of model training data and determining model frameworks corresponding to several initial process parameter optimization models, wherein the model training data includes process parameter setting values, product quality feedback, and mold parameters;

[0087] It should be noted that the model training data refers to various data sets used to train the process parameter optimization model, including process parameter set values, that is, process parameter values ​​pre-set during the production process; product quality feedback refers to various quality problems that occur in the production process of the product and the corresponding feedback information; additional information refers to other information that affects the production process in addition to the above data, such as production environment conditions; statistical process control data (SPC data, Statistical Process Control) refers to data collected through statistical process control methods, which is used to evaluate the stability and quality level of the production process; curve data refers to curve data of certain parameters changing over time during the production process, such as temperature curves, pressure curves, etc.; mold parameters refer to various parameters related to the mold, such as mold temperature, mold material, etc., thereby providing a rich data basis for subsequent model training, enabling the model to learn the relationship between different parameters and their impact on product quality, thereby improving the accuracy and reliability of the model.

[0088] It should be further explained that the model framework refers to a model framework constructed according to a specific algorithm or architecture, which is used for the subsequent model training process. The appropriate deep learning model framework can be selected according to the characteristics and objectives of the die-casting process. For example, AI models such as recurrent neural networks, long short-term memory networks, and convolutional neural networks can be used to provide a basic structure and algorithm framework for model training, so that the model can learn and optimize based on these frameworks. In addition, in order to accelerate model training, transfer learning can be used to use existing models in related fields for subsequent model training.

[0089] Furthermore, Figure 3 The full-featured general model and inference model can be existing large models that have been trained and can be used directly, while the fine-tuning model is further iteratively trained based on the model framework. For example, a lightweight expert model is obtained by fine-tuning the initial model framework based on the internal knowledge base, industry knowledge base, and historical data to improve the accuracy and robustness of parameter optimization. There are no restrictions here and adjustments can be made according to actual conditions.

[0090] Specifically, several groups of model training data are obtained, and the model frameworks corresponding to several initial process parameter optimization models are determined. In this embodiment, the system collects these data from multiple data sources and integrates them into several groups of model training data to prepare for subsequent model training.

[0091] Furthermore, according to different optimization objectives and data characteristics, a variety of different model frameworks are selected, such as machine learning-based model frameworks, deep learning-based model frameworks, etc., to improve the diversity and adaptability of the model.

[0092] Step S22, performing data preprocessing, data dimensionality reduction, and feature extraction on each of the model training data to obtain several groups of model structured feature data;

[0093] It should be noted that the data preprocessing refers to the cleaning, screening, standardization and other operations on the original model training data to remove noise and outliers in the data and improve the quality and availability of the data; the data dimensionality reduction refers to reducing the dimension of the data through data dimensionality reduction methods, removing redundant information, and improving the training efficiency and performance of the model; the feature extraction refers to extracting features that are important for model training from the original data, so as to better reflect the inherent laws and characteristics of the data, and finally converting the original data into structured feature data suitable for model training through the above data processing methods, thereby improving the learning effect and optimization ability of the model.

[0094] It should be further explained that the model structured feature data refers to data with a specific format and structure that can be used for model training after a series of preprocessing, data dimensionality reduction and feature extraction operations. It contains information that is important for model training, while removing irrelevant or redundant parts, thereby providing high-quality and efficient input for the process parameter optimization model, enabling the model to better learn and understand the patterns in the data, thereby improving the performance and optimization effect of the model.

[0095] Specifically, a variety of data preprocessing (such as data cleaning, data conversion, data standardization, and data filtering), data dimensionality reduction, and feature extraction methods are used to process each set of model training data to obtain several sets of model structured feature data, providing high-quality input for subsequent model training. Among them, the data dimensionality reduction includes an importance analysis between different defects and process parameters, and the correspondence between defects and key parameters is obtained, thereby reducing the dimensionality of the input, reducing misjudgments caused by unnecessary parameters, and improving the accuracy of large models in analyzing process conditions; the feature extraction project refers to extracting meaningful features from the original data, such as statistical features (mean, variance, etc.), time domain features, and frequency domain features; the data filtering identifies and processes outliers and missing values, uses statistical methods or machine learning methods to detect and eliminate abnormal data that does not meet process requirements, and filters out data irrelevant to the target based on process knowledge. In addition, statistical methods and visualization tools can be used to analyze the relationship between process parameters and product quality. For example, machine learning methods such as cluster analysis and association rule mining can be used to discover potential patterns in the data, and by analyzing SPC data and curve data, fluctuations and anomalies in the process can be identified, ultimately achieving structured processing of model training data.

[0096] In one embodiment, during data preprocessing, the system selects appropriate processing methods based on the data characteristics and optimization objectives. For example, filtering algorithms can be used to denoise data sets with significant noise, while normalization methods can be used for data sets with uneven data distribution. During data dimensionality reduction, methods such as principal component analysis (PCA) can be used to remove redundant information and retain key features. During feature extraction, statistical analysis and machine learning methods can be used to extract features that significantly impact the optimization objective, effectively improving the quality and availability of model training data and providing strong support for model optimization.

[0097] In another embodiment, the correlation between product quality issues and process parameters is analyzed. The analysis method is not limited to (PCA (Principal Component Analysis), covariance, probability analysis, etc.). By analyzing the key parameters of the quality issues, the number of input process parameters is reduced, thereby reducing the input interference to the large model and achieving the purpose of improving the accuracy.

[0098] Specifically, users provide the current process parameters, product defect types and degrees, and other relevant remarks to the large language model in a structured manner through a predefined data format (not limited to standard formats such as JSON and XML). Through structured input prompts, the LLM can accurately understand the user's intentions and generate structured output containing optimization parameter suggestions and adjustment reasons based on its internal pre-trained knowledge and reasoning ability. The output also uses a standard data format (not limited to standard formats such as JSON and XML) to clearly present the recommended adjusted process parameter values ​​and the logical explanations behind these adjustments. Therefore, through structured input and output methods, not only the efficiency and accuracy of human-computer interaction are improved, but also the process of optimizing process parameters is made more transparent and explainable.

[0099] Step S23: inputting the structured feature data of each model into a model framework corresponding to a plurality of initial process parameter optimization models for iterative training to obtain each process parameter optimization model.

[0100] Specifically, the structured feature data of each model and the sample labels corresponding to the structured feature data of each model are input into the several initial process parameter optimization models. In one embodiment, a suitable data model is selected according to the characteristics and objectives of the die-casting process. For example, process parameters, SPC data, curve data, mold parameters and other data can be integrated into a multidimensional time series data model to obtain the predicted values ​​output by each initial process parameter optimization model. Then, based on the predicted values ​​and the sample labels, the model loss value is calculated using a loss function. In this embodiment, the loss function can be set according to actual needs and is not specifically limited here. After the model loss value is calculated, the training process ends, and the error back propagation algorithm is used to update the model parameters in each initial process parameter optimization model before the next training is carried out. During the training process, it is determined one by one whether the updated initial process parameter optimization model meets the preset training end conditions. If so, the updated initial process parameter optimization model is used as the process parameter optimization model. If not, the model training is continued. The preset training end conditions include loss convergence and reaching the maximum iteration number threshold, etc., so that the model can better fit the training data and improve the accuracy and reliability of the model by continuously adjusting the model parameters.

