WO2023193458A1

WO2023193458A1 - Digital twin-based production line optimization method and apparatus, electronic device, and medium

Info

Publication number: WO2023193458A1
Application number: PCT/CN2022/136469
Authority: WO
Inventors: 陈录城; 诸葛慧玲; 张成龙; 周靖超; 王勇; 孟祥秀; 李晓璐
Original assignee: 卡奥斯工业智能研究院(青岛)有限公司; 海尔数字科技(青岛)有限公司; 海尔卡奥斯物联生态科技有限公司; 海尔智家股份有限公司
Priority date: 2022-04-07
Filing date: 2022-12-05
Publication date: 2023-10-12
Also published as: CN114493049A

Abstract

The present application provides a digital twin-based production line optimization method and apparatus, an electronic device, and a medium. The digital twin-based production line optimization method comprises: obtaining historical production data of a factory, and training a factory dynamic model according to the historical production data to obtain a virtual data model; inputting test production data of the factory into the virtual data model to obtain a prediction result corresponding to the test production data; determining a predicted benefit value corresponding to the prediction result, and determining whether the predicted benefit value meets an effective condition; and when the predicted benefit value meets the effective condition, mapping production parameters corresponding to the virtual data model into a production line of the factory so as to optimize data of the production line of the factory.

Description

Production line optimization methods, devices, electronic equipment and media based on digital twins

This application claims priority to the Chinese patent application with application number 202210358331.7, which was submitted to the China Patent Office on April 7, 2022. The entire content of this application is incorporated into this application by reference.

Technical field

This application relates to computer technology, for example, to production line optimization methods, devices, electronic equipment and media based on digital twins.

Background technique

With the sales of household appliances, the status of washing machines in household appliances is increasing day by day, becoming a necessary appliance, and the manufacturing of washing machines has become a major factory industry in the industry. In the production process of washing machines, since there are many types, forms and quantities of production materials, the control of production material costs directly determines the company's time competitiveness. Big data companies only use enterprise management systems and industrial Internet systems to monitor production materials, using barcodes and data collection methods to track production materials. Most of these systems present data on the entry and exit, inventory, production and location of production materials. Use data analysis to control and optimize the consumption, storage locations and surplus of production materials throughout the production process. Although it is possible to optimize the use of production materials and reduce the production cost of the production line to a certain extent, it lacks deep embedding of the use of materials throughout the entire life cycle of the production line, and cannot connect and correlate the production data of each production process of the production line. At the same time, it cannot dynamically adapt to the changing production environment, provide real-time decision-making basis to optimize the use of production line production materials and production process flow, and use data analysis to intuitively display the operating status of the production line and the use of materials.

Contents of the invention

This application provides production line optimization methods, devices, electronic equipment and media based on digital twins to realize the optimization of factory production lines using production data, which can more accurately and effectively improve material utilization and product quality, and improve the overall efficiency of the factory. income.

In the first aspect, this application provides a production line optimization method based on digital twins, which includes:

Obtain the historical production data of the factory, train the factory dynamic model based on the historical production data, and obtain a virtual data model;

Input the test production data of the factory into the virtual data model to obtain prediction results corresponding to the test production data;

Determine the predicted benefit value corresponding to the prediction result, and determine whether the predicted benefit value meets the valid conditions;

When the predicted benefit value meets the valid conditions, the production parameters corresponding to the virtual data model are mapped to the production line of the factory to perform data optimization on the production line of the factory.

In the second aspect, this application also provides a production line optimization device based on digital twins, which includes:

The model training module is configured to obtain the historical production data of the factory, train the factory dynamic model based on the historical production data, and obtain a virtual data model;

The prediction acquisition module is configured to input the test production data of the factory into the virtual data model to obtain prediction results corresponding to the test production data;

A valid judgment module, configured to determine the predicted benefit value corresponding to the prediction result, and determine whether the predicted benefit value meets the valid conditions;

The production line optimization module is configured to map the production parameters corresponding to the virtual data model to the production line of the factory when the predicted benefit value meets the valid conditions, so as to perform data optimization on the production line of the factory.

In a third aspect, embodiments of the present application also provide an electronic device, which includes:

one or more processors;

a storage device configured to store one or more programs;

When the one or more programs are executed by the one or more processors, the one or more processors are caused to implement the above-mentioned production line optimization method based on digital twins.

In a fourth aspect, embodiments of the present application also provide a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the above-mentioned production line optimization method based on digital twins is implemented.

Description of the drawings

Figure 1 is a schematic flow chart of a production line optimization method based on digital twins provided by an embodiment of the present application;

Figure 2 is a schematic flow chart of another production line optimization method based on digital twins provided by an embodiment of the present application;

Figure 3 is a schematic structural diagram of a digital twin-based production line optimization device provided by an embodiment of the present application;

FIG. 4 is a schematic structural diagram of an electronic device provided by an embodiment of the present application.

