Digital twin workshop management and control method and system based on data multi-layer fusion

Abstract

The invention relates to a digital twin workshop management and control method and system based on data multilayer fusion, and belongs to the field of service-oriented industrial system data science, and the method comprises the steps: obtaining the global multi-source heterogeneous data of a digital twin workshop, and carrying out the data classification, data processing and data storage of the global multi-source heterogeneous data, carrying out data-level fusion to obtain digital twin workshop standard data; extracting standard data of the digital twin workshop, performing semantic association, feature extraction and feature analysis on the standard data of the digital twin workshop, and performing feature level fusion to obtain key state features of the digital twin workshop; the method comprises the following steps: constructing a decision-making model oriented to a digital twin workshop, inputting key operation features of the digital twin workshop, performing decision-making level fusion based on the decision-making model, decision-making optimization and decision-making method solution, and outputting a workshop production management and control decision. The method can provide support for digital twin workshop operation data governance analysis and production management and control accurate decision.

Description

Technical Field

[0001] The present invention belongs to the field of service-oriented industrial system data science, and specifically relates to a digital twin workshop management and control method and system based on multi-layer data fusion. Background Art

[0002] With the continuous development of modern manufacturing, workshop production environments are becoming increasingly complex, especially in discrete production plants. The uncertainty and dynamism of the production process require companies to possess greater production flexibility and control precision. Traditional workshop management models suffer from data silos. Data such as equipment status, production plans, and material flow are scattered across independent systems, and format differences make data integration difficult. Furthermore, the production patterns implicit in massive amounts of data are not fully explored, making it difficult to support accurate decision-making. Static scheduling strategies based on empirical rules cannot adapt to real-time dynamic production changes. To address these issues, academia and industry have recently proposed a variety of intelligent manufacturing and industrial Internet solutions. These solutions incorporate technologies such as big data, cloud computing, and the Internet of Things (IoT) to achieve data integration and intelligent control in workshop management. Digital twins are a new paradigm for integrated workshop data control. By creating a virtual mirror image of the physical workshop, digital twins enable real-time mapping of the production process and optimized decision-making. However, existing digital twin workshop data fusion systems still have the following limitations: First, a data fusion paradigm is missing. There is a lack of unified representation and processing methods for sensor data at the physical layer and simulation data at the virtual layer, resulting in low efficiency in digital twin virtual-real collaboration. Second, the focus is too much on physical domain feature extraction, ignoring multi-domain virtual-real data mining, making it difficult to extract high-value feature information. Third, the optimization solutions generated in the virtual environment lack feedback on physical execution effects, making it difficult to form a continuously improving twin optimization mechanism and unable to flexibly adjust according to the production conditions of the physical workshop. These problems collectively lead to bottlenecks in data management and decision-making in digital twin workshops. Summary of the Invention

[0003] To address the above-mentioned issues, the present invention proposes a digital twin workshop management and control method and system based on multi-layer data fusion, aiming to enhance the data management and control capabilities of the digital twin workshop production process through multi-level data fusion. Effectively designed data-level, feature-level, and decision-level fusion methods for the digital twin workshop excavate the application value of the multi-source heterogeneous data of the digital twin workshop and achieve accurate and reliable decision support. Through the above-mentioned technical means, the present invention realizes data management and control of the entire process and entire field of the digital twin workshop, providing effective technical support for the multi-source heterogeneous data management and intelligent dynamic production decision-making goals of the digital twin workshop.

[0004] The technical solution adopted by the present invention to solve the above technical problems is:

[0005] A digital twin workshop management and control method based on multi-layer data fusion includes the following steps:

[0006] Step 1: Obtain the global multi-source heterogeneous data of the digital twin workshop, classify, process, and store the data, and perform data-level fusion to obtain the standard data of the digital twin workshop;

[0007] Step 2: Extract the standard data of the digital twin workshop, perform semantic association, feature extraction, and feature analysis on the standard data of the digital twin workshop, and perform feature-level fusion to obtain the key status features of the digital twin workshop;

[0008] Step 3: Build a decision-making model for the digital twin workshop, input the key state characteristics of the digital twin workshop, perform decision-level fusion based on the decision model, decision optimization, and decision method, and output the workshop production control decision.