[0101] Among them, the processed model structured feature data can be divided into training set, validation set and test set, and the model can be trained with the training set using an optimization algorithm (such as the Adam algorithm); the model performance is evaluated with the validation set. In order to improve the accuracy of the model, the model needs to be optimized, including adjusting the model's hyperparameters, adjusting the model's structure, and using different optimizers; the model performance is evaluated with the test set, and relevant indicators such as mean square error (MSE), mean absolute error (MAE) and R square are calculated to achieve model training and verification.

[0102] Furthermore, during iterative training, the system dynamically adjusts training parameters and parameter optimization values ​​based on the model's training progress and performance. For example, the learning rate and other parameters can be adjusted based on changes in the model's loss function to accelerate convergence. The model's structure and algorithm can also be adjusted based on metrics like accuracy and recall to improve performance. These dynamic adjustment mechanisms effectively enhance model training effectiveness and optimization capabilities, providing high-quality model support for subsequent process parameter optimization.

[0103] This embodiment obtains several sets of model training data and determines model frameworks corresponding to several initial process parameter optimization models, wherein the model training data includes process parameter set values, product quality feedback, and mold parameters. Data preprocessing, data dimensionality reduction, and feature extraction are then performed on each of the model training data to obtain several sets of model structured feature data. Each of the model structured feature data is then input into the model frameworks corresponding to the several initial process parameter optimization models for iterative training to obtain each of the process parameter optimization models. Furthermore, multiple sets of model training data containing rich information (such as process parameter set values, product quality feedback, statistical process control data, etc.) are obtained, enabling the model to learn the relationship between parameters and quality under different scenarios and conditions. This allows the trained process parameter optimization model to have stronger generalization capabilities and better adapt to various production environments and tasks, rather than being limited to a specific scenario or data set. When faced with new and unseen production situations, the model can also provide reasonable and effective optimization suggestions based on the learned knowledge and patterns, thereby improving the stability of the production process and the consistency of product quality, enhancing the accuracy of the model, and reducing reliance on expert experience by increasing the diversity of parameter optimization values.

[0104] In a feasible implementation manner, obtaining the model prompt words corresponding to the parameter information to be optimized includes:

[0105] Step S31, obtaining parameter information to be optimized;

[0106] Specifically, the parameter information to be optimized can be obtained from the production site or relevant data sources. Real-time data in the production process can also be collected through sensors to obtain the current process parameter setting values. Product quality feedback information can be obtained through the quality inspection system. Additional explanatory information can be obtained through manual input or system records, etc., thereby obtaining the parameter information to be optimized in a variety of ways to ensure the comprehensiveness and accuracy of the information. There is no restriction here and it can be set according to actual conditions.

[0107] Additionally, the parameter information to be optimized is format checked and validated to ensure data integrity and accuracy, for example, checking whether the values ​​of the process parameters are within a reasonable range and whether the defect type is a predefined type.

[0108] Step S32, performing data preprocessing and importance analysis on the parameter information to be optimized to obtain structured parameter information;

[0109] It should be noted that the importance analysis refers to evaluating the importance of each parameter information through a certain method to determine which parameters have an important impact on the optimization target, thereby providing a basis for the subsequent generation of model prompt words, thereby converting the original parameter information into structured parameter information, highlighting key information, removing irrelevant information, and improving the generation quality of model prompt words and the optimization effect.

[0110] It should be further explained that the structured parameter information refers to parameter information with a clear format and organizational form obtained after data preprocessing and importance analysis. It is obtained by extracting content that is important to the optimization goal from the original parameter information to be optimized, removing irrelevant or redundant parts, and organizing and formatting it according to certain rules and standards, so as to clearly reflect the relationship between each parameter and their impact on the optimization goal, providing accurate and efficient basic data for the subsequent generation of model prompt words, so that the optimization model can understand and process the input information more quickly, thereby improving the efficiency and quality of generating parameter optimization values.

[0111] Specifically, in one embodiment, during data preprocessing, the system selects appropriate processing methods based on the characteristics of the parameter information and the optimization goal. For example, the system preprocesses and formats the raw parameter information to be optimized to ensure that it meets the input requirements of the optimization model; cleans and standardizes the data, or converts unstructured data into structured data, thereby improving the model's processing efficiency and accuracy of the input data.

[0112] During the importance analysis process, statistical analysis methods (such as correlation analysis and principal component analysis) or machine learning methods (such as feature importance assessment algorithms) can be used to assess the importance of each parameter in the parameter information to be optimized. In one embodiment, importance analysis can be achieved by calculating the correlation coefficient between each parameter in the parameter information to be optimized and the product quality indicator, and then ranking the importance of each parameter based on the correlation coefficient. Alternatively, a machine learning model can be used to output a feature importance score corresponding to each parameter in the parameter information to be optimized, thereby determining the importance of each parameter in the parameter information to be optimized. This effectively improves the quality and usability of structured parameter information and provides strong support for the generation of model prompts.

[0113] In addition, the system combines specific production scenarios and optimization goals to dynamically adjust the content and format of model prompts to better adapt to different optimization needs.

[0114] Step S33: generating model prompt words corresponding to the parameter information to be optimized based on the structured parameter information.

[0115] Specifically, based on the characteristics of structured parameter information and the optimization goal, algorithms (such as machine learning models) or rules (such as preset prompt word generation templates) are used to generate model prompt words, providing clear and accurate input for subsequent process parameter optimization. To improve the quality of model output, various prompt engineering techniques can be employed, such as providing detailed background information and constraints; using example data for guidance; and employing diverse questioning methods to stimulate diverse thinking within large models.

[0116] Furthermore, model prompt words are sent in parallel to multiple LLM (Large Language Model) instances, and multi-threading, multi-process or distributed computing technologies are used to implement parallel reasoning of LLM, so that parameter adjustment suggestions for multiple model outputs can be obtained simultaneously, improving processing efficiency.

[0117] This embodiment obtains the parameter information to be optimized, and then performs data preprocessing and importance analysis on the parameter information to be optimized to obtain structured parameter information, and then generates model prompt words corresponding to the parameter information to be optimized based on the structured parameter information, and then extracts the parameters that really affect the optimization target by removing irrelevant and redundant information, so that the generated model prompt words more accurately reflect the optimization requirements, thereby guiding the process parameter optimization model to more accurately generate parameter optimization values, so that the optimization model can more effectively adjust and optimize the process parameters based on these high-quality prompt words, reduce optimization deviations caused by interference from irrelevant information, and thus improve the accuracy and effectiveness of the optimization, enhance the model's understanding ability and response speed, while improving the quality and reliability of the parameter optimization values, and reducing dependence on manual experience.

[0118] In a feasible implementation manner, the comparing the optimized values ​​of the process parameters to determine the target parameter optimization information includes:

[0119] Step S41, based on a preset integration strategy, integrating the optimized values ​​of the process parameters to obtain a parameter integration ranking;

[0120] It should be noted that the preset integration strategy refers to a scheme for comprehensively processing the optimization values ​​of multiple process parameters according to specific rules and methods, including the voting method: counting the output results of multiple LLMs and selecting the parameter adjustment scheme with the highest frequency; the weighted average method: assigning different weights according to the performance and reliability of the LLM, and then performing weighted averaging on the output results. There is no restriction here. Through the preset integration strategy, these scattered parameter values ​​can be summarized and sorted to form a comprehensive ordered list of parameter values, namely parameter integration sorting, which provides an ordered set of candidate parameter values ​​for subsequent conflict detection and optimization selection, so that the system can more efficiently screen out the optimal optimization parameter values.

