A process flow simulation method and system for a cloud manufacturing mode
By simulating the process flow in a cloud manufacturing model, and combining local data acquisition, environmental assessment, and cloud platform data matching, the simulation parameters were optimized, which solved the problem of insufficient simulation accuracy caused by the limitations of data acquisition and achieved higher simulation accuracy.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- GONGYEYUN MFG (SICHUAN) INNOVATION CENT CO LTD
- Filing Date
- 2022-08-03
- Publication Date
- 2026-05-22
Smart Images
Figure CN115221796B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of industrial virtual simulation technology, specifically to a process simulation method and system for cloud manufacturing. Background Technology
[0002] Process simulation technology is a commonly used virtual technology for industrial processes. By visually simulating industrial processes, it can effectively improve the resource utilization rate of enterprises, increase product production efficiency and market competitiveness. Simulation technology mainly simulates production based on the collected process flow-related data. However, it is inevitable that some parameter data cannot be collected immediately. Currently, the commonly used data acquisition method is to obtain data through big data comparison. However, due to certain drawbacks of the acquisition method, the final simulation effect deviates from the actual process flow.
[0003] In existing technologies, when performing process flow simulation, the methods for obtaining missing parts of the collected data are limited and the acquisition process is not rigorous enough, resulting in insufficient consistency between the final determined data and the actual situation, which in turn affects the accuracy of the simulation. Summary of the Invention
[0004] This application provides a process simulation method and system for cloud manufacturing, which addresses the technical problem in the prior art where the methods for obtaining missing parts of the collected data are limited and the acquisition process is not rigorous enough, resulting in insufficient consistency between the final determined data and the actual situation, thus affecting the accuracy of the simulation.
[0005] In view of the above problems, this application provides a process simulation method and system for cloud manufacturing.
[0006] Firstly, this application provides a process simulation method for cloud manufacturing, comprising: collecting simulation data of a local process according to process simulation parameter requirements to obtain process collection data; determining collection environment parameters based on the process collection data, and evaluating multiple environment parameters based on the collection environment parameters to obtain multi-environment predicted collection data; using a cloud manufacturing platform, matching process cooperation data based on the process collection data to obtain a cooperative process collection dataset; obtaining simulated process product demand parameters; calculating standard demand parameter deviation values based on the simulated process product demand parameters to obtain demand parameter deviation information; setting weights for the cooperative process collection dataset based on the demand parameter deviation information; using the multi-environment predicted collection data and the process collection data to form local training data, and performing federated learning using the local training data and the cooperative process collection dataset; optimizing model parameters according to the weights of the cooperative process collection dataset to obtain a parameter-optimized federated model, and determining process simulation parameter information.
[0007] Secondly, this application provides a process simulation system for cloud manufacturing, comprising: a data acquisition module for acquiring simulation data of a local process according to process simulation parameter requirements, thereby obtaining process acquisition data; a parameter evaluation module for determining acquisition environment parameters based on the acquired process data, and performing multi-environment parameter evaluation based on the acquired environment parameters to obtain multi-environment predicted acquisition data; a data matching module for using a cloud manufacturing platform to perform process cooperation data matching based on the acquired process data, thereby obtaining a cooperative process acquisition dataset; and a parameter acquisition module for obtaining simulated process products. The system includes: a requirement parameter module; a parameter calculation module for calculating standard requirement parameter deviations based on the simulated process product requirement parameters to obtain requirement parameter deviation information; a weight setting module for setting weights on the collaborative process data acquisition dataset based on the requirement parameter deviation information; a federated modeling module for using the multi-environment prediction data acquisition data and the process flow data acquisition data to form local training data, and using the local training data and the collaborative process data acquisition dataset for federated learning; and a model optimization module for optimizing model parameters according to the weights of the collaborative process data acquisition dataset to obtain a parameter-optimized federated model and determine the process flow simulation parameter information.
[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:
[0009] This application provides a process simulation method for cloud manufacturing. It involves collecting simulation data from a local process according to simulation parameter requirements, obtaining process data, determining environmental parameters, and evaluating multiple environmental parameters to obtain multi-environment predicted data. The method then uses a cloud manufacturing platform to match the process data with collaborative process data to obtain a collaborative process dataset. Simulated process product requirement parameters are obtained, and standard requirement parameter deviations are calculated to obtain requirement parameter deviation information. Weights are set for the collaborative process dataset based on this deviation. Furthermore, the multi-environment predicted data and the process data are used to form local training data, which is then federated with the collaborative process dataset. Model parameters are optimized according to the weights of the collaborative process dataset to obtain a parameter-optimized federated model. This method determines the process simulation parameter information and solves the technical problem in existing technologies where the methods for obtaining missing data are limited and the acquisition process is not rigorous enough, resulting in insufficient alignment between the final determined data and reality, thus affecting the accuracy of the simulation. This method achieves precise simulation that closely matches the requirements of process simulation. Attached Figure Description
[0010] Figure 1 This application provides a schematic diagram of a process simulation method for cloud manufacturing.