Detailed ways

The present application will be described below in conjunction with the drawings and embodiments. The specific embodiments described herein are merely illustrative of the application. For convenience of description, only parts relevant to the present application are shown in the drawings.

Figure 1 is a schematic flowchart of a digital twin-based production line optimization method provided by an embodiment of the present application. This method can be executed by a digital twin-based production line optimization device provided by an embodiment of the present application. The device can use software and/or Or implemented in hardware. In one embodiment, the apparatus may be integrated in an electronic device, such as a server. The following embodiments will be described by taking the device integrated in an electronic device as an example. Referring to Figure 1, the method may include the following steps:

S110. Obtain the historical production data of the factory, train the factory dynamic model based on the historical production data, and obtain a virtual data model.

For example, a factory is a production line enterprise that can process raw materials to produce commercial products. Among them, the factory's production process includes multiple production factors such as workers, equipment, materials, methods, environment and testing. The historical production data of the factory can be the data information generated by the factory in the actual production process. It can be the big data of the factory's production process and the operating parameters of the actual production equipment. The parameters include material information, personnel information, equipment information and Environmental information, the data information generated during the production process can be material data: it can be material ratio, corresponding input equipment, production time, experience links and output corresponding equipment, etc. It can also be personnel data: setting positions, Information such as the personnel ratio and shift sequence corresponding to the machine can also be equipment data: equipment connection methods, equipment operation logic, equipment maintenance and equipment operation rules, etc. It can also be environmental data: products corresponding to different environments. Line operating parameters adjustment range and other data. Among them, the material ratio in the historical production data can be the ratio of raw materials in the product production process, for example: making bread is the ratio of the weight of water, sugar, butter and various types of flour; the corresponding input device can be the production equipment model of the material. , function and other aspects of information; the production time can be the production time of the material running in the production equipment, for example: the time to make bread can be 30 minutes; the experience link can be the link that needs to be experienced to make product materials, for example: when making bread The materials need to be mixed, fermented, shaped, baked and packaged first. The set positions can be corresponding positions set in links that require manual intervention in the production process based on the machine's operating speed, such as: safety inspection posts, waste processing posts, operation monitoring posts, product packaging posts and other information data; the machine corresponding The staffing ratio can be different types of machines. According to the actual link requirements, the staffing required for each position. For example, the staffing ratio of a production workshop can be 8 people. The shift sequence can be the order of personnel in the factory during the operation of the machine, or it can be determined as 2 shifts or 3 shifts according to the product production speed. The connection method of the equipment can be to determine the length of the production line according to the contract supply demand of the product. For the production line with a longer duration, choose the physical connection method to ensure the stability of the production line. For the production line with the shorter duration, choose the network connection. . The equipment operation logic can be to determine the equipment operation principle based on the material proportions and links of the product. For example, for the production of fresh food, it is necessary to start the washing machine first and start the production link according to the degree of washing. Equipment maintenance and equipment operation rules can include equipment maintenance time, maintenance procedures, and basic operating principles during equipment operation. The factory dynamic model can be a basic factory operation model based on multiple equipment information, business logic structure and production process data on the actual production line of the factory. It can clearly display the production process of generating the corresponding products of the factory. The virtual data model can be a simulation model obtained by optimizing the factory dynamic model using historical production data. The factory dynamic model can be targeted and optimized based on actual needs and experimental data.

In the implementation, the historical production data of the factory is obtained from the factory to be optimized. The obtained historical production data can be processed to improve the data quality of the historical production data, and the training data set in the historical production data can be screened out according to actual needs. , the training data set can be obtained by filtering samples of historical production data according to the random forest algorithm. The factory dynamic model is trained according to the training data set to obtain a virtual data model, so that optimal production parameters can be determined from the production parameters of the virtual data model and the actual production parameters to optimize the factory production line. Among them, historical production data is acquired through collecting and monitoring the factory control system to obtain equipment process parameters, material types and codes, multi-type Programmable Logic Controller (PLC) and sensor-collected equipment and codes for the entire production process. Environmental parameters, including data processing of historical production data, and formatting of historical production data to achieve corresponding formatting unification.

In the embodiment of this application, the training data set in the historical production data can be used as the input of the factory dynamic model to train the parameters of the neural network, learn the operating rules and internal relationships between multiple parameters in the training data set, and obtain the corresponding The operating principle is used to obtain the virtual data model. During the training process, training can be based on a single factor in the historical production data, for example: marking the material ratio in the historical production data, inputting the material ratio and product quality into the factory dynamic model for training, and in the factory dynamic model The neural network learns the intrinsic relationship between material ratio and product quality, and obtains a virtual data model for inputting product material ratio, in which a single factor can be any type of data in historical production data; when the training process Multiple factors in historical production data are used for training, that is, need prediction factors are marked according to actual demand, and the marked factors and other data are input into the factory dynamic model for training. The neural network in the factory dynamic model learns the combination of marked factors and other data. The internal relationships among them are used to obtain a virtual data model of multiple factors. Among them, the virtual data model determines the information categories in the prediction results based on actual needs.