[0121] Specifically, the process of generating the parameter integration ranking can take into account various factors, such as the accuracy, reliability, and implementation difficulty of the parameter optimization values. In one possible implementation, each parameter optimization value is weighted and scored based on its historical performance and the confidence level of the model, and then ranked according to the scoring results. This ensures that each set of parameter values ​​in the parameter integration ranking has undergone a comprehensive evaluation, thereby improving the scientific nature and rationality of the parameter optimization value selection.

[0122] Step S42, determining whether the optimized values ​​of the process parameters meet the preset conflict range;

[0123] It should be noted that the preset conflict range refers to a pre-set conflict detection range when there may be a conflict between multiple parameter optimization values. Conflict refers to the situation where there are significant differences in the parameter adjustment direction or amplitude between different parameter optimization values, which may lead to the inability to achieve the optimization goal or a contradiction. In one embodiment, assuming that in the injection molding process, temperature is a key process parameter, different optimization models may give different temperature adjustment suggestions, such as: Model A recommends increasing the temperature by 5°C, Model B recommends reducing the temperature by 3°C, and Model C recommends keeping the current temperature unchanged. If the preset conflict range is ±3°C, then the suggestions of Model A and Model B exceed this range because the difference between them reaches 8°C. In this case, the system will consider that there is a conflict between the two parameter optimization values ​​and further processing is required.

[0124] It should be further noted that the purpose of determining whether the optimized values ​​of each process parameter fall within the preset conflict range is to identify those parameter optimization values ​​that may have problems or require further processing. This process determines whether a conflict exists by detecting whether the differences between the optimized parameter values ​​exceed a preset reasonable range. In this embodiment, the system checks each optimized value of the process parameter one by one according to the preset conflict range to determine whether a conflict exists.

[0125] Specifically, the preset conflict range can be adjusted based on actual production needs and optimization objectives, and is not limited here; it can be set based on actual circumstances. In one possible implementation, the system sets a reasonable conflict threshold based on historical data and expert experience. When the difference between two or more optimized parameter values ​​exceeds this threshold, the system marks them as conflicting. This ensures that the system can effectively identify conflicts in different scenarios, improving the stability and reliability of the optimization process.

[0126] Step S43: If yes, then based on the parameter integration ranking and expert scoring model, target parameter optimization information is obtained.

[0127] It should be noted that the expert scoring model refers to a model built based on expert experience and historical data, and is used to evaluate and score parameter optimization values.

[0128] Specifically, if so, each of the process parameter optimization values ​​is input into the expert scoring model to obtain the expert score corresponding to each of the process parameter optimization values ​​output by the expert scoring model, and then based on the expert score corresponding to each of the process parameter optimization values, the parameter integration ranking is re-sorted to obtain a comprehensive parameter ranking, and then based on the comprehensive parameter ranking, the target parameter optimization information is determined, and the optimal parameter optimization value is selected through the evaluation of the expert scoring model, or the conflicting parameter optimization values ​​are adjusted and integrated to generate a parameter optimization value with the best comprehensive performance.

[0129] This embodiment integrates the optimization values ​​of each process parameter based on a preset integration strategy to obtain a parameter integration ranking, and then determines whether the optimization values ​​of each process parameter meet the preset conflict range. If so, the target parameter optimization information is obtained based on the parameter integration ranking and the expert scoring model, and then multiple similar process parameter optimization values ​​are integrated to combine the advantages of multiple models and avoid the limitations and deviations that may exist in a single model, so that the integrated parameter optimization values ​​can more comprehensively consider various factors and constraints, thereby improving the comprehensiveness and reliability of the parameter optimization values, effectively resolving parameter optimization value conflicts, improving the scientificity and practicality of the parameter optimization values, and reducing dependence on a single model.

[0130] In a feasible implementation, obtaining target parameter optimization information based on the parameter integration ranking and expert scoring model includes:

[0131] Step S51, inputting the optimized values ​​of each process parameter into the expert scoring model to obtain the expert score corresponding to each optimized value of the process parameter output by the expert scoring model;

[0132] It should be noted that the expert scores directly reflect the quality of the parameter optimization values ​​and provide an important basis for the subsequent selection of parameter optimization values. The expert scores are determined by a combination of factors, including the effect of the process parameter optimization values ​​on product quality, the impact on production efficiency, the difficulty of implementation, and cost-effectiveness. This ensures the comprehensiveness and objectivity of the evaluation results and avoids evaluation bias caused by a preference for a single factor. In actual applications, companies can adjust the evaluation criteria and weights of the expert scoring model based on their own production goals and needs to make it more consistent with their actual optimization needs. This is not a restriction here.

[0133] Specifically, the optimized values ​​of each process parameter are input into the expert scoring model, and the expert scoring model extracts relevant features, such as the adjustment range, adjustment reason, etc., from the integrated parameter adjustment suggestions, and then scores each parameter adjustment suggestion. At the same time, the parameter adjustment suggestions are evaluated based on the industry knowledge rule base that has been injected with knowledge in advance, and finally the expert scores corresponding to the optimized values ​​of each process parameter output by the expert scoring model are obtained.

[0134] Step S52, re-ranking the parameter integration ranking based on the expert scores corresponding to the optimized values ​​of each process parameter to obtain a comprehensive parameter ranking;

[0135] It should be noted that the parameter integration ranking refers to the result of sorting all process parameter optimization values ​​according to certain rules in the preliminary evaluation stage. This ranking may be based on the preliminary evaluation results of the parameter optimization values, the confidence of the model or other preset standards. However, this preliminary ranking may not fully take into account the comprehensive performance of the parameter optimization values ​​in actual production. Therefore, the purpose of re-sorting the parameter integration ranking based on the expert scores output by the expert scoring model is to generate a more accurate parameter optimization value ranking that is more in line with actual production needs. The expert score provides a quantitative indicator for each parameter optimization value under multi-dimensional evaluation. By incorporating these scores into the sorting process, it can be ensured that the final parameter comprehensive ranking can more truly reflect the pros and cons of each parameter optimization value.

[0136] It should be further explained that the re-ranking process can adopt a variety of methods. In one embodiment, expert scores are used as the main ranking basis, and other auxiliary indicators (such as model confidence, innovativeness of parameter optimization values, etc.) are combined for comprehensive ranking. In this way, while ensuring the dominance of the main evaluation criteria (expert scores), other important factors are taken into account to improve the rationality and practicality of the ranking results.

[0137] Step S53: Determine the target parameter optimization information based on the comprehensive ranking of the parameters.

[0138] Specifically, because the large model itself has a certain degree of randomness and hallucination in its answers to different questioning methods and questions asked at different times, this situation will lead to a certain error rate in the answers. Therefore, a scoring method is used to use a model trained by multiple experts under multiple questioning methods. In the case of joint decision-making, an expert scoring system is used to score each parameter optimization value. The highest score becomes the optimal parameter optimization value. That is, the process parameter optimization value that ranks first in the comprehensive parameter ranking is used as the target parameter optimization information. The confidence level of the optimal parameter adjustment plan is evaluated and provided to the user for reference.