[0011] Figure 2 This application provides a schematic diagram of the collaborative process data acquisition process in a process flow simulation method for cloud manufacturing models.
[0012] Figure 3 This application provides a schematic diagram of the data federated learning process in a process simulation method for cloud manufacturing models;
[0013] Figure 4 This application provides a schematic diagram of a process flow simulation system for cloud manufacturing.
[0014] Figure labeling: a) Data acquisition module, b) Parameter evaluation module, c) Data matching module, d) Parameter acquisition module, e) Parameter calculation module, f) Weight setting module, g) Federated modeling module, h) Model optimization module. Detailed Implementation
[0015] This application provides a process simulation method and system for cloud manufacturing, which acquires process data and determines the acquisition environment parameters, then performs multi-environment parameter evaluation to obtain multi-environment prediction data, utilizes a cloud manufacturing platform to acquire a collaborative process acquisition dataset, obtains the simulated process product requirement parameters and calculates the deviation values of standard requirement parameters, sets the weights of the collaborative process acquisition dataset, builds local training data for federated learning, optimizes model parameters according to the weights of the collaborative process acquisition dataset, and determines the process simulation parameter information. This addresses the technical problem in existing technologies where the methods for acquiring missing parts of the acquired data are limited and the acquisition process is not rigorous enough, resulting in insufficient alignment between the final determined data and reality, thus affecting the accuracy of the simulation.
[0016] Example 1
[0017] like Figure 1 As shown, this application provides a process flow simulation method for cloud manufacturing models, the method comprising:
[0018] Step S100: Collect simulation data of the local process flow according to the process flow simulation parameter requirements to obtain the process flow data;
[0019] Specifically, this application provides a process flow simulation method for cloud manufacturing models. This method acquires process flow data and collaborative process data, filters and matches the collaborative process data based on local product requirements, performs federated learning on a third-party cloud platform to obtain a federated model, optimizes the model, and then simulates the process flow. Simulation data for the local process flow is collected based on process flow simulation parameter requirements. These simulation parameter requirements refer to the relevant parameter indicators needed for production simulation according to the process flow. The local process flow simulation data refers to the usage and generation data during the process design process, which can be divided into two categories: static data and dynamic data. The static data includes processing material data, equipment data, and standard process flow data, while the dynamic data includes intermediate process data and operation data. Collecting relevant data from the local process flow provides the process flow data and a basic information source for subsequent process flow simulation.
[0020] Step S200: Determine the environmental parameters based on the data collected in the process flow, and evaluate multiple environmental parameters based on the collected environmental parameters to obtain multi-environmental prediction data.
[0021] Specifically, based on the process flow data, environmental parameters are determined, the corresponding environment for the process flow data is obtained, and environmental parameters, including ambient temperature and humidity, are extracted. The impact of dynamic changes in these environmental parameters on the process flow parameters is further determined to obtain relevant influencing parameters, such as machine performance and raw material characteristics. Furthermore, the process flow parameters are estimated under multiple scenarios with the entire processing cycle as the time interval. Comparison and analysis with historical experience databases are used to determine the corresponding data fluctuations. Data integration processing is then performed to obtain the multi-environmental prediction data, providing training data for the subsequent construction of the federated model.
[0022] Furthermore, based on the data collected from the aforementioned process flow, environmental parameters are determined, and multi-environmental parameter evaluation is performed based on these parameters to obtain multi-environmental prediction data. Step S200 of this application also includes:
[0023] Step S210: Based on the collected environmental parameters, perform process flow influence relationship analysis to obtain process flow influence parameters;
[0024] Step S220: Perform a full-cycle variation and fluctuation analysis based on the process flow influencing parameters to determine multi-environment prediction parameters, wherein each environmental prediction parameter includes the fluctuation range of the process flow influencing parameters corresponding to each environment;
[0025] Step S230: Compare the fluctuation range of the process flow influencing parameters with the historical processing fluctuation database to estimate the process flow parameter fluctuation data, and use the process flow parameter fluctuation data of each environment to form the multi-environment prediction data collection.