[0009] A digital twin workshop management and control system based on multi-layer data fusion, including:

[0010] The data-level fusion module inputs the multi-source heterogeneous data of the digital twin workshop and outputs the standard data of the digital twin workshop, including:

[0011] A data classification module, used for globally acquiring and classifying entries of the multi-source heterogeneous data set of the digital twin workshop;

[0012] The data processing module is used to pre-process the data and remove noise and duplicate values ​​in the data;

[0013] The data storage module is used to store data in a standardized and hierarchical manner to obtain standardized digital twin workshop data;

[0014] The feature-level fusion module inputs the standard data of the digital twin workshop and outputs the key status features of the digital twin workshop, including:

[0015] The data association module is used to semantically associate virtual and real data of heterogeneous models of the digital twin workshop based on the ontology model, and build a global view of the digital twin workshop data;

[0016] Feature extraction module, used to extract and characterize multi-field features of the global data of the digital twin workshop;

[0017] Feature analysis module, used to mine the key features of the digital twin workshop, analyze and characterize the workshop's operating status;

[0018] The decision-level fusion module inputs the key status characteristics of the digital twin workshop and outputs workshop production control decisions, including:

[0019] Decision model module: used to build a model related to digital twin workshop decision making, which serves as the input of decision objects;

[0020] Decision optimization module, used to define decision constraints, decision conditions, decision goals, and optimize decision models;

[0021] The decision-making method solving module is used to select appropriate decision-making methods, analyze and solve decision-making problems, and obtain production control decisions.

[0022] The present invention has the following beneficial effects:

[0023] 1. Data management and control level: By designing a standardized modeling and representation method for workshop-wide data, data-level integration is achieved, and data management and control of the entire process and entire field of the digital twin workshop are realized, providing a reliable data foundation for subsequent analysis and decision-making.

[0024] 2. Analytical decision-making level: Design multi-dimensional data feature mining methods to achieve feature-level fusion and further explore the value of data; and design decision-making optimization methods for digital twin workshop data enhancement based on effective features to achieve decision-level fusion, provide accurate decision-making for production control, and help meet the 3C digital twin workshop status prediction needs and production scheduling goals.

[0025] 3. Production efficiency level: It provides a theoretical basis for enhancing the production control efficiency of digital twin workshops, helps to improve the flexibility and accuracy of production, improves the production efficiency of workshops, meets the needs of discrete production models, has high practical application value, and promotes the application of digital twin technology in the 3C industry and the development of new quality productivity. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 This is a step diagram of a digital twin workshop management and control method based on multi-layer data fusion of the present invention;

[0027] Figure 2 This is a module relationship diagram of a digital twin workshop management and control system based on multi-layer data fusion of the present invention. DETAILED DESCRIPTION

[0028] In order to make the objectives, technical solutions, and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are intended only to illustrate the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other. To achieve the above-mentioned objectives, the present invention adopts the following technical solutions.

[0029] The present invention uses a 3C workshop manufacturing system as an example to illustrate the present invention's method. The 3C workshop manufacturing system has uncertainties in the production process and requires higher control precision. The 3C workshop model has multi-level coupling, discrete flexibility, and internal data from multiple heterogeneous sources.

[0030] like Figure 1 As shown, the technical solution adopted by the present invention is: a digital twin workshop management and control method based on multi-layer data fusion, comprising the following steps:

[0031] Step 1: Acquire the global multi-source heterogeneous data of the digital twin workshop, synchronize it to the digital twin workshop virtual model, and establish a physical-virtual data mapping relationship. Based on the characteristics of the global multi-source heterogeneous data and the complex system hierarchy of the digital twin workshop, sort out the internal elements and data content of the digital twin workshop and classify the data; then perform preliminary preprocessing on the data to filter out noise and duplicate values; finally, store the data in a standard computer format to obtain standard digital twin workshop data. Based on data classification, data processing, and data storage, perform data-level fusion;