In the embodiments of this application, the factory dynamic model is trained based on historical production data, and during the training process of the virtual data model, the optimal solution of the material ratio in the historical production data is found during the simulation process through an enhanced algorithm, and the factory is optimized. The parameters on the production line provide the efficiency of material use and product quality in the factory, combined with the waste generated during the production process. On the basis of optimizing production parameters, the value stream map of the production line can be combined to optimize the production line production process, shorten the product development cycle, and improve production line production efficiency. Among them, the training of the virtual data model realizes the accurate simulation of the geometry, physics, behavior, rules, status and other characteristics of the factory's production line, and realizes the digital reconstruction of the actual production line activities.

S120. Input the factory's test production data into the virtual data model to obtain prediction results corresponding to the test production data.

For example, the test production data can be a data set for testing the virtual data model in historical production data, used to obtain prediction results corresponding to the test data output by the virtual data model, where the prediction results corresponding to the test production data can be used to obtain Calculate and evaluate the data of the predicted benefit value of the production parameters of the virtual data model. Among them, the data type and data information content in the prediction results are consistent with the historical production data. They all correspond to multiple data information on the production line of the actual factory, and Determine the data type and data information in the prediction results based on actual needs. For example, if the output efficiency of materials is used as the prediction target, the prediction results can include the output products corresponding to the input materials and the quantity of the output products. The prediction results can also include the energy consumption corresponding to the output materials.

In the implementation, the test production data is selected from the factory's historical production data, and the test production data is input into the virtual data model to simulate the production of the factory production line. When the factory production is simulated based on the test production data and the virtual data model, the virtual data model is obtained. The prediction results corresponding to the output test data are used to calculate the predicted benefit value corresponding to the prediction result based on the prediction results corresponding to the test data. Based on the predicted benefit value, it is judged whether the production parameters of the virtual data model are better than the actual production parameters of the factory, so that the judgment results can be Optimize the production parameters of the factory production line.

S130. Determine the predicted benefit value corresponding to the prediction result, and determine whether the predicted benefit value meets the valid conditions.

For example, the predicted benefit value corresponding to the prediction result can be the benefit value of the product corresponding to the factory calculated based on the data information in the prediction result, where the benefit value is generated through the input value of the material, energy consumption, and output from the prediction result. The output value of the product is determined. Among them, energy consumption is not only thermal energy consumed by materials on the factory production line for production line operations, but also includes employee labor costs, equipment consumption and other energy consumption. The preset benefit threshold can be set in advance based on actual needs and experimental data, and the preset benefit value threshold can be used to determine whether the predicted benefit meets the effective conditions.

During implementation, the predicted benefit value can be compared with the preset benefit threshold. If the predicted benefit value is greater than the preset benefit threshold, it is determined that the production parameters of the virtual data model corresponding to the predicted benefit value meet the factory's operating standards. According to the predicted benefit value Compare with the actual benefit value of the factory to determine whether the predicted benefit value meets the valid conditions. On the contrary, if the predicted benefit value is less than the preset benefit threshold, it is determined that the production parameters of the virtual data model corresponding to the predicted benefit value do not meet the factory's operating standards, then It is determined that the predicted benefit value does not meet the valid conditions, and the production parameters of the virtual data model cannot be mapped to the factory's production line.

S140. When the predicted benefit value meets the valid conditions, map the production parameters corresponding to the virtual data model to the factory's production line to perform data optimization on the factory's production line.

In implementation, the production parameters corresponding to the virtual data model can be parameters used to guide the virtual data model to simulate production line processing of test data. The production parameters include parameters such as the running time of the production line, material proportions, and production behavior rules. When the predicted benefit value is compared with the predicted benefit threshold and the actual benefit value, and it is determined that the predicted benefit value meets the valid conditions, it means that the production parameters of the virtual data model are better than the production parameters actually used by the factory, then the production parameters of the virtual data model are mapped To the factory's production line, the production parameters of the factory's production line are mapped to the production parameters of the virtual data model, so that the production parameters of the factory's production line are updated to optimize the data of the factory's production line. Among them, the data optimization of the factory's production line can be the optimization of the material ratio of the factory's production line, or the overall optimization of the factory's production line. The pertinence of the optimization is mainly reflected in the data preprocessing. Screening of samples and training of machine learning algorithms.

In the embodiment of the present application, the production parameters corresponding to the virtual data model may be data corresponding to factors in the training process based on the virtual data model. That is, if a single factor is trained, the production parameter may be a single factor in the marked historical production data. Data corresponding to factors, for example: when marking material ratios in historical production data, the production parameters corresponding to the virtual data model are the material ratios; when training multiple factors, the production parameters can be marked historical production data data information in.