[0026] Specifically, based on the acquired process flow data, environmental parameters are determined. These environmental parameters refer to the external environment where data is currently collected, such as temperature and humidity. A process flow impact analysis is performed on each of these environmental parameters. By defining the influence relationships between parameters, the process parameters affected by changes in the environmental parameters are identified. For example, the impact of temperature and humidity on the machine (heat dissipation at high temperatures and internal liquid condensation at low temperatures), and the impact of high humidity on machine operation smoothness; the impact of raw material characteristics (corrosiveness, liquefaction, vaporization, etc.). The process flow impact parameters are then acquired, and further, a full-cycle variation and fluctuation analysis is performed on these impact parameters. For a complete production cycle, which may be six months, a year, or longer, environmental parameters change in real time throughout the entire cycle. The analysis determines whether and to what extent the corresponding process flow impact parameters are affected.
[0027] Furthermore, based on the currently collected data and the analysis results, the fluctuation range of process flow impact parameters corresponding to each environment is estimated, and the multi-environment estimated parameters are obtained. The fluctuation range of the process flow impact parameters is further compared with the historical processing fluctuation database. By performing a mapping comparison between the two, the possible data changes of the process flow impact parameters are determined, and the estimated process flow parameter fluctuation data is obtained. Based on the entire processing cycle, the fluctuation data of the process flow parameters under multiple environments is estimated, and the multi-environment estimated collection data is obtained. The acquisition of the multi-environment estimated collection data lays a solid foundation for the subsequent construction of the federated model.
[0028] Step S300: Using the cloud manufacturing platform, perform process flow cooperation data matching based on the process flow data to obtain the cooperation process data collection dataset;
[0029] Specifically, during the data acquisition process, some data cannot be collected immediately and cannot be determined, such as the operational output data of each process step. Other scenarios can be predicted based on the relationship between the currently collected data and changes in environmental influences to improve the process flow data. Based on the cloud manufacturing platform, process flow cooperation data matching is performed according to the process flow data to obtain process flows in the same industry and with the same characteristics as the process to be simulated. These are then used as cooperative process flows. Related parameter data is acquired for these cooperative process flows to obtain the cooperative process data set. Using this cooperative process data set as a benchmark, missing parts of the process flow data are supplemented and improved, leading to subsequent analysis and processing, which can effectively improve the accuracy of the analysis results.
[0030] Furthermore, such as Figure 2 As shown, by utilizing a cloud manufacturing platform and matching process flow cooperation data with collected process flow data to obtain a collaborative process data collection dataset, step S300 of this application further includes:
[0031] Step S310: Obtain process flow information and process equipment information;
[0032] Step S320: Based on the process flow information and process equipment information, set the filtering feature relationship and filter in the cloud manufacturing platform database to obtain the first filtering dataset;
[0033] Step S330: Extract features from the collected data according to the process flow to obtain the features of the collected data;
[0034] Step S340: Filter the first filtered dataset using the collected data features to obtain the cooperative process collected dataset.
[0035] Specifically, the process flow information and process equipment information of the simulated process flow are acquired. The process flow information includes a complete process flow diagram and the relationships between each process. The process equipment information includes the process equipment required for each part of the process flow and the corresponding equipment parameter information. Based on the process flow information and the process equipment information, a filtering feature relationship is set. The filtering feature relationship includes the relationship between the features of each process and the relationship between each process flow and the corresponding process equipment. Since some data cannot be collected immediately, the completeness of the acquired process flow data is insufficient. In order to improve the actual fit of the process flow simulation, the actual data can be estimated through similar reference data to improve the acquired process flow data.
[0036] Furthermore, based on the cloud manufacturing platform database, data information is filtered according to the filtering feature relationship to obtain data similar to the process flow information and the process equipment information as the first filtering dataset. Further, the process flow data collection features are extracted, such as equipment usage and raw material attributes. The extracted data features are categorized and integrated. The first filtering dataset is then filtered using these data features as filtering criteria to obtain data similar to the data features as the cooperative process data collection data. The process flow data collection data is then supplemented and improved based on the cooperative process data collection data to enhance information completeness and improve the accuracy of subsequent process flow simulations.