[0032] Step 2: Extract standard data from the digital twin workshop, perform heterogeneous data semantic association on the physical entities, virtual entities, and interactive relationships of the digital twin workshop, and construct global data for the digital twin workshop; extract multi-field features from the global data of the digital twin workshop; combine deep learning algorithms to perform feature analysis on the temperature, speed, energy consumption of the equipment, as well as the time, accuracy, and material consumption of the product processing in the multi-field features, and obtain key state features of the digital twin workshop through time series prediction, such as equipment failure warning time, equipment maintenance time, and key process parameter deviations that cause product quality fluctuations. Based on data semantic association, feature extraction, and feature analysis, perform feature-level fusion;

[0033] Step 3: Construct a decision model for the digital twin workshop, which includes the parameterized decision indicators of various variables required for decision-making. These variables are the key state characteristics of the digital twin workshop obtained in step 2, such as equipment failure warning time, equipment maintenance time, etc.; select clear decisions to solve the problem, define physical constraints, twin constraints, decision conditions, and decision goals for decision optimization; select the corresponding decision method to achieve digital twin enhanced decision results. Based on the decision model, decision optimization, and decision method solution, perform decision-level fusion and output the workshop production control decision;

[0034] The detailed process of data-level fusion is as follows:

[0035] Acquire multi-source heterogeneous data from the entire digital twin workshop, including: Acquire all production data from the digital twin workshop , analyze the data content composition of the digital twin workshop, the data sources of the 3C physical workshop include the equipment management system (DMS), material management system (WMS), environmental management system (EMS), public network (PN), enterprise resource planning (ERP), and manufacturing execution management system (MES). Among them, production data Including data objects, data reality, data status, data structure, the above data are classified in detail according to the corresponding attributes of the data to obtain the global data composition ,in Indicates the twin data domain type, domain, representing twin workshop data, Domain, representing physical shop floor data. is the number of system categories from which the data originates, For the a data entry within an information system;

[0036] Data preprocessing includes: deduplication and denoising of the collected digital twin workshop data to obtain high-quality twin data. The specific steps include:

[0037] Data interpolation: Using regression interpolation method, the missing attribute is used as the dependent variable and other attributes are used as independent variables. A multiple linear regression model of the missing attribute and other attributes is established. The model is expressed as:

[0038] ,

[0039] in, is the estimated missing attribute value, such as product defect rate, is the estimated value of the multiple regression coefficient, Attributes related to the missing attributes, such as product size, product temperature, etc.

[0040] Data deduplication: Check data Check whether the primary key ID is identical. In workshops, the primary key ID is often the product number. Use the latest timestamp priority strategy to find the data where the update time field is a timestamp. Keep the record with the largest timestamp value, delete duplicate records, and only keep the latest version.

[0041] Data denoising: For single time series data of equipment vibration, temperature and other attributes, a moving window averaging denoising method is used with a window size of 9. The average of the data within the window is taken to replace the original value at the center of the original window to obtain smoothed data. The formula is as follows:

[0042] ,

[0043] in, for Moment data time point, is the sliding window size.

[0044] Data storage includes: sorting out the data-feature-decision information processing chain of the digital twin workshop for the heterogeneous data of the physical entity system (such as sensor group, AGV logistics system), virtual model system (such as process simulation module, digital twin), and business management system (such as MES, ERP) in the digital twin workshop, and using XML text language as the expression of data entity relationship. Constructing data mapping relationship is divided into four levels: Concept layer , direct object layer , indirect object layer and attribute layer . Among them, the concept layer is the abstract concept integration of the workshop's own resources, workshop inputs and outputs, and workshop production events in the 3C workshop manufacturing process; the direct object layer is the general model obtained by integrating the production process objects, which characterizes the common elements of the workshop and forms a general object description of the production process data-bearing subject; the indirect object layer is a detailed extension of the direct object. The objects here are the actual objects in the physical workshop and are also the instantiation results of the general objects of the production process; the attribute layer is the specific attributes clustered with the indirect objects, which is the basic unit of data collection in the physical workshop. Finally, for the basic unit of data collection, the XMI format is used to store data as XML text format data as a computer-recognizable data language.