[0139] This embodiment inputs the optimized values ​​of each process parameter into the expert scoring model to obtain the expert scores corresponding to the optimized values ​​of each process parameter output by the expert scoring model, and then re-sorts the parameter integration ranking based on the expert scores corresponding to the optimized values ​​of each process parameter to obtain a comprehensive parameter ranking, and then determines the target parameter optimization information based on the comprehensive parameter ranking, and then objectively and accurately evaluates each parameter optimization value to ensure that each parameter optimization value has undergone rigorous scientific evaluation, thereby effectively improving the scientific nature and reliability of the parameter optimization value, avoiding optimization deviations caused by the limitations or randomness of a single model, and then more comprehensively evaluating the pros and cons of each parameter optimization value, selecting the parameter optimization value with the best performance in multiple aspects, thereby improving the practicality and feasibility of the parameter optimization value, and resolving parameter optimization value conflicts, avoiding optimization failures or production problems caused by parameter optimization value conflicts.

[0140] In a feasible implementation, the training process of the expert scoring model includes:

[0141] Step S61, obtaining several sets of expert parameter adjustment knowledge information and several sets of parameter adjustment optimization values, and converting each set of expert parameter adjustment knowledge information into expert knowledge rules;

[0142] It should be noted that the expert parameter adjustment knowledge information refers to the knowledge and experience about process parameter adjustment accumulated by experienced process engineers and technicians in long-term production practice, and the knowledge and experience usually exist in an unstructured form, such as technical documents, operating manuals, engineers' verbal descriptions, etc.

[0143] It should be further explained that the parameter adjustment optimization value refers to the process parameter value actually recorded during the production process after expert optimization, including process parameter setting values, product quality feedback, production environment conditions, etc., thereby combining the expert's experience and knowledge with actual production data to provide a rich knowledge base for subsequent model training.

[0144] Specifically, obtain several sets of expert parameter adjustment knowledge information and several sets of parameter adjustment optimization values, such as organizing interviews with experienced die-casting process engineers, mold designers and equipment maintenance personnel; collecting their cause analysis and parameter adjustment experience on various die-casting defects (such as pores, cracks, deformation, etc.); recording their parameter adjustment optimization values ​​under different production conditions (such as different alloys, molds, and equipment status) to obtain several sets of expert parameter adjustment knowledge information and several sets of parameter adjustment optimization values.

[0145] Furthermore, the process of converting expert parameter adjustment knowledge information into expert knowledge rules is to convert unstructured expert experience into structured rules that can be understood and processed by computers. In one embodiment, these rules can be expressed in the form of "IF-THEN", for example, "IF the porosity defect is serious AND the mold temperature is too low, THEN increase the mold temperature" and "IF the product produces a cold shut, then increase the molten metal temperature and increase the injection speed". Through the above conversion process, the expert's experience and knowledge can be systematically applied to process parameter optimization, thereby improving the scientificity and practicality of the parameter optimization value.

[0146] In this embodiment, the system obtains expert parameter adjustment knowledge information through interviews, document analysis, etc., and converts it into specific expert knowledge rules in combination with parameter adjustment optimization values, preparing for subsequent expert knowledge base establishment and model training.

[0147] Furthermore, the process of converting expert knowledge into rules requires specialized knowledge and experience to ensure the accuracy and effectiveness of the rules. In one possible implementation, domain experts can be invited to participate in the rule conversion process. Through expert review and verification, each rule can be ensured to accurately reflect the experts' experience and knowledge. The system also tests and verifies the converted rules to ensure their feasibility and effectiveness in practical applications.

[0148] Step S62: establishing an expert knowledge base based on each of the expert knowledge rules, and injecting knowledge from the expert knowledge base into the initial expert scoring model;

[0149] It should be noted that the expert knowledge base is a system for storing and managing expert knowledge rules, which provides rich knowledge resources for process parameter optimization, integrates and organizes scattered rules to form a systematic knowledge system, which is not only convenient for management and query, but also can provide unified knowledge support for subsequent model training, so that the model can fully consider the experience and knowledge of experts when evaluating the optimization value of process parameters, thereby improving the accuracy and reliability of the evaluation.

[0150] Specifically, based on the expert knowledge rules, an expert knowledge base is established. In order to ensure the timeliness of the knowledge, the expert knowledge is verified by using parameters to adjust the optimization values ​​and simulated data to verify the effectiveness of the rules. According to the verification results, the rules are adjusted and optimized to ensure their accuracy and applicability, and the expert knowledge base is updated regularly.

[0151] Furthermore, the rules from the expert knowledge base are injected one by one into the initial expert scoring model, enabling the model to learn and apply these rules during training, thereby generating evaluation results that better meet actual production needs. This knowledge injection process can be implemented through a variety of technical means, such as rule engines and knowledge graphs. For example, using a rule engine to embed expert knowledge rules into the model's evaluation logic allows the model to reference these rules when evaluating each parameter optimization value. At the same time, the system dynamically manages and updates the injected knowledge to ensure the timeliness and accuracy of the knowledge base.

[0152] Step S63: inputting the optimized values ​​of the parameter adjustments into the initial expert scoring model for iterative training to obtain the expert scoring model.

[0153] Specifically, each of the parameter adjustment optimization values ​​is obtained, and then each of the parameter adjustment optimization values ​​is subjected to data preprocessing, such as data cleaning, normalization, etc., so that each of the parameter adjustment optimization values ​​and the sample labels corresponding to each of the parameter adjustment optimization values ​​are input into the initial expert scoring model to obtain the predicted value output by the initial expert scoring model, and then based on the predicted value and the sample label, the model loss value is calculated using the loss function. In this embodiment, the loss function can be set according to actual needs and is not specifically limited here. After the model loss value is calculated, the training process ends, and the error back propagation algorithm is used to update the model parameters in the initial expert scoring model, and then the next training is carried out. During the training process, it is determined whether the updated initial expert scoring model meets the preset training end conditions. If so, the updated initial expert scoring model is used as the expert scoring model. If not, the model training continues, wherein the preset training end conditions include loss convergence and reaching the maximum number of iterations threshold.

[0154] It should be noted that the initial expert scoring model in the above-mentioned iterative training is an initial expert scoring model that has been knowledge-injected.

[0155] This embodiment obtains several groups of expert parameter adjustment knowledge information and several groups of parameter adjustment optimization values, converts each of the expert parameter adjustment knowledge information into expert knowledge rules, and then establishes an expert knowledge base based on each of the expert knowledge rules, and injects knowledge from the expert knowledge base into the initial expert scoring model, so that each of the parameter adjustment optimization values ​​is input into the initial expert scoring model for iterative training to obtain the expert scoring model, thereby ensuring that the model can fully consider the experience and knowledge of experts when evaluating the process parameter optimization values, improve the accuracy of the model, and make the evaluation results of the model closer to the optimization requirements in actual production. At the same time, when the model scores a certain parameter optimization value, it can clearly display the basis and logic of the scoring, so that production personnel can better understand the evaluation process and results of the model, enhance the interpretability of the model, and reduce dependence on experts. Then, by quickly evaluating a large number of parameter optimization values, the optimal parameter optimization value is selected, reducing the time and cost of manual evaluation and improving the efficiency of the optimization process.

[0156] In a feasible implementation manner, after comparing the optimized values ​​of the process parameters and determining the target parameter optimization information, the method further includes:

[0157] Step S71, performing visualization processing on the target parameter optimization information to obtain visualization parameter information;

[0158] It should be noted that the target parameter optimization information refers to the final process parameter setting values ​​obtained after optimization processing. These parameter values ​​are considered to be the most suitable parameter settings for current production needs after comprehensive consideration of the optimization goals and actual production conditions.