[0037] Step S400: Obtain the product requirement parameters for the simulation process;
[0038] Step S500: Calculate the standard requirement parameter deviation value based on the simulated process product requirement parameters to obtain requirement parameter deviation information;
[0039] Specifically, the required parameters of the simulated process product are acquired. These required parameters refer to the basic indicator data for completing the process, and the parameter ranges of various key product required parameters used in production and process control are determined. For example, for welding processes, the required parameters of the process product include electrode diameter, number of welding layers, power supply type, and welding current. The welding current and number of welding layers required to achieve the expected product vary depending on the welding raw materials or different external environments. Furthermore, the standard required parameters of the simulated process product are acquired based on big data. Using the required parameters of the simulated process product as a benchmark, parameter information is mapped to correspondence between the two. Then, the parameter deviation value of the corresponding required parameters is calculated to obtain the deviation value between the required parameters of the simulated process product and the standard required parameters. This deviation value is used as the required parameter deviation information, where there is a one-to-one correspondence between the required parameters of the simulated process product. The acquisition of the required parameter deviation information provides a basic basis for the matching analysis of the cooperative process data.
[0040] Step S600: Set weights for the cooperative process data set based on the demand parameter deviation information;
[0041] Specifically, by calculating the deviation between the simulated process product demand parameters and the standard demand parameters, the demand parameter deviation information is obtained. Furthermore, the deviation information and the collaborative process data collection dataset are compared with the standard demand parameters to calculate the percentage of parameter deviation. Based on the formula Parameter Deviation Percentage = Demand Parameter Deviation Value / Standard Demand Parameter Value, the percentage of parameter deviation and the percentage of collaborative parameter deviation are obtained. Weights are then set based on the similarity between the two; the higher the similarity, the higher the weight of the collaborative data corresponding to the product demand parameters. This weighting of the collaborative process data collection dataset provides a basis for subsequent model optimization after federated learning.
[0042] Furthermore, based on the aforementioned demand parameter deviation information, the weighting of the collaborative process data acquisition dataset is set. Step S600 of this application also includes:
[0043] Step S610: Based on the demand parameter deviation information, determine the parameter deviation percentage, wherein the parameter deviation percentage = demand parameter deviation value / standard demand parameter value;
[0044] Step S620: Calculate the standard demand deviation value for the collected dataset of the cooperative process to determine the proportion of deviation of cooperative parameters;
[0045] Step S630: Calculate the similarity based on the proportion of deviation of cooperation parameters and the proportion of parameter deviation, and set the weights for the corresponding cooperation process acquisition data in the cooperation process acquisition dataset based on the similarity.
[0046] Specifically, based on the aforementioned demand parameter deviation information, and according to the formula Parameter Deviation Ratio = Demand Parameter Deviation Value / Standard Demand Parameter Value, the parameter deviation ratios of multiple parameters in the simulated process product demand parameters are calculated to determine the parameter deviation ratios. Further, the parameter deviation values of the acquired collaborative process data set and the standard demand parameters are determined. By performing a mapping comparison between the two, the standard demand parameters corresponding to the collaborative process data set are determined. Then, based on the formula Collaborative Parameter Deviation Ratio = Collaborative Parameter Deviation Value / Standard Demand Parameter Value, the collaborative parameter deviation ratio is determined. Based on the deviation ratio and the parameter deviation ratio, the similarity of the corresponding parameter deviation values is calculated, and further similarity analysis is performed to determine the weights. For example, assuming that the local production product parameter requirement is the lowest cost, the data with lower corresponding parameter data in the cooperation data has a higher weight, and the cooperation data is matched according to the lowest cost requirement; assuming that the local production product requirement is high wear and corrosion resistance, the data with the corresponding parameter in the cooperation data has a higher weight, so as to complete the weight setting of the corresponding cooperation process acquisition data in the cooperation process acquisition data. By setting the weights of the cooperation process acquisition data, the matching degree with the requirement parameter information can be effectively improved.
[0047] Step S700: Use the multi-environment prediction data and the process flow data to form local training data, and use the local training data and the cooperative process data dataset to perform federated learning;
[0048] Step S800: Optimize model parameters according to the weights of the dataset collected from the collaborative process, obtain the parameter-optimized federated model, and determine the process simulation parameter information.
[0049] Specifically, the local training data is constructed based on the multi-environment prediction data and the process flow data. A third-party cloud platform is then used as an intermediate auxiliary platform for federated learning. The local training data is transmitted to the third-party cloud platform to determine the standard model for collaborative training in federated learning. Based on the collaborators of the collaborative process flow dataset, multiple collaborative model training data are acquired and sent to the third-party cloud platform. Federated learning is performed based on all received model training parameters to determine the federated model. Federated learning with data from the same industry and with the same characteristics as the process flow to be simulated via the third-party cloud platform effectively improves the reliability of the simulation data. Furthermore, using the weights of the collaborative process flow data as a benchmark, parameter data with high similarity to local product requirements has higher weights to determine the proportion of multiple collaborative data. Based on this benchmark, collaborative data is matched, and training data is selected from the collaborative process flow data for optimized training of the federated model to determine the parameter-optimized federated model. Based on this, the process flow simulation is completed. Multi-sample data matching analysis is performed to improve the reliability and accuracy of the prediction results. Optimization based on weights improves the fit between the parameter-optimized federated model and local simulation requirements.