[0045] The detailed process of feature-level fusion is as follows:

[0046] Data semantic association: Define the multi-dimensional context of heterogeneous data and divide workshop data into three categories: physical domain (such as real-time vibration data), virtual domain (such as simulation prediction value), and interactive domain (such as OPC UA communication process data). Label the source domain for data in different fields , collection timestamp , spatial location , semantic tags Read the acquisition timestamp, spatial location, and semantic tag information of the physical and virtual domain data, and use the OPC UA communication protocol to establish a unified measurement value for data in different domains to achieve semantic association of data in different domains. Align the scales of the physical and virtual domains based on the unified measurement value to ensure that the two types of data are comparable under the same dimension. For example, the real-time vibration data of the mobile phone production line and the vibration data of the virtual workshop simulation. Finally, the concept layer in step 1 is aligned. , direct object layer , indirect object layer and attribute layer Update all XML text data under the digital twin workshop to ensure the relevance and consistency of data in different fields;

[0047] Feature extraction: Multi-domain feature extraction of global data of the digital twin workshop, including:

[0048] ① Time domain feature extraction: Taking the moving parts of the placement machine in the SMT production line as the extraction object, the conventional indicators of the time series data are extracted, including maximum value, maximum absolute value, minimum value, mean value, peak-to-peak value, absolute mean value, root mean square value, root mean square amplitude, standard deviation, kurtosis, skewness, margin index, waveform index, pulse index, peak index, and kurtosis index, a total of 16 time domain features;

[0049] ② Frequency domain feature extraction: Through fast Fourier transform, the equipment vibration signal is converted into frequency domain representation, including the amplitude and phase information of all frequency components. The fast Fourier transform formula is as follows:

[0050] ,

[0051] in, for Moment data time point, is the number of sampling points, For the The complex value of the frequency domain components, is a complex exponential rotation on the unit circle. Extract the conventional frequency domain indicators, including the center of gravity frequency, frequency root mean square, average frequency, and frequency variance, a total of 4 frequency domain features;

[0052] Feature analysis:

[0053] ① Data feature training: A data feature analysis model based on CNN-BiGRU-Attention is constructed based on the 16 extracted time-domain features and 4 frequency-domain features of device vibration, with the dependent variable being the device's remaining life. First, the original feature set is divided into a training set and a test set. Next, the feature analysis model is constructed, consisting of a one-dimensional convolution module, a bidirectional gated recurrent network module, and an attention module. The one-dimensional convolution module convolves and pools the time-domain features of the one-dimensional signal with the transformed frequency-domain features to extract global features. The bidirectional gated recurrent network module simultaneously processes the positive and negative relationships between these features in the time domain, fully leveraging the sequential information within the features to capture long-term trends in the data, such as regular fluctuations and degradation. The attention module identifies and prioritizes features among the 16 time-domain and 4 frequency-domain features that have the greatest impact on the dependent variable, ensuring effective feature selection and accurate extraction of key state features. Model initialization parameters, such as the convolution kernel size, convolution function, number of hidden layers, virtual-real collaborative loss function, and number of iterations, are set, and the feature analysis model is trained to obtain the initial training model.

[0054] ② Data feature prediction: Select a test set, make predictions on the initial training model, output feature prediction results, and predict the loss function based on the deviation between the predicted value and the actual value. , as shown below.

[0055] ,

[0056] in, for The actual value of the moment data at a certain time point, for The time point of the predicted value of the moment data, is the number of sampling points. Based on the loss function calculation results and the original model parameters, the prediction error of the virtual model is fine-tuned and corrected using real-time feedback from physical sensors. The training calculation process and parameter update process are repeated until the preset number of iterations is reached, resulting in a trained feature analysis model. Finally, the trained feature analysis model is applied to the test set and the test results are output.