[0159] Specifically, key parameters in the target parameter optimization information, such as temperature, pressure, and injection speed, are displayed in the form of a line graph, clearly showing the changing trends of the parameters before and after optimization. At the same time, production quality indicators such as defect rate and production efficiency before and after optimization are compared through bar charts, intuitively demonstrating the actual effects of the optimization. Furthermore, the system can also display the difference between the actual value of the current parameter and the recommended value in the form of a dashboard, helping users quickly determine whether the parameter needs further adjustment. This converts this complex parameter information into intuitive and easy-to-understand visualization forms, such as charts, curves, and dashboards, so that users can quickly understand and evaluate the optimization results. This allows users to clearly display the direction and magnitude of parameter adjustment and also demonstrate the difference before and after optimization through comparative analysis, providing users with a more intuitive reference basis. The system can use a variety of visualization technologies to convert target parameter optimization information into visual parameter information, facilitating subsequent user review and feedback. There is no restriction on the type of visualization, and it can be set according to actual conditions.

[0160] Step S72: Pushing the visualization parameter information to a target user for viewing by the target user, and obtaining user feedback information corresponding to the visualization parameter information;

[0161] It should be noted that the user feedback information includes satisfaction evaluation of the optimization results, suggestions for parameter adjustment, and adjustments to the optimization target. In one embodiment, the system provides parameter adjustment suggestions to the target user, allowing the target user to choose whether to adopt the adjustment suggestions to obtain user feedback information, such as enabling the system to obtain defect descriptions supplemented by staff, etc., without limitation. The target user can be a production site operator, process engineer, quality management personnel, etc.

[0162] Specifically, the visualization parameter information is pushed to the target user in real time through channels such as a user interface or a mobile device, and a feedback interface is provided to obtain user feedback information corresponding to the visualization parameter information, so that the user can submit feedback information at any time.

[0163] In one embodiment, the visual parameter information can be displayed to target users in the form of a real-time updated dashboard through a dedicated production management platform, and users can provide feedback through a feedback button on the platform.

[0164] Step S73: Optimizing the process parameter optimization flow based on the user feedback information.

[0165] Specifically, the process parameter optimization process is optimized based on the user feedback information, that is, the user's actual experience and opinions are integrated into the optimization process to improve the applicability and effectiveness of the optimization process, because the user feedback information provides an evaluation of the optimization results from the perspective of actual production. This information can help the system discover deficiencies in the optimization process, such as deviations from the optimization goals, irrationality of parameter optimization values, etc., so that by adjusting and optimizing the optimization process, the system can better meet actual production needs and improve the quality and reliability of the optimization results.

[0166] In one embodiment, based on the collected user feedback information, each link in the optimization process is analyzed and adjusted, such as the setting of optimization goals, the selection of model training data, the generation of parameter optimization values, etc., so as to achieve continuous improvement of the optimization process. For example, if the user is generally satisfied with the optimized production quality indicators but is concerned about the increase in costs, the optimization goals can be readjusted and cost control can be incorporated into the optimization goal system. At the same time, the system will also supplement or adjust the training data of the optimization model based on the user's specific suggestions for parameter adjustment to improve the model's adaptability to actual production conditions. For example, if the user feedback indicates that the adjustment range of a certain parameter is too large, resulting in new problems in actual production, the system can adjust the relevant weights in the model training data, retrain the optimization model, and generate more reasonable parameter optimization values. Through this dynamic adjustment mechanism based on user feedback, the system can continuously optimize the process parameter optimization process and improve the practicality and reliability of the optimization results.

[0167] This embodiment obtains visual parameter information by visualizing the target parameter optimization information, and then pushes the visual parameter information to the target user for the target user to view, and obtains user feedback information corresponding to the visual parameter information, so as to optimize the process parameter optimization process based on the user feedback information, and then through visual processing, presents the complex target parameter optimization information to the user in an intuitive and easy-to-understand form, such as charts, curves, etc., so that the user can understand the optimization results more quickly, including the adjustment direction, amplitude and expected effect of the parameters, etc., thereby improving the comprehensibility of the optimization results, and helping users to better accept and apply the parameter optimization values, thereby improving the practicality and effectiveness of the optimization process.

[0168] In a feasible implementation manner, before obtaining the model prompt word corresponding to the parameter information to be optimized, the method further includes:

[0169] Step S81, obtaining parameter information to be optimized, and inputting the parameter information to be optimized into a process parameter defect recognition model to obtain a process parameter defect type output by the process parameter defect recognition model;

[0170] It should be noted that the process parameter defect recognition model refers to a model specifically designed to analyze and identify potential defects caused by improper process parameter settings during the production process. In one embodiment, the model can identify the types of defects that may be caused by the current parameter settings by learning the relationship between normal and abnormal process parameters and product quality in historical data.

[0171] It should be further explained that the process parameter defect types refer to various product quality problems that may be caused by improper process parameter settings during the production process, including dimensional deviations, surface defects, insufficient material performance, appearance problems, and functional failures, etc., which are not limited here.

[0172] Step S82, based on the process parameter defect type, determining a parameter optimization path corresponding to the parameter information to be optimized by a preset greedy algorithm;

[0173] It should be noted that the preset greedy algorithm refers to an algorithm that takes the best choice under the current state in each step of selection, in the expectation that the final result is also the optimal parameter optimization value. The specific greedy algorithm varies depending on the application scenario and is not limited here.

[0174] It should be further explained that the parameter optimization path refers to the path that starts from the current process parameter settings and reaches the optimization goal through a series of orderly adjustment steps. The parameter optimization path includes a variety of different parameter optimization values, such as: prediction strategy based on historical experience library: parameter adjustment is performed according to the historical experience library; single model prediction guidance: a neural network prediction model is used to predict the impact of different parameter settings on product quality and guide parameter adjustment; simulation optimization: the effect of parameter adjustment is evaluated through simulation technology, and the best solution is selected; multiple large model prediction strategies: multiple large language models or other advanced analysis models are used to provide deeper insights and optimization suggestions. There is no restriction on the parameter optimization path here, and it can be set according to actual conditions.

[0175] Specifically, based on the process parameter defect type, a preset greedy algorithm is used to determine the parameter optimization path corresponding to the parameter information to be optimized. In one embodiment, the preset greedy algorithm steps are as follows: defining selection criteria: determining the criteria or rules to be followed for each selection step; local optimal selection: selecting the currently optimal option based on the defined criteria at each step; iterative selection: repeating the local optimal selection until the termination condition is met; and result verification: checking whether the final result satisfies the global optimal or acceptable solution. Furthermore, the above solution is used as the parameter optimization path corresponding to the parameter information to be optimized.

[0176] In one embodiment, for a process parameter in the parameter information to be optimized, a process of determining a parameter optimization path of the cooling time by using a preset greedy algorithm is as follows:

[0177] Define selection criteria: The system obtains the current process parameters (such as temperature, pressure, time, etc.) and the corresponding product quality indicators (such as yield rate and defect rate), and combines the process parameter defect type (such as "excessive temperature fluctuation" and "insufficient pressure") to identify the specific problems to be optimized, namely the selection criteria.

