Virtual production process control optimization method and system based on digital twinning and storage medium
Through digital twin technology, the virtual production process is optimized, and the problems of flexible production control and customer demand response in intelligent manufacturing are solved, visual tracking and feedback of the production process are realized, and the flexibility and responsiveness of the production system are improved.
Patent Information
- Application Number
- CN202410817887.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-24
- Publication Date
- 2024-11-26
AI Technical Summary
In the field of intelligent manufacturing, it is difficult for the existing technology to achieve effective control of flexible production systems and timely response to customer needs, especially in complex production processes, which cannot be fully visualized tracking and feedback.
Using a virtual production process control method based on digital twins, by building virtual production processes, building and optimizing digital twin models, collecting data in real time, reducing dimensionality and calculating production process control optimization indicators, optimizing production processes, and providing a visual system through AR equipment Track and feedback of flexible production modes with storage media.
It realizes optimized control of the production process, improves the flexibility and responsiveness of the production system, can promptly push and feedback production services, meet customer needs, and improves production efficiency and product quality.
Smart Images

Figure CN119026825A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of intelligent manufacturing process optimization, and in particular to a virtual production process control optimization method, system and storage medium based on digital twins. Background Art
[0002] Manufacturing technology is the most active field in the development of contemporary science and technology. It is an important means of product renewal, production development and international technological competition. All industrialized countries have listed advanced manufacturing technology as a national high-tech key technology and priority development project, and have attached great importance and attention to it. Manufacturing is the pillar of various industries, and the development of various industries depends on the support of the manufacturing industry, including the provision of cutting-edge and professional equipment, the provision of advanced general equipment, etc. The country's economic independence and industrial self-sufficiency are largely reflected in the level of manufacturing technology. In a broad sense, intelligent manufacturing is the deep integration of new generation information technology and advanced manufacturing technology, throughout the entire life cycle of products, manufacturing and services, and the optimal integration of corresponding systems to achieve digitalization, networking and intelligence, and continue to improve the product quality, efficiency and service level of enterprises, and promote the innovation, green, coordinated, open and shared development of the manufacturing industry.
[0003] In the field of intelligent manufacturing, in response to the market's increasingly strong demand for flexible production systems, the industry has currently seen a variety of representative flexible production solutions at both the equipment and software levels. Representative solutions at the equipment level include AGV. In the manufacturing industry, the application of AGV in intelligent production warehousing is known as the most flexible link in the production automation system, and it is also the preferred simple and effective automatic material transportation method in the production field. AGV is widely used in scenes such as production line material handling, finished product delivery to the warehouse, and raw material delivery to the production line; in the circulation industry, the application of Kiva-like robots is becoming more and more widespread. At the software level, warehouse management systems (WMS) and warehouse control systems (WCS) are also crucial for the embodiment of flexibility. The WMS system can help enterprises achieve refined control within the four walls from product entry to delivery, connect with the upper ERP system, and track the entire process of production and cost management of warehouse business in real time. The complex production field has problems such as untimely response and feedback to customer needs and inability to allow customers to visually track the production process in all directions. Summary of the invention
[0004] Purpose of the invention: To propose a virtual production process control optimization method, system and storage medium based on digital twins, so as to effectively solve the above-mentioned problems existing in the prior art.
[0005] First, a virtual production process control optimization method based on digital twin is proposed, and the steps are as follows:
[0006] S1. Build a virtual production process;
[0007] S2. Run the virtual production process and build a digital twin model; use the digital twin model to control the virtual production process and collect process data in real time;
[0008] S3. Perform a dimensionality reduction operation on the process data, calculate a production process control optimization index based on the process data after dimensionality reduction, and evaluate the quality of the digital twin model by the production process control optimization index:
[0009] If the quality of the current digital twin model meets the predetermined value, the current digital twin model is retained as the control optimization basis of the virtual production process;
[0010] If the quality of the current digital twin model does not meet the predetermined value, the current digital twin model is discarded, and step S2 is re-executed to iterate a digital twin model that meets the conditions.