[0057] The detailed process of decision-level fusion is as follows:

[0058] Decision model construction: Consider various variables in the production process, including the equipment status of the 3C physical workshop, production tasks, resource consumption, personnel scheduling and simulation model parameters of the virtual workshop, virtual-real synchronization delay, etc., describe the operation status, virtual-real operation deviation and decision goals in the digital twin workshop, and build a decision model for the production of the digital twin workshop. It is known that there are 10 production tasks in the 3C physical workshop To be processed, the workshop includes 6 processing equipment , each task has an indefinite processing step , for the The first step required for task processing Each process has different processing time on different equipment ,in The virtual workshop production parameters are defined through the above physical workshop, and the corresponding virtual workshop decision model is constructed. The decision content of this model is to allocate different machines to different task processes according to the production task requirements to form a production plan.

[0059] Decision optimization: Define constraints, decision goals, and decision conditions based on the decision model. Constraints are constraints that must be met by the operation of the workshop itself, including the limited production resources of the physical workshop, production capacity restrictions, and simulation model restrictions of the virtual workshop, such as the number of placement machine heads and placement pressure prediction errors. The decision goal is to minimize the total production time of the production plan. Decision conditions are conditions that must be met for workshop decisions, such as the same equipment can only process one workpiece at a time. Parameter optimization and improvement are performed based on the decision model;

[0060] Decision-making method solution: The genetic algorithm in the meta-heuristic algorithm is selected to solve the decision-making method of the digital twin workshop. First, the virtual workshop simulation data, including the workshop equipment status, processing time, etc., is introduced to determine the multiple processes of the digital twin workshop, the process of each process, and the corresponding processing equipment. The chromosomes in the genetic algorithm are defined as a double gene string of process code and equipment processing time code, and the population is the decision solution set. First, an initial decision solution set with a population size of 100 is generated, and the completion time of these solution sets is calculated. , the calculation formula is as follows:

[0061] ,

[0062] For each decision-making individual in the parent population, perform crossover and mutation operations in the genetic algorithm to generate a new child population, and also calculate the completion time. , select the parent-child population The smallest 100 individuals form a new generation. The above is considered as one iteration. Repeat 100 generations to get the final population. The smallest solution is taken as the final decision plan. The decision plan is sent to the physical workshop for decision execution, and the actual completion time of the physical workshop is obtained. The sensor feeds back the decision results to the virtual workshop in real time, and the actual execution effect of the physical workshop is determined. Fine-tune the virtual shop floor model until it is operational on the physical shop floor Equal to the calculation of the virtual workshop The above steps enable digital twin workshop management and control.

[0063] like Figure 2 As shown, the present invention provides a digital twin workshop management and control system based on multi-layer data fusion, including:

[0064] (1) Data-level fusion module: inputs multi-source heterogeneous data of the digital twin workshop and outputs clean, standardized and mineable data of the workshop. This includes:

[0065] A data classification module, used for globally acquiring and classifying entries of the multi-source heterogeneous data set of the digital twin workshop;

[0066] The data processing module is used to pre-process the data and remove noise and duplicate values ​​in the data;

[0067] The data storage module is used to store data in a standardized and hierarchical manner to obtain standardized digital twin workshop data;

[0068] (2) Feature-level fusion module: inputs the value data to be mined and outputs the key state features of the digital twin workshop. This includes:

[0069] The data association module is used to semantically associate virtual and real data of heterogeneous models of the digital twin workshop based on the ontology model, and build a global view of the digital twin workshop data;

[0070] Feature extraction module, used to extract and characterize multi-field features of the global data of the digital twin workshop;

[0071] Feature analysis module, used to mine the key features of the digital twin workshop, analyze and characterize the workshop's operating status;

[0072] (3) Decision-making level fusion module, which inputs the key operational characteristics of the digital twin workshop and outputs the workshop production control decision. This includes:

[0073] The decision model module is used to build a model related to digital twin workshop decision making as the input of the decision object;

[0074] Decision optimization module, used to define decision constraints, decision conditions, decision goals, and optimize decision models;

[0075] The decision-making method solving module is used to select appropriate decision-making methods, analyze and solve decision-making problems, and obtain production control decisions.