[0050] Furthermore, such as Figure 3 As shown, using the multi-environment prediction data and the process flow data, a local training data is formed. Federated learning is then performed using the local training data and the collaborative process data dataset. Step S700 of this application further includes:
[0051] Step S710: Train the local parameter evaluation model based on the multi-environment prediction data and the process flow data to obtain the local model training parameters;
[0052] Step S720: Encrypt the local model training parameters and the local parameter evaluation model and send them to the third-party training cloud platform;
[0053] Step S730: Using the third-party training cloud platform, determine the cooperative training standard model based on the local parameter evaluation model, and send the cooperative training standard model to the collaborator determined by the cooperative process data collection dataset;
[0054] Step S740: Each partner uses local collaborative process data to train the collaborative training standard model and obtain collaborative model training parameters;
[0055] Step S750: All partners will encrypt and send the obtained cooperative model training parameters and local cooperative process acquisition data to the third-party training cloud platform. The third-party training cloud platform will perform federated model learning based on all received model parameters and process acquisition data.
[0056] Specifically, the local training data is composed of the multi-environment prediction data and the process flow data. The constructed local parameter evaluation model is trained, and the local model training parameters are obtained. These local model training parameters refer to the training parameters of the process flow corresponding to the local training data. The local model training parameters and the local parameter evaluation model are then encrypted and sent to the third-party training cloud platform. The third-party training cloud platform is an intermediate auxiliary platform that performs comprehensive matching analysis of the local training data and the collaborative training data. Using the local parameter evaluation model as the original model, the third-party training cloud platform is used to determine the collaborative training standard model. This collaborative training standard model refers to a standard model that refines the parameter indicators of the local parameter evaluation model. Multiple partners corresponding to the collaborative process flow data are identified, and the collaborative training standard model is sent to the corresponding partners.
[0057] Furthermore, each partner trains the cooperative training standard model based on the locally collected cooperative process data, obtaining the cooperative model training parameters. These parameters refer to the training parameters for the missing parts of the process flow determined by each partner based on the locally collected cooperative process data. Each partner then encrypts and sends the obtained cooperative model training parameters and the locally collected cooperative process data to the third-party training cloud platform. The third-party training cloud platform performs federated learning on the cooperative training standard model based on all received model parameters and process data to improve the model and obtain the final federated model. The federated model includes the complete process flow and corresponding raw and generated data. Through distributed joint training of the model, the final model data coverage is more comprehensive and its fit with reality is higher.
[0058] Furthermore, the step S800 of this application, which optimizes model parameters according to the weights of the dataset collected through collaborative processes to obtain a parameter-optimized federated model, further includes:
[0059] Step S810: Use the cooperative process to collect dataset weights and determine the proportion of federated learning training values and the proportion of federated model parameters;
[0060] Step S820: Optimize the parameters of the federated model by selecting parameters from the training parameters of the cooperative model according to the proportion of the federated model parameters;
[0061] Step S830: Select federated learning training data from the collaborative process data collection dataset according to the proportion of federated learning training values, perform federated model parameter optimization training based on the selected collaborative model training parameters, and obtain the parameter-optimized federated model.
[0062] Specifically, the federated model is optimized based on the weights of the collaborative process data acquisition dataset. The proportions of the federated learning training values and the federated model parameters are determined using the weights of the collaborative process data acquisition dataset as a benchmark. The proportion of the federated learning training values represents the degree of approximation between each piece of collaborative process data and the local data requirements; the higher the weight, the larger the proportion. The proportion of the federated model parameters represents the weight of missing parameters in the process flow data acquisition. Using the proportion of the federated model parameters as a parameter selection criterion, corresponding parameters are selected from the collaborative model training parameters to optimize the federated model. Furthermore, based on the proportion of the federated learning training values, data is selected proportionally from the collaborative process data acquisition dataset as the federated learning training data. The optimized federated model is then used for model optimization training. The optimized federated model is then used as the parameter-optimized federated model. By selecting the collaborative model training parameters and the federated learning training data for federated model optimization, the final model contains more complete process data and better meets local simulation requirements.