[0178] Local Optimization: Based on the type of process parameter defect, the system offers four optional parameter optimization paths: Historical Experience Library Adjustment, which utilizes historical optimization solutions for similar defects; Single-Model Prediction, which generates recommended values ​​using a single process parameter optimization model; Multi-Model Collaborative Prediction, which comprehensively analyzes the output of multiple large models; and Simulation Optimization, which simulates the effects of different parameter combinations using digital twins. The system then instantly evaluates the expected impact of each parameter optimization path (e.g., estimated quality improvement, cost reduction), selecting the path with the highest benefit relative to the specific issue being optimized. For example, if the multi-model collaborative prediction strategy offers the highest benefit (e.g., 70% quality improvement + 30% cost reduction), it will be prioritized.

[0179] Iterative execution and termination: Apply the parameter optimization path with the highest benefit in the current step, update the process parameters and re-test the quality indicators. It should be noted that this is the step where the parameter optimization path has been determined and the parameter optimization is performed according to the parameter optimization path (that is, step S82 has been separated). Please refer to Figure 5 If the criteria are not met, that is, if the specific problem to be optimized in the "Define Selection Criteria" step is not achieved, then the optimization process returns to the "Local Optimum Selection" step to continue. If the criteria are met or the number of iterations exceeds the limit (e.g., 10), the process is terminated and the final parameters are output.

[0180] Result verification: Dynamic fallback mechanism: If the actual effect of a path deviates from expectations (for example, the quality decreases instead of improving), the system automatically falls back to the previous step and switches to the suboptimal path to try again.

[0181] In another specific embodiment, during the injection molding process, the system detects that the current parameter information to be optimized (melt temperature 200°C, injection pressure 80 MPa, holding time 5 seconds) causes sink marks on the product surface. This is identified as an "insufficient holding pressure" defect through the process parameter defect recognition model. The specific problem to be optimized is "how to improve the sink mark to more than 70% by adjusting the pressure." Based on the greedy algorithm, the system first estimates the benefits of the four optimization schemes relative to the specific problem to be optimized. For example, the benefit of the historical experience strategy is "shrinkage improvement to 70%," the benefit of the single-model strategy is "shrinkage improvement to 54%," the benefit of the multiple large-model prediction strategy is "shrinkage improvement to 85%," and the benefit of the digital twin simulation strategy is "shrinkage improvement to 60%." (It should be noted here that the estimated benefits of the four optimization schemes relative to the specific problem to be optimized are not necessarily the actual benefits obtained after the four optimization schemes are actually run. They are only possible benefits close to the true values ​​obtained through rapid evaluation of the greedy algorithm, thereby improving the efficiency of path selection.) For example, the system selects the multiple large-model prediction strategies with the highest benefits as the parameter optimization path and then performs actual parameter optimization according to the multiple large-model prediction strategies (i.e., executing the steps of "obtaining model prompt words corresponding to the parameter information to be optimized, wherein the model prompt words are generated based on the structured parameter information corresponding to the parameter information to be optimized; inputting the model prompt words into several process parameter optimization models to obtain process parameter optimization values ​​output by each of the process parameter optimization models; and selectively comparing the process parameter optimization values ​​to determine the target parameter optimization information").

[0182] Furthermore, if new problems such as flash appear after adjusting the target parameter optimization information of the multiple large model prediction strategies, the system will automatically fall back to the last valid solution and try again until the quality target is achieved or the iteration limit is reached. Figure 5 .

[0183] Step S83: If the parameter optimization path is a plurality of large model prediction strategies, then the model prompt words corresponding to the parameter information to be optimized are obtained.

[0184] Specifically, if the parameter optimization path is a multi-model prediction strategy, then the model prompt words corresponding to the parameter information to be optimized can be obtained by referring to Figure 5 ,The process shown in the figure is an automated defect parameter optimization ,process.

[0185] This embodiment obtains parameter information to be optimized and inputs the parameter information to be optimized into a process parameter defect recognition model to obtain the process parameter defect type output by the process parameter defect recognition model, and then determines the parameter optimization path corresponding to the parameter information to be optimized through a preset greedy algorithm based on the process parameter defect type. If the parameter optimization path is a plurality of large model prediction strategies, the model prompt words corresponding to the parameter information to be optimized are obtained, and then the problem or defect type existing in the current process parameter setting is accurately identified, which helps to quickly locate the source of the problem, provide a clear direction for subsequent optimization work, and automatically determine the best parameter optimization path. At the same time, through the automated parameter optimization value selection, human intervention is reduced, decision-making efficiency is improved, and decision-making errors caused by human factors are reduced, thereby enhancing the targeted optimization and reducing dependence on expert experience.

[0186] In a feasible embodiment, before inputting the model prompt words into a plurality of process parameter optimization models to obtain the process parameter optimization values ​​output by each of the process parameter optimization models, the method further includes:

[0187] Step S91: if the process parameter optimization model is an external large model, obtaining the external model interface corresponding to the process parameter optimization model;

[0188] It should be noted that the term "external large models" refers to highly complex and advanced algorithmic models developed by third-party organizations or companies. These models typically run on cloud computing platforms or remote servers and can be accessed and interacted with through network interfaces. These include, but are not limited to, deep learning models, reinforcement learning models, or other machine learning models, capable of processing large amounts of data and providing advanced analytical and predictive capabilities. These models, owing to their sheer scale and powerful computing power, are capable of handling complex optimization tasks and providing more accurate and reliable optimization results.

[0189] Specifically, in intelligent manufacturing systems, when process parameters need to be optimized, if there is no corresponding optimization model internally or more advanced technology needs to be used, or when the target user chooses to directly use an external model for parameter optimization, the system will choose to call an external large model to complete this task. Therefore, it is necessary to obtain an external model interface to be able to communicate and exchange data with these remotely running large models, thereby utilizing their optimization capabilities.

[0190] Step S92: calling the external large model according to the external model interface to execute the step of inputting the model prompt words into a plurality of process parameter optimization models through the external large model to obtain the process parameter optimization values ​​output by each of the process parameter optimization models.

[0191] Specifically, the method provided by the external model interface is used to pass the model prompt words as parameters to the external large model, and then the model prompt words are transmitted from the local system to the external large model on the remote server through the network, so that after the external large model receives the input data, it executes its internal algorithm for analysis and calculation to generate optimized process parameter values, and then receives the optimization results from the external large model, stores them in the local system, and applies them to the application layer for subsequent process adjustments and production decisions, thereby achieving the purpose of utilizing the powerful computing and analysis capabilities of the external large model to achieve the purpose of accurate optimization of process parameters.

[0192] In this embodiment, if the process parameter optimization model is an external large model, the external model interface corresponding to the process parameter optimization model is obtained, and then the external large model is called according to the external model interface, so as to execute the step of inputting the model prompt words into several process parameter optimization models through the external large model to obtain the process parameter optimization values ​​output by each process parameter optimization model, thereby improving the accuracy of process parameter optimization by calling the external large model that has been trained and verified with a large amount of data and has high accuracy and reliability, thereby improving product quality and production efficiency. At the same time, due to the use of the external model interface, different optimization models can be easily switched to adapt to different production needs and conditions, thereby reducing development and maintenance costs, and enabling the system to quickly adapt to changes, improving the system's adaptability, and simplifying the integration and deployment process of the process parameter optimization model, thereby reducing implementation time and accelerating project progress.