[0011] In a further embodiment of the first aspect, step S1 specifically includes:
[0012] S1-1, obtaining the 3D coordinates of the positioning objects in each link of the actual production process, performing feature information matching on the 3D coordinates of the positioning objects in each link, and selecting the three-dimensional coordinates of the points of interest;
[0013] S1-2, transforming the three-dimensional coordinates of the interest point using the KL standard basis according to the minimum mean square error criterion, and reducing the dimension to obtain a two-dimensional coordinate value;
[0014] S1-3, mapping the two-dimensional coordinate values to the interval 0-1, obtaining corresponding feature parameter vectors, and storing them in a feature information library.
[0015] In a further embodiment of the first aspect, step S1-1 performs feature information matching on the 3D coordinates of the positioning object in each link, specifically including:
[0016] S1-1-1. Divide the working area of each machine room in the workshop into a category i , i = 1, 2, 3, ..., the device in the C working area is a sample G in this category i (x) = min||pp * || k = 1, 2, 3, ..., N;
[0017] S1-1-2. Calculate the distance between the real-time positioning coordinates p of the mobile object and each sample x4 in category m, and find the minimum distance g i (p), the distance is defined as the distance between the real-time positioning coordinates and the category ω, and the expression is as follows:
[0018] G i(x) = min||pp * || k = 1, 2, 3, ..., N;
[0019] S1-1-3. Find the minimum distance gi(p) between the positioning coordinates and category a:
[0020] N gi (p) = ming(p);
[0021] S1-1-4. Determine g j (p)≤l;
[0022] If the distance is less than a certain distance, the positioning coordinate x is classified into this category and the match is successful, otherwise the decision is rejected.
[0023] In a further embodiment of the first aspect, the digital twin model in step S2 is constructed in multiple;
[0024] Trained through different machine training data groups; the data set not used for training is the validation set; through the validation set and reasonable validation indicators, the quality of different digital twin models is compared and the optimal digital twin model is found. The optimal digital twin model is combined with real-time lloT data for simulation and optimization, and an optimization plan is proposed for production line control.
[0025] In a further embodiment of the first aspect, the true value and absolute value of FE are displayed by a curve to compare the efficiency of different digital twin models:
[0026]
[0027] Where y1 is the true value of the training data set, It is the predicted value calculated by the training model; the smaller the absolute value of FE, the higher the efficiency of the current digital twin model; the digital twin model with the highest efficiency is selected as the optimal digital twin model.
[0028] In a further embodiment of the first aspect, the production process control optimization indicators in step S3 include model accuracy rate MAR, root mean square error RMSE, and variance explanation rate VIR.
[0029] In a further embodiment of the first aspect, the expression of the model accuracy rate MAR is as follows:
[0030]
[0031] Where y1 is the true value of the validation data set, is the average value, is the predicted value calculated by the trained model;
[0032] The model accuracy rate MAR is used to reflect the deviation between the predicted result and the actual value. The larger the MAR, the better, and the maximum value is 1.
[0033] In a further embodiment of the first aspect, the expression of the root mean square error RMSE is as follows:
[0034]
[0035] Where y1 is the true value of the validation data set, is the average value, It is the predicted value calculated by the training model; RMSE is based on the quadratic distance between the actual value and the predicted value and is used to evaluate the performance of the regression model; the smaller the RMSE, the better, and the minimum value is 0.
[0036] In a further embodiment of the first aspect, the expression of the variance explanation rate VIR is as follows:
[0037]
[0038] Where g1 is the true value of the validation data set, is the average value, is the predicted value calculated by the trained model;
[0039] The VIR index is used to evaluate the degree to which the model prediction is better than the random prediction: when the value is greater than 0, the model prediction is better than the random prediction; the larger the VIR, the better, and the maximum value is 1.