In the embodiment of this application, the historical production data of the factory is obtained, and the dynamic model of the factory is trained according to the historical production data to obtain a virtual data model; the test production data of the factory is input into the virtual data model to obtain the prediction results corresponding to the test production data; Determine the predicted benefit value corresponding to the prediction result, and determine whether the predicted benefit value meets the valid conditions; when the predicted benefit value meets the valid conditions, map the production parameters corresponding to the virtual data model to the factory's production line, so as to predict the factory's production line. Perform data optimization. That is, in the embodiment of this application, the virtual data model is trained through historical production data to simulate the factory production line, and the benefit value calculated by borrowing the prediction results corresponding to the test data output by the model is used to determine whether the virtual data model satisfies the valid conditions, which will satisfy the valid conditions. The production parameters in the conditional virtual data model are mapped to the factory production line to realize the optimization of the factory production line using production data, which can more accurately and effectively improve material utilization and product quality, and overall increase the factory's profitability.

The production line optimization method based on digital twins provided by the embodiment of the present application is described below. As shown in Figure 2, the method may include the following steps:

S210. Obtain the historical production data of the factory, train the factory dynamic model based on the historical production data, and obtain a virtual data model.

Before obtaining the historical production data of the factory, it also includes:

Obtain the equipment information, business logic structure and production process flow of the factory; build the equipment assembly line of the factory based on the equipment information and business logic structure; form a factory dynamic model of the factory based on the equipment assembly line and production process flow.

For example, the equipment information of the factory can be the information of all electromechanical equipment used on the factory's production line, and can be information such as equipment model and equipment parameters, which is used to provide hardware unit information for the factory's production line. The business logic structure can be the frame design information and functional module information of the factory's production line, which is used to arrange the correlation between the equipment locations and functional modules of the factory. The production process flow can be the production information of the factory's production line and the implementation method of the production line. It can be the equipment parameters, material types, environmental parameters and other information during the factory's production line production. The equipment pipeline can connect factory equipment through communication connections, wire connections, physical connections, etc., in order to form a complete framework design and corresponding functions corresponding to the business logic results.

During implementation, the factory's equipment information, business logic structure, and production process flow are obtained from the factory's production line database, and the factory's equipment assembly line is built based on the factory's equipment information and business logic structure. Among them, the construction of the factory's equipment assembly line can be a virtual construction on the application software based on the factory's equipment information and business logic structure, or a physical construction of different proportions in the laboratory based on the factory's equipment information and business logic structure, or it can be It can be a construction method that combines virtual and physical objects. Input the information from the production process into the built equipment assembly line to form a factory dynamic model based on the production process.

In the embodiment of this application, the action of obtaining historical production data can be real-time. It can be that after the factory's production line outputs products once, the historical production data is updated once, and the virtual data model is trained once based on the historical production data. Among them, the training of the virtual data model is constantly iterative based on data updates. Only when the update of historical production data is stopped will the training of the virtual data model in the target direction stop.

The factory dynamic model is trained based on historical production data to obtain a virtual data model, including:

Perform data preprocessing on historical production data to obtain training data; train the factory dynamic model based on the training data to obtain a virtual data model.

For example, the training data can be data for targeted training of the factory dynamic model, and data that can meet the training needs can be obtained through data preprocessing of historical production data. The data preprocessing can be a single-factor analysis of historical production data. Screening can also be used to screen historical production data by multiple factors in order to obtain training data. A single factor can be factors such as material ratio, or it can be multiple factors related to the factory's benefit value.

In implementation, the historical production data of the factory is obtained from the factory to be optimized, and the obtained historical production data can be processed, that is, the training data set in the historical production data is screened out according to actual needs to improve the historical production data. Quality can be obtained by filtering samples of historical production data according to the random forest algorithm to obtain a training data set. The factory dynamic model is trained according to the training data set to obtain a virtual data model, so that optimal production parameters can be determined from the production parameters of the virtual data model and the actual production parameters to optimize the factory production line.

Perform data preprocessing on historical production data to obtain training data, including:

Perform data preprocessing on historical production data to obtain clean data. Data preprocessing includes eliminating redundant data and extracting data features; training data is filtered out from the clean data based on the data features of the clean data.

For example, eliminating redundant data can be to eliminate duplicate data in historical production data, where redundant data refers to duplicate data in historical production data, and the same data can be stored in different data files. Extracting data features can be based on training requirements to extract data features corresponding to any factors in historical production data. It can be data on motor rotors and assembly heat pipes in historical production data. Cleaning data can be production data obtained by eliminating redundant data and extracting data features from historical production data.

In the implementation, the historical production data of the factory is obtained from the factory to be optimized, and the historical production data of the factory is preprocessed. Redundant data operations are eliminated on the historical production data of the factory, and duplicate data in the historical production data of the factory are deleted. , and then extract data features from the deduplicated historical production data, which can be to obtain the frequency in the motor rotor data, the assembly heat pipe data, and obtain the cleaning data corresponding to the historical production data. According to the data characteristics of the cleaning data, training data is selected from the cleaning data. This can be done by filtering out the frequency noise data in the data of the motor rotor and filtering out the training data without noise. It can also be done by filtering out the data on the assembly of heat pipes. Data with abnormal heat dissipation and heat are filtered out, and training data with normal heat dissipation from the data of assembled heat pipes are screened out. Among them, the corresponding data features can be extracted from the historical production data according to the optimization target direction of the factory dynamic model, and the factory dynamic model can be optimized and trained in the target direction to obtain a virtual data model.