Claims

1. A digital twin workshop management and control method based on multi-layer data fusion, characterized in that: The following steps are involved: Step 1: Obtain the global multi-source heterogeneous data of the digital twin workshop, classify, process, and store the data, and perform data-level fusion to obtain the standard data of the digital twin workshop; Step 2: Extract the standard data of the digital twin workshop, perform semantic association, feature extraction, and feature analysis on the standard data of the digital twin workshop, and perform feature-level fusion to obtain the key status features of the digital twin workshop; Step 3: Build a decision-making model for the digital twin workshop, input the key state characteristics of the digital twin workshop, perform decision-level fusion based on the decision model, decision optimization, and decision method, and output the workshop production control decision.

2. A digital twin workshop management and control method based on multi-layer data fusion according to claim 1, characterized in that: In step 1, obtain the global multi-source heterogeneous data of the digital twin workshop, including: obtaining all the production data of the digital twin workshop , for the above production data Detailed classification is performed according to the corresponding attributes of the data to obtain global multi-source heterogeneous data ,in Indicates the twin data domain type, domain, representing twin workshop data, domain, representing physical shop floor data, is the number of system categories from which the data originates, For the A data entry within an information system.

3. The digital twin workshop management and control method based on multi-layer data fusion according to claim 1 is characterized in that: In step 1, data processing includes: deduplication and denoising of the collected multi-source heterogeneous data of the digital twin workshop. The specific steps include: Data interpolation: Using regression interpolation method, the missing attribute is used as the dependent variable and other attributes are used as independent variables. A multiple linear regression model of the missing attribute and other attributes is established. The model is expressed as: ， in, is the estimated missing attribute value, such as product defect rate, is the estimated value of the multiple regression coefficient, Attributes related to the missing attributes, including product size and product temperature; Data deduplication: Check data Check whether the primary key ID is the same. In the workshop, the primary key ID is the product number. Use the latest timestamp priority strategy to find the data with the update time field as the timestamp. Keep the record with the largest timestamp value, delete the duplicate records and keep only the latest version. Data denoising: For a single time series data, a moving window averaging denoising method is used. The average of the data in the window is taken to replace the original value at the center of the original window to obtain smoothed data. The formula is as follows: ， in, for Moment data time point, is the sliding window size.

4. The digital twin workshop management and control method based on multi-layer data fusion according to claim 1 is characterized in that: In step 1, data storage includes: sorting out the digital twin workshop data-feature-decision information processing chain for the heterogeneous data of the physical entity system, virtual model system, and business management system in the digital twin workshop, using XML text language as the expression of data entity relationship, and building data mapping relationship, which is divided into four levels: concept layer , direct object layer , indirect object layer and attribute layer .

5. The digital twin workshop management and control method based on multi-layer data fusion according to claim 1 is characterized in that: In step 2, semantic association includes: defining the multi-dimensional context of heterogeneous data, dividing workshop data into three categories: physical domain, virtual domain, and interactive domain, and marking the source domain for different domain data. , collection timestamp , spatial location , semantic tags Each data information reads the collection timestamp, spatial location, and semantic label of the physical domain and virtual domain data, uses the communication protocol to establish a unified measurement value for data in different domains, and realizes the semantic association of data in different domains.

6. The digital twin workshop management and control method based on multi-layer data fusion according to claim 1 is characterized in that: In step 2, feature extraction, including: ① Time domain feature extraction: extract indicators of time series data, including maximum value, maximum absolute value, minimum value, mean value, peak-to-peak value, absolute mean value, root mean square value, root mean square amplitude, standard deviation, kurtosis, skewness, margin index, waveform index, pulse index, peak index, and kurtosis index; ② Frequency domain feature extraction: Through fast Fourier transform, the signal is converted into frequency domain representation, including the amplitude and phase information of all frequency components. The fast Fourier transform formula is as follows: ， in, for Moment data time point, is the number of sampling points, For the The complex value of the frequency domain components, is a complex exponential rotation on the unit circle, obtained from the calculation Frequency domain indicators are extracted from the data, including center of gravity frequency, frequency root mean square, average frequency, and frequency variance.