[0193] For example, to help understand the implementation process of the process parameter optimization method, please refer to Figure 5 , Figure 5 A brief example diagram of the overall process flow of process parameter optimization provided for the process parameter optimization method of this application.

[0194] Specifically, the figure shows an implementation process for optimizing process parameter defects. The process begins with defect discovery and then uses neural network technology to identify defect types. Once the defects are classified, the system determines the next course of action based on the classification results.

[0195] Furthermore, if the identified defect type is a machine problem, manual intervention is required for processing, such as pushing it to staff for further manual intervention to resolve it. For non-machine problems, that is, parameter problems, the system will use a greedy algorithm to select parameter optimization values ​​to determine the best optimization path. This parameter optimization value may involve one or more of a neural network prediction strategy, multiple large model prediction strategies, or a historical experience library prediction strategy. There is no restriction here, and independent predictions or multiple-selection predictions can be performed. After determining the optimization path, the system will check whether the parameters of the selected parameter optimization values ​​meet the preset basic rules. If the parameters meet the rules, the system will evaluate whether the defect has been resolved; if the defect still exists, the process will return to the neural network defect type identification stage and reclassify the defect. If the parameters do not meet the rules, the system will also return to the step of using a greedy algorithm to select parameter optimization values ​​to determine the best optimization path until the parameters meet the rules.

[0196] For example, during the injection molding process, if a porosity defect is detected on the product surface, the system first uses a neural network to determine whether this is due to equipment failure or improper parameter settings. If it is a parameter issue, the system uses a greedy algorithm to select parameter optimization values ​​to determine the best optimization path, outputting the corresponding optimization parameters based on the strategy in the best optimization path. Furthermore, if the recommended parameter adjustment complies with production rules and the defect is resolved after implementation, the process ends. If the defect still exists, the defect type needs to be re-identified and the optimization process continues. If the recommended parameter adjustment does not comply with production rules, the system returns to the step of using a greedy algorithm to select parameter optimization values ​​to determine the best optimization path until the parameters comply with the rules. By combining artificial intelligence and expert knowledge, continuous learning and adjustment can be carried out to improve the stability of the production process and product quality, reducing reliance on manual labor.

[0197] It should be noted that the examples in the figures are only used to understand the present application and do not constitute a limitation on the process parameter optimization method of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0198] It should be understood that the order of execution of the steps in the above embodiments does not necessarily mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0199] This application also provides a process parameter optimization device, please refer to Figure 6 , the process parameter optimization device includes:

[0200] An acquisition module 61 is configured to acquire a model prompt word corresponding to the parameter information to be optimized, wherein the model prompt word is generated based on the structured parameter information corresponding to the parameter information to be optimized;

[0201] An output module 62 is used to input the model prompt words into a plurality of process parameter optimization models to obtain the process parameter optimization values ​​output by each of the process parameter optimization models;

[0202] The optimization module 63 is used to compare the optimized values ​​of the process parameters and determine the target parameter optimization information.

[0203] The process parameter optimization device is also used for:

[0204] Acquire several sets of model training data and determine model frameworks corresponding to several initial process parameter optimization models, wherein the model training data includes process parameter set values, product quality feedback, and mold parameters;

[0205] Performing data preprocessing, data dimensionality reduction, and feature extraction on each of the model training data to obtain several sets of model structured feature data;

[0206] The structured feature data of each model is input into a model framework corresponding to a number of initial process parameter optimization models for iterative training to obtain each process parameter optimization model.

[0207] The process parameter optimization device is also used for:

[0208] Get the parameter information to be optimized;

[0209] Performing data preprocessing and importance analysis on the parameter information to be optimized to obtain structured parameter information;

[0210] Based on the structured parameter information, a model prompt word corresponding to the parameter information to be optimized is generated.

[0211] The process parameter optimization device is also used for:

[0212] Based on a preset integration strategy, the optimized values ​​of the process parameters are integrated to obtain a parameter integration ranking;

[0213] Determining whether the optimized values ​​of the process parameters meet the preset conflict range;

[0214] If so, target parameter optimization information is obtained based on the parameter integration ranking and expert scoring model.

[0215] The process parameter optimization device is also used for:

[0216] Inputting the optimized values ​​of each process parameter into the expert scoring model to obtain the expert score corresponding to each optimized value of the process parameter output by the expert scoring model;

[0217] Re-ranking the parameter integration ranking based on the expert scores corresponding to the optimized values ​​of each process parameter to obtain a comprehensive parameter ranking;

[0218] Based on the comprehensive ranking of the parameters, the target parameter optimization information is determined.

[0219] The process parameter optimization device is also used for:

[0220] Acquire several sets of expert parameter adjustment knowledge information and several sets of parameter adjustment optimization values, and convert each set of expert parameter adjustment knowledge information into expert knowledge rules;

[0221] Based on the expert knowledge rules, an expert knowledge base is established, and the expert knowledge base is injected into the initial expert scoring model;

[0222] The optimized values ​​of the parameter adjustments are input into the initial expert scoring model for iterative training to obtain the expert scoring model.

[0223] The process parameter optimization device is also used for:

[0224] Performing visualization processing on the target parameter optimization information to obtain visualization parameter information;

[0225] Pushing the visualization parameter information to a target user for viewing by the target user, and obtaining user feedback information corresponding to the visualization parameter information;

[0226] Based on the user feedback information, the process parameter optimization process is optimized.

[0227] The process parameter optimization device is also used for:

[0228] Acquiring parameter information to be optimized, and inputting the parameter information to be optimized into a process parameter defect recognition model to obtain a process parameter defect type output by the process parameter defect recognition model;

[0229] Based on the process parameter defect type, determining a parameter optimization path corresponding to the parameter information to be optimized by a preset greedy algorithm;

[0230] If the parameter optimization path is a plurality of large model prediction strategies, the model prompt words corresponding to the parameter information to be optimized are obtained.

[0231] The process parameter optimization device is also used for:

[0232] If the process parameter optimization model is an external large model, obtaining the external model interface corresponding to the process parameter optimization model;

[0233] According to the external model interface, the external large model is called to execute the step of inputting the model prompt words into several process parameter optimization models through the external large model to obtain the process parameter optimization values ​​output by each process parameter optimization model.

[0234] The process parameter optimization device provided in this application, employing the process parameter optimization method of the above-described embodiment, can solve the technical problems described in the background art. Compared with the prior art, the beneficial effects of the process parameter optimization device provided in this application are the same as those of the process parameter optimization method provided in the above-described embodiment, and the other technical features of the process parameter optimization device are the same as those disclosed in the above-described embodiment and are not further described here.

[0235] The present application provides a process parameter optimization device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the process parameter optimization method in the above-mentioned embodiment one.

[0236] Reference below Figure 7 , which shows a schematic diagram of the structure of a process parameter optimization device suitable for implementing the embodiments of the present application. The process parameter optimization device in the embodiments of the present application may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), in-vehicle terminals (such as in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 7 The process parameter optimization device shown is merely an example and should not limit the functions and scope of use of the embodiments of the present application.