S220: Input the factory's test production data into the virtual data model to obtain prediction results corresponding to the test production data.

S230. Determine the input value of the material in the prediction result, the energy consumption, and the output value of the output product in the prediction result.

In implementation, the value of the input amount of materials can be calculated by testing the input amount of materials in the production data, the unit prices of multiple types, and the material proportions to calculate the value of the input materials corresponding to the output products. Energy consumption can be the energy consumed by the production line from materials to output products. Energy consumption is not only the thermal energy consumed by the materials on the factory production line for production line operations, but also includes employee labor costs, equipment consumption, etc. Various energy consumption. The output value of the output product in the prediction result can be the test production data input into the virtual data model for production line production simulation, and the data and the value of the produced product during the production line production simulation process of the virtual data model are obtained. Input the test production data into the virtual data model for production line production simulation, and obtain the data and output data of the virtual data model during the production line production simulation process, and obtain the prediction results of the test production data by the virtual data model, so as to facilitate the prediction results based on Calculate the predicted benefit value based on the input value of the material, energy consumption, and output value of the output product in the prediction result, and determine whether the production parameters of the virtual data model can be mapped to the factory's production line.

S240. Determine the predicted benefit value corresponding to the prediction result based on the input value of the material in the prediction result, energy consumption and the output value of the output product in the prediction result, and determine whether the predicted benefit value meets the valid conditions.

During implementation, the test production data is input into the virtual data model for production line production simulation, and the data and output data during the production line production simulation process of the virtual data model are obtained, and the prediction results of the test production data by the virtual data model are obtained. According to the input value of the material in the prediction result, energy consumption and the output value of the output product in the prediction result, corresponding calculations are performed to obtain the predicted benefit value corresponding to the prediction result. The difference between the output value of the output product and the value of energy consumption and material input can be used as the predicted benefit value based on the prediction results. Whether the effective conditions are met is determined based on the predicted benefit value, the preset benefit threshold and the actual benefit value.

The predicted benefit value corresponding to the prediction result is determined based on the input value of the material in the prediction result, energy consumption and the output value of the output product in the prediction result, including:

Among them, V is the predicted benefit value, I is the input amount of the material, S is the unit price of the material, E is the energy consumption in the production process, Y is the output of the output product, P is the market price of the output product, and i is the i-th kind Output product, n is a positive integer greater than 1, _Yi is the output of the i-th output product, and _Pi is the market price of the i-th output product.

During implementation, the test production data is input into the virtual data model for production line production simulation, and the data and output data during the production line production simulation process of the virtual data model are obtained, and the prediction results of the test production data by the virtual data model are obtained. Calculate the output value of the output product in the prediction result based on the output of the output product, the market price of the output product, and the type of output product, and calculate the input volume value of the material based on the input volume of the material and the unit price of the material. The output price of the output product using the prediction results is subtracted from the input price of the material and energy consumption in sequence according to formula (1) to obtain the predicted benefit value corresponding to the test production data.

Determine whether the predicted benefit value meets valid conditions, including:

Determine whether the predicted benefit value is greater than the preset benefit threshold; when the predicted benefit value is greater than the preset benefit threshold, determine whether the preset benefit threshold is greater than the actual benefit value; when the preset benefit threshold is not greater than the actual benefit value, determine whether the predicted benefit value is greater than Actual benefit value; when the predicted benefit value is greater than the actual benefit value, the predicted benefit value meets the valid conditions.

For example, a preset benefit threshold can be set in advance based on actual needs and experimental data, and the preset benefit value threshold can be used to determine whether the production parameters of the virtual data model corresponding to the predicted benefit value meet the factory's production line operation standards. For example, the predicted benefit value can be compared with the preset benefit threshold. If the predicted benefit value is greater than the preset benefit threshold, it is determined that the production parameters of the virtual data model corresponding to the predicted benefit value meet the factory's production line operation standards. Otherwise, if the measured benefit value If the benefit value is less than the preset benefit threshold, it is determined that the production parameters of the virtual data model corresponding to the predicted benefit value do not meet the factory's production line operation standards. If the predicted benefit value is equal to the preset benefit threshold, then the virtual data model corresponding to the predicted benefit value is determined. The production parameters just meet the factory's production line operation standards, but do not meet the effective conditions. The preset benefit threshold may be preset based on the operating standard data of the production parameters of the factory's production line, or may be determined based on the average of the actual benefit values of the factory's production line within a preset time period. The actual benefit value can be the real value of the actual benefit obtained by testing the production data on the current factory's production line, which is used to compare with the predicted benefit value to determine the quality of the production parameters of the factory's production line and the production parameters of the virtual data model. It is also the data information for judging the effective conditions.