[0063] Furthermore, step S830 of this application also includes:
[0064] Step S831: Optimize the federated model based on the parameters to evaluate the multi-environment prediction data and the process flow data, and obtain the evaluation results of the multi-environment prediction data.
[0065] Step S832: Correct the multi-environment prediction data based on the evaluation results of the multi-environment prediction data, and then use the parameter optimization federated model for evaluation to obtain the multi-environment prediction data whose evaluation results meet the requirements.
[0066] Step S833: Simulate the process flow using the process flow data and multi-environment prediction data.
[0067] Specifically, the initially collected multi-environment prediction data and the process flow data are input into the parameter optimization federated model. The evaluation results of the multi-environment prediction data are obtained through data evaluation. Due to the influence of external environmental factors during data collection, the collected data will inevitably be affected by certain deviations. Using the evaluation results of the multi-environment prediction data as a reference, and based on the correspondence between the multi-environment prediction data and the evaluation results, the multi-environment prediction data is corrected for deviations. Further evaluation is then performed based on the parameter optimization federated model to determine the accuracy of the corrected data, thereby obtaining multi-environment prediction data whose evaluation results meet the requirements. Furthermore, using the process flow data and the multi-environment prediction data as data sources, process flow simulation is performed, which can effectively improve the fit between the process flow simulation results and the actual process flow.
[0068] Example 2
[0069] Based on the same inventive concept as the process simulation method for cloud manufacturing mode in the foregoing embodiments, such as Figure 4 As shown, this application provides a process simulation system for cloud manufacturing, the system comprising:
[0070] Data acquisition module a, which is used to acquire simulation data of the local process flow according to the process flow simulation parameter requirements, and obtain process flow acquisition data;
[0071] Parameter evaluation module b is used to determine the collection environment parameters based on the data collected in the process flow, and to perform multi-environment parameter evaluation based on the collection environment parameters to obtain multi-environment predicted collection data.
[0072] Data matching module c is used to utilize the cloud manufacturing platform to perform process flow cooperation data matching based on process flow data collected to obtain a cooperative process data collection dataset.
[0073] Parameter acquisition module d, which is used to obtain the product requirement parameters of the simulation process;
[0074] The parameter calculation module e is used to calculate the standard requirement parameter deviation value based on the simulation process product requirement parameters, and obtain the requirement parameter deviation information.
[0075] Weight setting module f, which is used to set weights for the cooperative process data collection dataset based on the demand parameter deviation information;
[0076] The federated modeling module g is used to form local training data by utilizing the multi-environment prediction data and the process flow data, and to perform federated learning using the local training data and the cooperative process data dataset.
[0077] The model optimization module h is used to optimize model parameters according to the weights of the collaborative process data set, obtain a parameter-optimized federated model, and determine the process simulation parameter information.
[0078] Furthermore, the system also includes:
[0079] An impact analysis module is used to analyze the impact relationship of the process flow based on the collected environmental parameters, and to obtain the impact parameters of the process flow.
[0080] The parameter determination module is used to perform a full-cycle variation and fluctuation analysis based on the process flow influencing parameters, and to determine multi-environmental prediction parameters, wherein each environmental prediction parameter includes the fluctuation range of the process flow influencing parameters corresponding to each environment.
[0081] The data assembly module is used to compare the fluctuation range of the process flow influencing parameters with the historical processing fluctuation database, estimate the fluctuation data of the process flow parameters, and use the fluctuation data of the process flow parameters in various environments to form the multi-environment prediction and collection data.
[0082] Furthermore, the system also includes:
[0083] A deviation analysis module is used to determine the percentage of parameter deviation based on the deviation information of the demand parameters, wherein the percentage of parameter deviation = deviation value of demand parameters / standard demand parameter value;
[0084] The deviation value calculation module is used to calculate the standard requirement deviation value of the cooperative process data set and determine the deviation ratio of the cooperative parameters.
[0085] The data weight setting module is used to calculate the similarity based on the proportion of deviation of the cooperation parameters and the proportion of parameter deviation, and to set the weight of the corresponding cooperation process acquisition data in the cooperation process acquisition dataset according to the similarity.
[0086] Furthermore, the system also includes:
[0087] Information acquisition module, the information acquisition module is used to acquire process flow information and process equipment information;
[0088] An information filtering module is used to filter the cloud manufacturing platform database based on the process flow information and process equipment information by setting filtering feature relationships to obtain a first filtered dataset.
[0089] The feature extraction module is used to extract features from the data collected according to the process flow to obtain the features of the collected data.