[0040] As a second aspect of the present invention, a visualization system for running the virtual production process control optimization method based on digital twin as described in the first aspect is proposed, and the visualization system includes:
[0041] A virtual production process construction unit is used to collect various data in the actual production process and construct a virtual production process;
[0042] A digital twin model construction unit, used to construct a digital twin model, use the digital twin model to control the virtual production process, and collect process data in real time;
[0043] The optimization result output unit is used to perform a dimensionality reduction operation on the process data, calculate the production process control optimization index based on the process data after dimensionality reduction, and evaluate the quality of the digital twin model by the production process control optimization index:
[0044] If the quality of the current digital twin model meets the predetermined value, the current digital twin model is retained as the control optimization basis of the virtual production process;
[0045] If the quality of the current digital twin model does not meet the predetermined value, the current digital twin model is discarded and a new digital twin model that meets the conditions is iterated again;
[0046] AR devices use a visual interface to help users select the scope of production services, and push and respond to production service feedback to users in a timely manner.
[0047] As a third aspect of the present invention, a computer-readable storage medium is proposed, which stores at least one executable instruction. When the executable instruction is run on an electronic device, the electronic device executes the virtual production process control optimization method based on digital twins as described in the first aspect.
[0048] Beneficial effects: This application uses the reduced-dimensional data to track and simulate the constructed virtual production process through two indicators: the scope flexibility and response flexibility measurement of the production system. The AR-based production equipment management visualization helps customers select the scope of production services, and timely pushes and responds to production service feedback to customer needs. BRIEF DESCRIPTION OF THE DRAWINGS
[0049] Figure 1 It is a flow chart of a virtual production process control optimization method in an embodiment of the present invention.
[0050] Figure 2 It is a schematic diagram of constructing a virtual production process in an embodiment of the present invention.
[0051] Figure 3 It is a schematic diagram of a visualization system for running a virtual production process control optimization method in an embodiment of the present invention. DETAILED DESCRIPTION
[0052] In the following description, a large number of specific details are provided to provide a more thorough understanding of the present invention. However, it is apparent to those skilled in the art that the present invention can be implemented without one or more of these details. In other examples, in order to avoid confusion with the present invention, some technical features known in the art are not described.
[0053] like Figure 1 As shown, this embodiment discloses a specific process of a virtual production process control optimization method, which is divided into three steps:
[0054] Step 1: Build a virtual production process. This embodiment uses the working area in a discrete data center or warehouse as the application scenario, and implements the equipment monitoring function in the working area. First, design the visualization content and visualization UI of a certain device in the working area. According to the visualization design, build a virtual visualization model in Unity, and finally use indoor positioning technology and AR to realize the mobile monitoring function of the device.
[0055] When the production personnel are within a certain range of the DMC-80H device in the work center, the indoor positioning system is used to obtain the indoor positioning coordinate information of the mobile device, and this positioning information is used as the positioning coordinate information of the mobile device camera. The camera positioning information is transmitted to the system server and matched with the positioning feature information of the device (target point of interest). If the match is successful, the virtual monitoring information of the device is pushed based on the estimated camera posture of the mobile device and fused with the device for display.
[0056] After building the AR application scene in Unity, you can publish it to the Android system and install the application. After opening the application, there are two selectable buttons on the home page, namely device basic data and device operation data. Figure 2 shown.
[0057] The 3D coordinates of the positioning objects in each link of the production process are obtained, and then the three-dimensional coordinates of the points of interest are transformed using the KL standard basis according to the minimum mean square error criterion to obtain the reduced-dimensional feature information (two-dimensional). Finally, the two-dimensional coordinate values are mapped to the 0-1 interval (normalization processing) to obtain the corresponding feature parameter vector, which is stored in the feature information library.
[0058] Next, the process of positioning feature information matching is executed: the feature matching algorithm will vary depending on the indoor environment. The main matching algorithms include probability method, Pearson correlation coefficient matching method, K nearest neighbor method and kernel function method. Considering the characteristics of the indoor environment and the form of feature matching, this case adopts the nearest neighbor decision algorithm with rejection decision. The specific steps are as follows:
[0059] Divide each machine room work area in the workshop into a category i ,i=1,2,3,...,The device in the C working area is a sample of this category:
[0060] G i (x) = min||pp * ||*k=1,2,3,...,N Formula (1)
[0061] Calculate the distance between the real-time positioning coordinates p of the mobile object and each sample x4 in category m, and find the minimum distance g i (p), which is defined as the distance between the real-time positioning coordinates and the category ω.