During implementation, the predicted benefit value can be compared with the preset benefit threshold. If the predicted benefit value is greater than the preset benefit threshold, it is determined that the production parameters of the virtual data model corresponding to the predicted benefit value meet the factory's operating standards. Under standard conditions, the predicted benefit value is compared with the actual benefit value to determine whether the predicted benefit value meets the valid conditions. Among them, before judging the size of the predicted benefit value and the actual benefit value, the size between the actual benefit value and the preset benefit threshold can be determined in advance. If the actual benefit is less than the preset benefit threshold, there is no need to compare the predicted benefit value and the actual benefit value. The size is in line with the factory's operating standards, that is, it meets the effective conditions. If the actual benefit value is greater than the preset benefit threshold, you need to continue to judge and compare the size between the predicted benefit value and the actual benefit value. After the predicted benefit value is greater than the actual benefit value , determine that the predicted benefit value meets the valid conditions. If the predicted benefit value is less than the preset benefit threshold, it is determined that the production parameters of the virtual data model corresponding to the predicted benefit value do not meet the factory's operating standards, it is determined that the predicted benefit value does not meet the valid conditions, and the production parameters of the virtual data model cannot be mapped to Factory production line.

S250. When the predicted benefit value meets the valid conditions, map the production parameters corresponding to the virtual data model to the factory's production line to perform data optimization on the factory's production line.

In the embodiment of this application, the historical production data of the factory is obtained, and the dynamic model of the factory is trained according to the historical production data to obtain a virtual data model; the test production data of the factory is input into the virtual data model to obtain the prediction results corresponding to the test production data; Determine the predicted benefit value corresponding to the prediction result, and determine whether the predicted benefit value meets the valid conditions; when the predicted benefit value meets the valid conditions, map the production parameters corresponding to the virtual data model to the factory's production line, so as to predict the factory's production line. Perform data optimization. That is, in the embodiment of this application, the virtual data model is trained through historical production data to simulate the factory production line, and the benefit value calculated by borrowing the prediction results corresponding to the test production data output by the model is calculated to determine that the virtual data model meets the effective conditions, which will satisfy The production parameters in the virtual data model of effective conditions are mapped to the factory production line to realize the optimization of the factory production line using production data, which can more accurately and effectively improve material utilization and product quality, and overall increase the factory's profitability.

Figure 3 is a schematic structural diagram of a digital twin-based production line optimization device provided by an embodiment of the present application. As shown in Figure 3, the digital twin-based production line optimization device includes:

The model training module 310 is configured to obtain the historical production data of the factory, train the factory dynamic model according to the historical production data, and obtain a virtual data model; the prediction acquisition module 320 is configured to input the test production data of the factory into the The virtual data model obtains the prediction results corresponding to the test production data; the validity judgment module 330 is configured to determine the predicted benefit value corresponding to the prediction result, and determine whether the predicted benefit value meets the valid conditions; the production line optimization module 340 , is set so that when the predicted benefit value meets the valid conditions, the production parameters corresponding to the virtual data model are mapped to the production line of the factory to perform data optimization on the production line of the factory.

In one embodiment, the model training module 310 trains the factory dynamic model based on the historical production data to obtain the virtual data model, which includes:

Perform data preprocessing on the historical production data to obtain training data; train the factory dynamic model according to the training data to obtain the virtual data model.

In one embodiment, the model training module 310 performs data preprocessing on the historical production data to obtain clean data. The data preprocessing includes eliminating redundant data and extracting data features; according to the data features of the clean data, The training data is filtered out from the cleaning data.

In one embodiment, the validity judgment module 330 determines whether the predicted benefit value meets valid conditions, including:

Determine whether the predicted benefit value is greater than the preset benefit threshold; when the predicted benefit value is greater than the preset benefit threshold, determine whether the preset benefit threshold is greater than the actual benefit value; when the preset benefit threshold is not greater than the When the actual benefit value is stated, it is determined whether the predicted benefit value is greater than the actual benefit value; when the predicted benefit value is greater than the actual benefit value, the predicted benefit value satisfies the valid condition.

In one embodiment, the validity judgment module 330 determines the predicted benefit value corresponding to the prediction result, including:

Determine the input value of the material in the prediction result, energy consumption and the output value of the output product in the prediction result; according to the input value of the material in the prediction result, energy consumption and the output of the output product in the prediction result The value determines the predicted benefit value corresponding to the predicted result.

In one embodiment, the validity judgment module 330 determines the predicted benefit value V corresponding to the prediction result based on the input value of the material in the prediction result, energy consumption, and the output value of the output product in the prediction result, including:

Among them, V is the predicted benefit value, I is the input amount of the material, S is the unit price of the material, E is the energy consumption in the production process, Y is the output of the output product, P is the market price of the output product, and i is the i-th kind Output products, n is a positive integer greater than 1, Y _i is the output of the i-th output product, and P _i is the market price of the i-th output product.