[0090] A data filtering module is used to filter the collected data from the first filtering dataset using the characteristics of the collected data to obtain the cooperative process collected dataset.
[0091] Furthermore, the system also includes:
[0092] The local model training parameter acquisition module is used to perform local parameter evaluation model training based on the multi-environment prediction data and the process flow data to obtain local model training parameters.
[0093] The information sending module is used to encrypt and send the local model training parameters and the local parameter evaluation model to a third-party training cloud platform.
[0094] The model determination and transmission module is used to determine the cooperative training standard model based on the local parameter evaluation model using the third-party training cloud platform, and to transmit the cooperative training standard model to the cooperative party determined by the cooperative process data collection dataset.
[0095] The cooperative model training parameter acquisition module is used by each partner to train the cooperative training standard model using local cooperative process data to obtain cooperative model training parameters.
[0096] The model federated learning module is used by all partners to encrypt and send the obtained cooperative model training parameters and local cooperative process acquisition data to a third-party training cloud platform. The third-party training cloud platform performs model federated learning based on all received model parameters and process acquisition data.
[0097] Furthermore, the system also includes:
[0098] The information proportion determination module is used to determine the proportion of training values and the proportion of parameters of the federated learning model by using the weights of the dataset collected by the cooperative process.
[0099] A model parameter optimization module is used to optimize the parameters of the federated model by selecting parameters from the training parameters of the cooperative model according to the proportion of the federated model parameters.
[0100] The federated model optimization module is used to select federated learning training data from the collaborative process data collection dataset according to the proportion of the federated learning training values, and perform federated model parameter optimization training based on the selected collaborative model training parameters to obtain the parameter-optimized federated model.
[0101] Furthermore, the system also includes:
[0102] The data evaluation module is used to evaluate the multi-environment prediction data and the process flow data based on the optimized federated model according to the parameters, and to obtain the evaluation results of the multi-environment prediction data.
[0103] The model evaluation module is used to correct the multi-environment prediction data based on the evaluation results of the multi-environment prediction data, and then use the parameter optimization federated model for evaluation to obtain the multi-environment prediction data whose evaluation results meet the requirements.
[0104] The process simulation module is used to simulate the process using the process flow data and multi-environment prediction data.
[0105] Through the foregoing detailed description of a process flow simulation method for cloud manufacturing, those skilled in the art can clearly understand the process flow simulation method and system for cloud manufacturing in this embodiment. As for the apparatus disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and relevant parts can be referred to the method section.
[0106] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A process flow simulation method for cloud manufacturing, characterized in that, The method includes: According to the process flow simulation parameter requirements, the local process flow simulation data is collected to obtain the process flow data. Based on the data collected according to the process flow, the environmental parameters are determined, and based on the environmental parameters, multiple environmental parameters are evaluated to obtain multi-environmental prediction data. By using a cloud manufacturing platform, the collaborative process data is matched with the process flow data collected from the process flow to obtain a collaborative process data collection dataset. Obtain the product requirement parameters for the simulation process; Based on the simulated process product requirement parameters, the standard requirement parameter deviation value is calculated to obtain the requirement parameter deviation information; The weights of the collaborative process data collection dataset are set based on the aforementioned demand parameter deviation information. The local training data is composed of the multi-environment prediction data and the process flow data, and federated learning is performed using the local training data and the cooperative process data dataset. The model parameters are optimized by using the weights of the dataset collected from the collaborative process to obtain a parameter-optimized federated model and determine the process simulation parameter information. Using a cloud manufacturing platform, collaborative process data is matched with data collected from the process flow to obtain a collaborative process data set, including: Obtain process flow information and process equipment information; Based on the process flow information and process equipment information, the filtering feature relationship is set and the data is filtered in the cloud manufacturing platform database to obtain the first filtered dataset. Based on the process flow, data features are extracted from the collected data to obtain the collected data features. The cooperative process data collection dataset is obtained by filtering the first filtered dataset using the characteristics of the collected data.
2. The method as described in claim 1, characterized in that, Based on the data collected according to the process flow, environmental parameters are determined, and multi-environmental parameter evaluation is performed based on the collected environmental parameters to obtain multi-environmental prediction data, including: Based on the collected environmental parameters, an analysis of the influence relationship of the process flow is performed to obtain the process flow influencing parameters; Based on the process flow impact parameters, a full-cycle variation and fluctuation analysis is performed to determine multiple environmental prediction parameters. Each environmental prediction parameter includes the fluctuation range of the process flow impact parameters corresponding to each environment. The fluctuation range of the process flow impact parameters is compared with the historical processing fluctuation database to estimate the fluctuation data of the process flow parameters. The fluctuation data of the process flow parameters in each environment are used to form the multi-environment prediction data.