[0237] like Figure 7As shown, the process parameter optimization device may include a processing device 1001 (e.g., a central processing unit, graphics processing unit, etc.), which can perform various appropriate actions and processes based on programs stored in a read-only memory 1002 or programs loaded from a storage device 1003 into a random access memory 1004. The random access memory 1004 also stores various programs and data required for the operation of the process parameter optimization device. The processing device 1001, the read-only memory 1002, and the random access memory 1004 are interconnected via a bus 1005. An input / output interface 1006 is also connected to the bus. Typically, the following systems can be connected to the input / output interface 1006: an input device 1007 including, for example, a touch screen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; a storage device 1003 including, for example, a magnetic tape or hard disk; and a communication device 1009. The communication device 1009 can allow the process parameter optimization device to communicate with other devices wirelessly or wired to exchange data. Although the figure shows a process parameter optimization device with various systems, it should be understood that it is not required to implement or have all of the systems shown. More or fewer systems may be implemented or have alternatively.

[0238] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program comprising program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device 1003, or installed from a read-only memory 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are performed.

[0239] The process parameter optimization device provided in this application, employing the process parameter optimization method described in the aforementioned embodiment, can resolve the technical problems described in the background art. Compared to the prior art, the beneficial effects of the process parameter optimization device provided in this application are the same as those of the process parameter optimization method described in the aforementioned embodiment. Other technical features of the process parameter optimization device are the same as those disclosed in the aforementioned embodiment and are not further elaborated upon here.

[0240] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0241] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0242] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, a computer program) stored thereon, and the computer-readable program instructions are used to execute the process parameter optimization method in the above-mentioned embodiment.

[0243] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0244] The computer-readable storage medium may be included in the process parameter optimization device; or it may exist independently without being assembled into the process parameter optimization device.

[0245] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the process parameter optimization device, the process parameter optimization device:

[0246] Obtaining a model prompt word corresponding to the parameter information to be optimized, wherein the model prompt word is generated based on the structured parameter information corresponding to the parameter information to be optimized;

[0247] Inputting the model prompt words into a plurality of process parameter optimization models to obtain the process parameter optimization values ​​output by each of the process parameter optimization models;

[0248] The optimized values ​​of the process parameters are compared and selected to determine the target parameter optimization information.

[0249] Computer program code for performing the operations of the present application may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0250] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, program segment or a part of code, and the module, program segment or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented by a dedicated hardware-based system that performs the specified function or operation, or can be implemented by a combination of dedicated hardware and computer instructions.

[0251] The modules described in the embodiments of the present application may be implemented in software or hardware, wherein the name of a module does not necessarily limit the unit itself.

[0252] The computer-readable storage medium provided in this application stores computer-readable program instructions (i.e., a computer program) for executing the aforementioned process parameter optimization method, thereby resolving the technical problems described in the background art. Compared to the prior art, the beneficial effects of the computer-readable storage medium provided in this application are similar to those of the process parameter optimization method provided in the aforementioned embodiments, and are not further elaborated here.

[0253] An embodiment of the present application provides a computer program product, including a computer program, which implements the steps of the process parameter optimization method as described above when executed by a processor.

[0254] The computer program product provided in this application can solve the technical problems in the background technology. Compared with the prior art, the beneficial effects of the computer program product provided in the embodiment of this application are the same as the beneficial effects of the process parameter optimization method provided in the above embodiment, which will not be repeated here.

[0255] The above description is only part of the embodiments of the present application and does not limit the patent scope of the present application. All equivalent structural transformations made by using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect application in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A process parameter optimization method, characterized in that: include: Obtaining a model prompt word corresponding to the parameter information to be optimized, wherein the model prompt word is generated based on the structured parameter information corresponding to the parameter information to be optimized; Inputting the model prompt words into a plurality of process parameter optimization models to obtain the process parameter optimization values ​​output by each of the process parameter optimization models; Comparing the optimized values ​​of the process parameters to determine the target parameter optimization information; The step of comparing the optimized values ​​of the process parameters to determine target parameter optimization information includes: Based on the preset integration strategy, the optimized values ​​of each process parameter are integrated to obtain a parameter integration ranking; it is determined whether the optimized values ​​of each process parameter meet the preset conflict range; if so, the target parameter optimization information is obtained based on the parameter integration ranking and the expert scoring model.

2. The process parameter optimization method according to claim 1, wherein: The training process of each process parameter optimization model includes: Acquire several sets of model training data and determine model frameworks corresponding to several initial process parameter optimization models, wherein the model training data includes process parameter set values, product quality feedback, and mold parameters; Performing data preprocessing, data dimensionality reduction, and feature extraction on each of the model training data to obtain several groups of model structured feature data; The structured feature data of each model is input into a model framework corresponding to a number of initial process parameter optimization models for iterative training to obtain each process parameter optimization model.

3. The process parameter optimization method according to claim 1, wherein: The step of obtaining the model prompt words corresponding to the parameter information to be optimized includes: Get the parameter information to be optimized; Performing data preprocessing and importance analysis on the parameter information to be optimized to obtain structured parameter information; Based on the structured parameter information, a model prompt word corresponding to the parameter information to be optimized is generated.

4. The process parameter optimization method according to claim 1, wherein: The target parameter optimization information is obtained based on the parameter integration ranking and expert scoring model, including: Inputting the optimized values ​​of each process parameter into the expert scoring model to obtain the expert score corresponding to each optimized value of the process parameter output by the expert scoring model; Re-ranking the parameter integration ranking based on the expert scores corresponding to the optimized values ​​of each process parameter to obtain a comprehensive parameter ranking; Based on the comprehensive ranking of the parameters, the target parameter optimization information is determined.

5. The process parameter optimization method according to claim 1, wherein: The training process of the expert scoring model includes: Acquire several sets of expert parameter adjustment knowledge information and several sets of parameter adjustment optimization values, and convert each set of expert parameter adjustment knowledge information into expert knowledge rules; Based on the expert knowledge rules, an expert knowledge base is established, and the expert knowledge base is injected into the initial expert scoring model; The optimized values ​​of the parameter adjustments are input into the initial expert scoring model for iterative training to obtain the expert scoring model.

6. The process parameter optimization method according to claim 1, wherein: After comparing the optimized values ​​of the process parameters and determining the target parameter optimization information, the method further includes: Performing visualization processing on the target parameter optimization information to obtain visualization parameter information; Pushing the visualization parameter information to a target user for viewing by the target user, and obtaining user feedback information corresponding to the visualization parameter information; Based on the user feedback information, the process parameter optimization process is optimized.

7. The process parameter optimization method according to claim 1, wherein: Before obtaining the model prompt word corresponding to the parameter information to be optimized, the method further includes: Acquiring parameter information to be optimized, and inputting the parameter information to be optimized into a process parameter defect recognition model to obtain a process parameter defect type output by the process parameter defect recognition model; Based on the process parameter defect type, determining a parameter optimization path corresponding to the parameter information to be optimized by a preset greedy algorithm; If the parameter optimization path is a plurality of large model prediction strategies, the model prompt words corresponding to the parameter information to be optimized are obtained.

8. The process parameter optimization method according to claim 1, wherein: Before inputting the model prompt words into a plurality of process parameter optimization models to obtain the process parameter optimization values ​​output by each of the process parameter optimization models, the method further includes: If the process parameter optimization model is an external large model, obtaining the external model interface corresponding to the process parameter optimization model; According to the external model interface, the external large model is called to execute the step of inputting the model prompt words into several process parameter optimization models through the external large model to obtain the process parameter optimization values ​​output by each process parameter optimization model.

9. A process parameter optimization device, characterized in that: The process parameter optimization device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the process parameter optimization method according to any one of claims 1 to 8.

10. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the process parameter optimization method according to any one of claims 1 to 8 are implemented.

Images