In one embodiment, before the model training module 310 obtains the historical production data of the factory, it also includes:

Obtain the equipment information, business logic structure and production process flow of the factory; build the equipment assembly line of the factory based on the equipment information and the business logic structure; form the equipment assembly line based on the equipment assembly line and the production process flow. Factory dynamic model of the factory.

The device of the embodiment of this application obtains the historical production data of the factory, trains the dynamic model of the factory based on the historical production data, and obtains a virtual data model; inputs the test production data of the factory into the virtual data model to obtain prediction results corresponding to the test production data; Determine the predicted benefit value corresponding to the prediction result, and determine whether the predicted benefit value meets the valid conditions; when the predicted benefit value meets the valid conditions, map the production parameters corresponding to the virtual data model to the factory's production line, so as to predict the factory's production line. Perform data optimization. That is, in the embodiment of this application, the virtual data model is trained through historical production data to simulate the factory production line, and the benefit value calculated by borrowing the prediction results corresponding to the test data output by the model is used to determine whether the virtual data model satisfies the valid conditions, which will satisfy the valid conditions. The production parameters in the conditional virtual data model are mapped to the factory production line to realize the optimization of the factory production line using production data, which can more accurately and effectively improve material utilization and product quality, and overall increase the factory's profitability.

FIG. 4 is a schematic structural diagram of an electronic device provided by an embodiment of the present application. 4 illustrates a block diagram of an exemplary electronic device 12 suitable for implementing embodiments of the present application. The electronic device 12 shown in FIG. 4 is only an example and should not bring any limitations to the functions and scope of use of the embodiments of the present application.

As shown in Figure 4, electronic device 12 is embodied in the form of a general-purpose computing device. Components of electronic device 12 may include, but are not limited to: one or more processors or processing units 16, storage system memory 28, and a bus 18 connecting various system components including storage system memory 28 and processing unit 16.

Bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, a graphics accelerated port, a processor, or a local bus using any of a variety of bus structures. For example, these architectures include, but are not limited to, Industry Standard Architecture (ISA) bus, Micro Channel Architecture (MAC) bus, Enhanced ISA bus, Video Electronics Standards Association (Video Electronics Standards) Association, VESA) local bus and Peripheral Component Interconnect (PCI) bus.

Electronic device 12 includes a variety of computer system readable media. These media can be any available media that can be accessed by electronic device 12, including volatile and nonvolatile media, removable and non-removable media.

Storage system memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and/or cache memory 32. Electronic device 12 may include other removable/non-removable, volatile/non-volatile computer system storage media. By way of example only, storage system 34 may be used to read and write to non-removable, non-volatile magnetic media (not shown in Figure 4, commonly referred to as a "hard drive"). Although not shown in FIG. 4, a disk drive may be provided for reading and writing to removable non-volatile disks (e.g., "floppy disks"), and for removable non-volatile optical disks (e.g., Compact Discs). Read-Only Memory (CD-ROM), Digital Video Disc-Read Only Memory (DVD-ROM) or other optical media) that reads and writes optical disc drives. In these cases, each drive may be connected to bus 18 through one or more data media interfaces. The storage system memory 28 may include at least one program product having a set of (eg, at least one) program modules configured to perform the functions of embodiments of the present application.

A program/utility 40 having a set (at least one) of program modules 42, including but not limited to an operating system, one or more application programs, other program modules, and Program data, each or a combination of these examples may include an implementation of a network environment. Program modules 42 generally perform functions and/or methods in the embodiments described herein.

Electronic device 12 may also communicate with one or more external devices 14 (e.g., keyboard, pointing device, display 24, etc.), may also communicate with one or more devices that enable a user to interact with electronic device 12, and/or with Any device (eg, network card, modem, etc.) that enables the electronic device 12 to communicate with one or more other computing devices. This communication may occur through input/output (I/O) interface 22. Moreover, the electronic device 12 can also communicate with one or more networks (such as a local area network (Local Area Network, LAN), a wide area network (Wide Area Network, WAN), and/or a public network, such as the Internet) through the network adapter 20. As shown, network adapter 20 communicates with other modules of electronic device 12 via bus 18 . It should be understood that, although not shown in the figures, other hardware and/or software modules may be used in conjunction with the electronic device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, disk arrays (Redundant Arrays). of Independent Disks, RAID) systems, tape drives and data backup storage systems, etc.

The processing unit 16 executes a variety of functional applications and data processing by running programs stored in the storage system memory 28, for example, implementing the digital twin-based production line optimization method provided by the embodiment of the present application, which method includes:

Obtain the historical production data of the factory, train the factory dynamic model based on the historical production data, and obtain a virtual data model; input the test production data of the factory into the virtual data model, and obtain the prediction results corresponding to the test production data. ; Determine the predicted benefit value corresponding to the prediction result, and determine whether the predicted benefit value meets the valid conditions; when the predicted benefit value meets the valid conditions, map the production parameters corresponding to the virtual data model to the factory in the production line to perform data optimization on the production line of the factory.