3. The method as described in claim 1, characterized in that, The weighting of the collaborative process data collection dataset is set based on the aforementioned demand parameter deviation information, including: Based on the aforementioned demand parameter deviation information, the percentage of parameter deviation is determined, wherein the percentage of parameter deviation = demand parameter deviation value / standard demand parameter value; The standard requirement deviation value is calculated for the dataset collected from the cooperative process to determine the proportion of deviation in cooperative parameters; Based on the proportion of deviation of the cooperation parameters and the proportion of parameter deviation, the similarity is calculated, and the weights are set for the corresponding cooperation process acquisition data in the cooperation process acquisition dataset according to the similarity.
4. The method as described in claim 1, characterized in that, Using the multi-environment prediction data and the process flow data, local training data is constructed. Federated learning is then performed using the local training data and the collaborative process data dataset, including: Based on the multi-environment prediction data and the process flow data, a local parameter evaluation model is trained to obtain local model training parameters. The local model training parameters and the local parameter evaluation model are encrypted and sent to a third-party training cloud platform. Using the third-party training cloud platform, a collaborative training standard model is determined based on the local parameter evaluation model, and the collaborative training standard model is sent to the collaborators determined by the collaborative process data collection dataset; Each partner uses local collaborative process data to train the collaborative training standard model and obtain collaborative model training parameters; All partners will encrypt and send the obtained collaborative model training parameters and local collaborative process data to a third-party training cloud platform. The third-party training cloud platform will then perform federated model learning based on all received model parameters and process data.
5. The method as described in claim 4, characterized in that, The step of optimizing model parameters according to the weights of the dataset collected through collaborative processes to obtain a parameter-optimized federated model includes: The weights of the dataset collected using the aforementioned collaborative process are used to determine the proportion of numerical values in federated learning training and the proportion of parameters in the federated model. Based on the proportion of parameters in the federated model, parameters are selected from the training parameters of the cooperative model to optimize the parameters of the federated model. According to the proportion of the federated learning training values, federated learning training data is selected from the dataset collected by the collaborative process. Based on the selected collaborative model training parameters, federated model parameter optimization training is performed to obtain the parameter-optimized federated model.
6. The method as described in claim 5, characterized in that, The method includes: The federated model is optimized based on the parameters to evaluate the multi-environmental prediction data and the process flow data, and the evaluation results of the multi-environmental prediction data are obtained. Based on the evaluation results of the multi-environment prediction data, the multi-environment prediction data is corrected, and then the parameter optimization federated model is used for evaluation to obtain the multi-environment prediction data whose evaluation results meet the requirements. The process flow is simulated using the data collected from the aforementioned process flow and the data collected from multiple environmental predictions.
7. A process simulation system for cloud manufacturing, characterized in that, The system includes: The data acquisition module is used to acquire simulation data of the local process flow according to the process flow simulation parameter requirements, and obtain process flow acquisition data. The parameter evaluation module is used to determine the collection environment parameters based on the data collected in the process flow, and to perform multi-environment parameter evaluation based on the collection environment parameters to obtain multi-environment predicted collection data. The data matching module is used to utilize the cloud manufacturing platform to perform process flow cooperation data matching based on the process flow data collected, and obtain a cooperative process collection dataset. A parameter acquisition module is used to obtain the product requirement parameters of the simulation process. The parameter calculation module is used to calculate the standard requirement parameter deviation value based on the simulation process product requirement parameters, and obtain the requirement parameter deviation information. A weight setting module is used to set weights for the cooperative process data collection dataset based on the demand parameter deviation information. The federated modeling module is used to form local training data by utilizing the multi-environment prediction data and the process flow data, and to perform federated learning using the local training data and the cooperative process data dataset. The model optimization module is used to optimize model parameters according to the weights of the collaborative process data collection dataset, obtain a parameter-optimized federated model, and determine the process simulation parameter information. Information acquisition module, the information acquisition module is used to acquire process flow information and process equipment information; An information filtering module is used to filter the cloud manufacturing platform database based on the process flow information and process equipment information by setting filtering feature relationships to obtain a first filtering dataset. The feature extraction module is used to extract features from the data collected according to the process flow to obtain the features of the collected data. A data filtering module is used to filter the collected data from the first filtering dataset using the characteristics of the collected data to obtain the cooperative process collected dataset.