[0062] G i (x) = min||pp * || k=1,2,3,…,N Formula (2)
[0063] Find the minimum distance g between the positioning coordinates and category ai (p)
[0064] Ng i (p) = ming(p) Formula (3)
[0065] Determine g j (p)≤l Formula (4)
[0066] If the distance is less than a certain distance, the positioning coordinate x is classified into this category and the match is successful, otherwise the decision is rejected.
[0067] Step 2: Run the virtual production process and build a digital twin model; use the digital twin model to control the virtual production process and collect process data in real time.
[0068] Step three: perform a dimensionality reduction operation on the process data, calculate a production process control optimization index based on the reduced dimensionality process data, and evaluate the quality of the digital twin model by the production process control optimization index.
[0069] The digital twin technology is used to process the collected data, and the quality of the digital model is comprehensively evaluated through four indicators: model accuracy ratio (MAR), root mean square error (RMSE), variance interpretation rate (VIR) and PPMCC.
[0070] In order to evaluate the convergence speed of different algorithms and observe the convergence process, the fitting error formula (5) can be used. The true value and absolute value of FE are displayed by the curve to compare the efficiency of different algorithms. y1 is the true value of the training data set, is the predicted value. The smaller the absolute value of FE is, the better.
[0071]
[0072] The digital twin model is trained with different machine training data sets. The data set that is not used for training is called the validation set. Through the validation set and reasonable validation indicators, the quality of different models can be compared and the optimal model can be found. That is, the digital twin model for production control. The digital twin model will be combined with real-time lloT data for simulation and optimization, and an optimization plan will be proposed for production line control.
[0073] In order to comprehensively evaluate the quality of the digital production model, the following four evaluation criteria are used for a multi-dimensional comprehensive evaluation, including model accuracy ratio (MAR), root mean square error (RMSE), variance interpretation rate (VIR) and PPMCC. The calculation of the indicators is shown in formulas (6) to (9). For example, the output of a specified product, x is the predicted label (independent variable). y1 is the true value of the validation data set, is the average value, is the predicted value calculated by the trained model.
[0074]
[0075] (Equation 6) is the calculation formula of MAR, which is the difference between 1 and the first-order deviation rate. MAR can reflect the deviation between the predicted result and the actual value. The larger the MAR, the better, and the maximum value is 1.
[0076]
[0077] (Equation 7) is the calculation formula of RMSE. RMSE is based on the quadratic distance between the actual value and the predicted value and is used to evaluate the performance of the regression model. The smaller the RMSE, the better, and the minimum value is 0.
[0078]
[0079] Formula 8 is the calculation formula of VIR, which is transformed by the coefficient of determination. In the manufacturing industry, the target prediction value usually fluctuates in a range much smaller than its absolute value, which often leads to the situation that even a random prediction with an average historical value will appear to have a "very good" MAR accuracy level. Therefore, the VIR index is used to evaluate the degree to which the model prediction is better than the random prediction. When the value is greater than 0, the model prediction is better than the random prediction. The larger the VIR, the better, and the maximum value is 1.
[0080] If the quality of the current digital twin model meets the predetermined value, the current digital twin model is retained as the control optimization basis of the virtual production process;
[0081] If the quality of the current digital twin model does not meet the predetermined value, the current digital twin model is discarded, and step 2 is re-executed to iterate a digital twin model that meets the conditions.
[0082] like Figure 3As shown, as a preferred embodiment, a visualization system 400 for running a virtual production process control optimization method based on digital twins is also disclosed. The visualization system 400 includes a virtual production process construction unit 401, a digital twin model construction unit 402, an optimization result output unit 403, and an AR device 404.