The implementation of this application also provides a computer-readable storage medium on which a computer program is stored, wherein when the program is executed by a processor, the digital twin-based production line optimization method is implemented, and the method includes:

The computer storage medium in the embodiment of the present application may be any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. Examples of computer-readable storage media (a non-exhaustive list) include: electrical connections having one or more conductors, portable computer disks, hard drives, RAM, ROM, Erasable Programmable Read Only Memory , EPROM or flash memory), optical fiber, CD-ROM, optical storage device, magnetic storage device, or any suitable combination of the above. As used herein, a computer-readable storage medium may be any tangible medium that contains or stores a program for use by or in connection with an instruction execution system, apparatus, or device.

A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave carrying computer-readable program code therein. Such propagated data signals may take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device .

Program code embodied on a computer-readable medium can be transmitted using any appropriate medium, including but not limited to wireless, wire, optical cable, radio frequency (Radio Frequency, RF), etc., or any suitable combination of the above.

Computer program code for performing operations of the present application may be written in one or more programming languages, including object-oriented programming languages such as Java, Smalltalk, C++, and conventional procedures, or a combination thereof. programming language such as "C" or a similar programming language. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. Where a remote computer is involved, the remote computer may be connected to the user's computer through any kind of network, including a LAN or WAN, or may be connected to an external computer (such as through the Internet using an Internet service provider).

Claims

Production line optimization methods based on digital twins include:

Obtain the historical production data of the factory, train the factory dynamic model based on the historical production data, and obtain a virtual data model;

Input the test production data of the factory into the virtual data model to obtain prediction results corresponding to the test production data;

Determine the predicted benefit value corresponding to the prediction result, and determine whether the predicted benefit value meets the valid conditions;

In response to the predicted benefit value meeting the valid condition, the production parameters corresponding to the virtual data model are mapped to the production line of the factory to perform data optimization on the production line of the factory.
The method according to claim 1, wherein said training the factory dynamic model according to the historical production data to obtain the virtual data model includes:

Perform data preprocessing on the historical production data to obtain training data;

The factory dynamic model is trained according to the training data to obtain the virtual data model.
The method according to claim 2, wherein said performing data preprocessing on the historical production data to obtain training data includes:

Perform data preprocessing on the historical production data to obtain clean data, where the data preprocessing includes eliminating redundant data and extracting data features;

The training data is filtered out from the cleaning data according to the data characteristics of the cleaning data.
The method according to claim 1, wherein determining whether the predicted benefit value meets a valid condition includes:

Determine whether the predicted benefit value is greater than a preset benefit threshold;

In response to the predicted benefit value being greater than the preset benefit threshold, determining whether the preset benefit threshold is greater than the actual benefit value;

In response to the preset benefit threshold not being greater than the actual benefit value, determining whether the predicted benefit value is greater than the actual benefit value;

In response to the predicted benefit value being greater than the actual benefit value, the predicted benefit value satisfies a valid condition.
The method according to claim 1, wherein determining the predicted benefit value corresponding to the predicted result includes:

Determine the input value of materials, energy consumption in the prediction results and the output value of the output products in the prediction results;

The predicted benefit value corresponding to the prediction result is determined based on the input value of the material in the prediction result, energy consumption and the output value of the output product in the prediction result.
The method according to claim 5, wherein the predicted benefit value corresponding to the prediction result is determined based on the input amount value of the material in the prediction result, energy consumption and the output value of the output product in the prediction result, including :

Among them, V is the predicted benefit value, I is the input amount of the material, S is the unit price of the material, E is the energy consumption in the production process, Y is the output of the output product, P is the market price of the output product, and i is the i-th kind Output product, n is a positive integer greater than 1, Yi is the output of the i-th output product, and Pi is the market price of the i-th output product.
The method according to claim 1, wherein before obtaining the historical production data of the factory, it further includes:

Obtain the equipment information, business logic structure and production process flow of the factory;

Establish the equipment assembly line of the factory according to the equipment information and the business logic structure;

A factory dynamic model of the factory is formed based on the equipment assembly line and the production process flow.
Production line optimization devices based on digital twins include:

The model training module is configured to obtain the historical production data of the factory, train the factory dynamic model based on the historical production data, and obtain a virtual data model;

The prediction acquisition module is configured to input the test production data of the factory into the virtual data model to obtain prediction results corresponding to the test production data;

A valid judgment module, configured to determine the predicted benefit value corresponding to the prediction result, and determine whether the predicted benefit value meets the valid conditions;

The production line optimization module is configured to map the production parameters corresponding to the virtual data model to the production line of the factory in response to the predicted benefit value satisfying the valid condition, so as to perform data optimization on the production line of the factory.
An electronic device including:

at least one processor;

a storage device configured to store at least one program;

When the at least one program is executed by the one or more processors, the at least one processor is caused to implement the digital twin-based production line optimization method as described in any one of claims 1 to 7.
A computer-readable storage medium stores a computer program. When the program is executed by a processor, the digital twin-based production line optimization method as described in any one of claims 1 to 7 is implemented.