7. The digital twin workshop management and control method based on multi-layer data fusion according to claim 6 is characterized in that: In step 2, feature analysis, includes: ① Data feature training: extract the time domain features and frequency domain features, take the dependent variables, and build a feature analysis model; first, divide the original feature set into a training set and a test set, and then build a feature analysis model. The model includes a one-dimensional convolution module, a bidirectional gated recurrent network module, and an attention module. The one-dimensional convolution module convolves and pools the time domain features of the one-dimensional signal itself with the transformed frequency domain features to extract global features; the bidirectional gated recurrent network module simultaneously processes the positive and reverse relationships of the above features in the time domain, and uses the before and after time sequence information in the features to capture the long-term trend of the data; the attention module distinguishes and strengthens the features in the time domain and frequency domain features that have a greater impact on the dependent variable, sets the convolution kernel size, convolution function, number of hidden layers, virtual-real collaborative loss function, and number of iterations, and trains the feature analysis model to obtain the initial training model; ② Data feature prediction: Select a test set, make predictions on the initial training model, output feature prediction results, and predict the loss function based on the deviation between the predicted value and the actual value. , as shown below: ， in, for The actual value of the moment data at a certain time point, for The time point of the predicted value of the moment data, is the number of sampling points; according to the calculation results of the loss function and the original model parameters, the prediction error of the virtual model is corrected according to the real-time feedback to make prediction corrections.

8. The digital twin workshop management and control method based on multi-layer data fusion according to claim 1 is characterized in that: In step 3, the decision model is constructed, including: considering various variables in the production process, describing the operating status, virtual and real operation deviations and decision goals in the digital twin workshop, and building a decision model for the production of the digital twin workshop.

9. The digital twin workshop management and control method based on multi-layer data fusion according to claim 1 is characterized in that: In step 3, decision optimization includes: defining constraints, decision goals and decision conditions according to the decision model. Constraints are constraints that the workshop operation itself must meet. The decision goal is to minimize the total production time of the production plan. Decision conditions are conditions that the workshop decision must meet. Parameter optimization and improvement are performed based on the decision model.

10. The digital twin workshop management and control method based on multi-layer data fusion according to claim 1 is characterized in that: In step 3, the decision-making method is solved, including: selecting the genetic algorithm in the meta-heuristic algorithm to solve the decision-making method of the digital twin workshop, first introducing the virtual workshop simulation data, including the workshop equipment status and processing time, determining multiple processes of the digital twin workshop, the process of each process and the corresponding processing equipment, defining the chromosomes in the genetic algorithm as a double gene string of process code and equipment processing time code, and the population as the decision solution set. First, the initial decision solution set is calculated for these solution sets. ; For each decision-making individual in the parent population, perform crossover and mutation operations in the genetic algorithm to generate a new child population, and also calculate the completion time. .

11. A digital twin workshop management and control method system based on multi-layer data fusion, characterized in that: include: The data-level fusion module inputs the multi-source heterogeneous data of the digital twin workshop and outputs the standard data of the digital twin workshop, including: A data classification module, used for globally acquiring and classifying entries of the multi-source heterogeneous data set of the digital twin workshop; The data processing module is used to pre-process the data and remove noise and duplicate values ​​in the data; The data storage module is used to store data in a standardized and hierarchical manner to obtain standardized digital twin workshop data; The feature-level fusion module inputs the standard data of the digital twin workshop and outputs the key status features of the digital twin workshop, including: The data association module is used to semantically associate virtual and real data of heterogeneous models of the digital twin workshop based on the ontology model, and build a global view of the digital twin workshop data; Feature extraction module, used to extract and characterize multi-field features of the global data of the digital twin workshop; Feature analysis module, used to mine the key features of the digital twin workshop, analyze and characterize the workshop's operating status; The decision-level fusion module inputs the key status characteristics of the digital twin workshop and outputs workshop production control decisions, including: Decision model module: used to build a model related to digital twin workshop decision making, which serves as the input of decision objects; Decision optimization module, used to define decision constraints, decision conditions, decision goals, and optimize decision models; The decision-making method solving module is used to select appropriate decision-making methods, analyze and solve decision-making problems, and obtain production control decisions.

Images