[0083] The virtual production process construction unit 401 is used to collect various data in the actual production process and construct a virtual production process. The digital twin model construction unit 402 is used to construct a digital twin model, use the digital twin model to control the virtual production process, and collect process data in real time. The optimization result output unit 403 is used to perform a dimensionality reduction operation on the process data, calculate the production process control optimization index based on the process data after dimensionality reduction, and evaluate the quality of the digital twin model by the production process control optimization index: if the quality of the current digital twin model meets the predetermined value, the current digital twin model is retained as the control optimization basis of the virtual production process; if the quality of the current digital twin model does not meet the predetermined value, the current digital twin model is discarded, and a digital twin model that meets the conditions is re-iterated. The AR device 404 uses a visual interface to select the production service scope for the user, and pushes and responds to production service feedback to the user in a timely manner.
[0084] As a preferred embodiment, a computer-readable storage medium is proposed, in which at least one executable instruction is stored. When the executable instruction is run on an electronic device, the electronic device performs the operation of the virtual production process control optimization method as described in the above embodiment. More specific examples of computer-readable storage media in the present disclosure may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. The computer-readable storage medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The readable signal medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in combination with an instruction execution system, an apparatus, or a device.
[0085] In summary, the present invention proposes a virtual production process control optimization method, system and storage medium based on digital twins, which is aimed at the problem that the response and feedback to customer needs in complex production fields are not timely and customers cannot visually track the production process in an all-round way. First, build flexible capabilities in intelligent manufacturing. According to the level of flexibility, flexible capabilities can be divided into innovation capabilities, adaptability and buffering capabilities. Innovation capabilities are to solve complex creative changes, adaptability is to solve random changes in the environment, and buffering capabilities are to reduce and cache changes. Secondly, virtual production is realized by mapping flexible capabilities to flexible production modes. Then, the reduced-dimensional data is tracked and simulated through two indicators of the scope flexibility and response flexibility measurement of the production system to construct a virtual production process. And the production equipment management visualization based on AR helps customers select the scope of production services, and timely pushes and responds to production service feedback to customer needs.
[0086] As described above, although the present invention has been shown and described with reference to specific preferred embodiments, it should not be construed as limiting the present invention itself. Various changes in form and details may be made without departing from the spirit and scope of the present invention as defined in the appended claims.
Claims
1. A virtual production process control optimization method based on digital twin, characterized in that: include: S1. Build a virtual production process; S2. Run the virtual production process and build a digital twin model; Using the digital twin model to control the virtual production process and collect process data in real time; S3. Perform a dimensionality reduction operation on the process data, calculate a production process control optimization index based on the process data after dimensionality reduction, and evaluate the quality of the digital twin model by the production process control optimization index: If the quality of the current digital twin model meets the predetermined value, the current digital twin model is retained as the control optimization basis of the virtual production process; If the quality of the current digital twin model does not meet the predetermined value, the current digital twin model is discarded, and step S2 is re-executed to iterate a digital twin model that meets the conditions.
2. The virtual production process control optimization method based on digital twin according to claim 1 is characterized in that: Step S1 specifically includes: S1-1, obtaining the 3D coordinates of the positioning objects in each link of the actual production process, performing feature information matching on the 3D coordinates of the positioning objects in each link, and selecting the three-dimensional coordinates of the points of interest; S1-2, transforming the three-dimensional coordinates of the interest point using the KL standard basis according to the minimum mean square error criterion, and reducing the dimension to obtain a two-dimensional coordinate value; S1-3, mapping the two-dimensional coordinate values to the interval 0-1, obtaining corresponding feature parameter vectors, and storing them in a feature information library.
3. The virtual production process control optimization method based on digital twin according to claim 2 is characterized in that: In step S1-1, feature information matching is performed on the 3D coordinates of the positioning objects in each link, specifically including: S1-1-1. Divide the working area of each machine room in the workshop into a category i , i = 1, 2, 3, ..., the device in the C working area is a sample G in this category i (x) = min||pp * || k = 1, 2, 3, ..., N; S1-1-2. Calculate the distance between the real-time positioning coordinates p of the mobile object and each sample x4 in category m, and find the minimum distance g i (p), the distance is defined as the distance between the real-time positioning coordinates and the category ω, and the expression is as follows: G i (x)=min||p-p * || k=1,2,3,…,N; S1-1-3. Find the minimum distance g between the positioning coordinates and category a i (p): Of i (p)=ming(p); S1-1-4. Determine g j (p)≤l; If the distance is less than a certain distance, the positioning coordinate x is classified into this category and the match is successful, otherwise the decision is rejected.
4. The virtual production process control optimization method based on digital twin according to claim 1 is characterized in that: The digital twin model described in step S2 is constructed in multiple ways; Trained through different machine training data groups; the data set not used for training is the validation set; through the validation set and reasonable validation indicators, the quality of different digital twin models is compared and the optimal digital twin model is found. The optimal digital twin model is combined with real-time lloT data for simulation and optimization, and an optimization plan is proposed for production line control.
5. The virtual production process control optimization method based on digital twin according to claim 4 is characterized in that: The true and absolute values of FE are displayed through curves to compare the efficiency of different digital twin models: Where y1 is the true value of the training data set, It is the predicted value calculated by the training model. The smaller the absolute value of FE is, the higher the efficiency of the current digital twin model is. The digital twin model with the highest efficiency is selected as the optimal digital twin model.
6. The virtual production process control optimization method based on digital twin according to claim 1 is characterized in that: The production process control optimization indicators in step S3 include model accuracy MAR, root mean square error RMSE, and variance explanation rate VIR; The expression of the model accuracy rate MAR is as follows: Where y1 is the true value of the validation data set, is the average value, is the predicted value calculated by the trained model; The model accuracy rate MAR is used to reflect the deviation between the predicted result and the actual value. The larger the MAR, the better, and the maximum value is 1.
7. The virtual production process control optimization method based on digital twin according to claim 6 is characterized in that: The expression of the root mean square error RMSE is as follows: Where y1 is the true value of the validation data set, is the average value, is the predicted value calculated by the trained model; RMSE is based on the quadratic distance between the actual value and the predicted value and is used to evaluate the performance of the regression model; the smaller the RMSE, the better, and the minimum value is 0.
8. The virtual production process control optimization method based on digital twin according to claim 6 is characterized in that: The expression of the variance explanation rate VIR is as follows: Where y1 is the true value of the validation data set, is the average value, is the predicted value calculated by the trained model; The VIR index is used to evaluate the degree to which the model prediction is better than the random prediction: when the value is greater than 0, the model prediction is better than the random prediction; the larger the VIR, the better, and the maximum value is 1.
9. A visualization system for running the virtual production process control optimization method based on digital twins as described in any one of claims 1 to 8, characterized in that: The visualization system includes: A virtual production process construction unit is used to collect various data in the actual production process and construct a virtual production process; A digital twin model construction unit, used to construct a digital twin model, use the digital twin model to control the virtual production process, and collect process data in real time; The optimization result output unit is used to perform a dimensionality reduction operation on the process data, calculate the production process control optimization index based on the process data after dimensionality reduction, and evaluate the quality of the digital twin model by the production process control optimization index: If the quality of the current digital twin model meets the predetermined value, the current digital twin model is retained as the control optimization basis of the virtual production process; If the quality of the current digital twin model does not meet the predetermined value, the current digital twin model is discarded and a new digital twin model that meets the conditions is iterated again; AR devices use a visual interface to help users select the scope of production services, and push and respond to production service feedback to users in a timely manner.
10. A computer-readable storage medium, characterized in that: The storage medium stores at least one executable instruction. When the executable instruction is executed on the electronic device, the electronic device executes the virtual production process control optimization method based on digital twins as described in any one of claims 1 to 8.
Citation Information
Patent Citations
Ship digital workshop simulation method and system based on digital twinning
CN113887016A
Production line virtual reconstruction and optimization method based on digital twinning
CN115688370A
Digital twinning system and method for personalized product-in-process customization production line
CN116523119A
Industrial production system based on digital twinning
CN116700162A
Rapid construction method of digital twinborn model in manufacturing process
CN116822353A
