The following article comes from Zhizaoyuan, the author Xiaozhi
Author: Li Peigen, Gao Liang | Source: Zhizao Yuan, adapted from: "Introduction to Intelligent Manufacturing"
「 1. The concept of digital twins"
Gartner, the world's most authoritative IT research and consulting company, regards digital twin as one of the important technologies in the top ten strategic technology development trends in 2019, and its description of digital twin is: digital twin is the digital embodiment of real-world entities or systems.
Figure 1 The conceptual ideal of PLM, the original conceptual model of the digital twin and its terminology
There are many definitions of digital twins. Professor Tao Fei believes in the review of Nature that digital twins, as a key way to realize two-way mapping, dynamic interaction and real-time connection between virtual and real, can map the attributes, structure, state, performance, functions and behaviors of physical entities and systems to the virtual world, forming a high-fidelity dynamic multi-dimensional/multi-scale/multi-physical model of the physical world, which provides an effective means for observing, understanding, understanding, controlling and transforming the physical world.
CIMdata recommends the definition: "A digital twin (i.e., digital clone): is a physical entity-based description of a system that enables the creation, management, and application of data, models, and information from trusted sources across the entire system lifecycle." This is a simple definition, but without a true understanding of the keywords (system description, lifecycle, trusted source, model), it can be misleading.
「 2. Digital Twin Model"
1) Conceptual model of the digital twin
Based on the literal definition of the digital twin, Figure 2 shows a five-dimensional conceptual model of the digital twin.
Figure 2 Five-dimensional conceptual model of the digital twin
The 5D Conceptual Model of the Digital Twin is a universal reference architecture that can be applied to different applications in different domains. Secondly, its five-dimensional structure can be integrated and integrated with new information technologies such as the Internet of Things, big data, and artificial intelligence, to meet the needs of cyber-physical system integration, cyber-physical data fusion, and two-way connection and interaction between virtual and real. Thirdly, the integration of twin data (DD) integrates information data and physical data to meet the consistency and synchronization requirements of information space and physical space, and can provide more accurate and comprehensive data support for all elements, all processes, and all services.
Service (SS) encapsulates all kinds of data, models, algorithms, simulations, and results required by different fields, different levels, and different businesses in the process of digital twin application, and provides them to users in the form of application software or mobile app, so as to realize convenient and on-demand use of services. Connectivity (CN) enables ubiquitous industrial interconnection between physical entities, virtual entities, services, and data, thereby supporting real-time interconnection and convergence of virtual and real realities. Virtual entities (VEs) depict and describe physical entities from multiple dimensions, spatial scales, and time scales.
2) The system architecture of the digital twin
Figure 3 illustrates a common reference architecture for a digital twin system. A typical digital twin system includes five levels: user domain, digital twin, measurement and control entity, physical domain and cross-domain functional entity.
Figure 3 Common reference architecture for digital twin systems
3) Maturity model of the digital twin
The digital twin is not only a mirror image of the physical world, but also accepts real-time information from the physical world, and in turn drives the physical world in real time, and evolves into a prophet, seer and even superbody of the physical world. This evolution process is called maturity evolution, that is, the growth and development of the digital twin will go through several processes, such as digitization, interaction, prophet, seer, and common intelligence (Figure 4).
Figure 4 Digital twin maturity model
(1) Digitization. Digitization is the process of digitizing the physical world. This process entails the representation of physical objects as digital models that computers and networks can recognize. Modeling technology is one of the core technologies of digitalization, such as surveying and mapping, geometric modeling, mesh modeling, system modeling, process modeling, organization modeling and other technologies. The Internet of Things (IoT) is another core technology of "digitalization", which changes the state of the physical world itself into something that can be perceived, recognized, and analyzed by computers and networks.
(2) Interaction. Interaction mainly refers to real-time dynamic interactions between digital objects and their physical objects. The Internet of Things (IoT) is the core technology that enables the interaction between the virtual and the real. One of the responsibilities of the digital world is to predict and optimize, and at the same time intervene in the physical world based on the results of the optimization, so instructions need to be passed to the physical world. The new state of the physical world needs to be transmitted to the digital world in real time, as a new initial value and a new boundary condition for the digital world. In addition, this interaction includes interaction between digital objects, which relies on digital threads to achieve this.
(3) Prophets. Prophetic refers to the use of simulation technology to make dynamic predictions of the physical world. This requires digital objects not only to express the geometry of the physical world, but also to integrate physical laws and mechanisms into the digital model. Simulation technology not only establishes a digital model of physical objects, but also calculates, analyzes and predicts the future state of physical objects through physical laws and mechanisms according to the current state.
(4) Foresight. If the "prophet" predicts the future of the digital twin based on the definite laws and complete mechanisms of physical objects, then the "prophet" is based on incomplete information and unclear mechanisms, and predicts the future through industrial big data and machine learning technology. If digital twins are to be more and more intelligent and intelligent, they should not be limited to human deterministic knowledge of the physical world, because human beings themselves do not rely solely on deterministic knowledge to comprehend the world.
(5) Common wisdom. Co-intelligence is to realize the wisdom exchange and sharing between different digital twins through cloud computing technology, and its implicit premise is that the wisdom of each component within a single digital twin is first shared. The so-called "single" digital twin is an artificially defined scope, and multiple digital twin cells can form a larger and higher-level digital twin through "common intelligence", and this number and level can be infinite.
「 3. Key Technologies for Digital Twins"
Modeling, simulation, and digital threads based on data fusion are the three core technologies of digital twins.
1) Modeling
Digital modeling technology originated in the 50s of the 20th century, and the purpose of modeling is to simplify and model our understanding of the physical world or problems. The purpose or essence of digital twins is to eliminate the uncertainty of various physical entities, especially complex systems, through digitization and modeling. Therefore, the establishment of a digital model or information modeling technology of a physical entity is the source and core technology for creating and realizing digital twins, and it is also the core of the "digitalization" stage.
The model development of digital twins is divided into 4 stages, which represent the general understanding of the development of digital twins in the industry, as shown in Figure 5.
Fig.5 Four stages of digital twin model establishment
The first stage is the physical model stage, and there is no virtual model to correspond to it. NASA builds a mock-up of the spaceship's twin on the ground during the spacecraft's flight. This life-size model played a key role in saving Apollo 13.
Phase 2 is a virtual model where the physical model has its counterpart partially implemented, but there is no data communication between them. In fact, this stage cannot be called the stage of digital twin, it is generally accurate to say that it is a digital model of the physical object. In addition, although there is a virtual model, this virtual model may reflect all the entities derived from it, such as the 2D/3D model of the design result, which also uses digital form to express the physical model, but the two are not directly corresponding to the individual.
The third stage is in the life cycle of the entity model, there is a corresponding virtual model, but the virtual model is partially realized, which is like the shadow of the physical model, which can also be called the digital shadow model, and there can be limited two-way data communication between the virtual model and the solid model, that is, the entity state data collection and the virtual model information feedback. The current digital twin modeling technology can better meet the requirements of this stage.
The fourth stage is the full digital twin stage, that is, the physical model and the virtual model are completely one-to-one. The virtual model fully expresses the physical model, and the integration between the two is realized, and the self-awareness and self-disposal between the virtual model and the physical model are realized, and the state between them can be synchronized with real-time fidelity.
It is worth noting that sometimes there can be a virtual model before the physical model, which is also an advanced stage of the application of digital twin technology.
A physical entity does not correspond to just one digital twin and may require multiple digital twins that are described from different perspectives or perspectives. It's easy to think of a physical entity as a digital twin. If it's just geometric, this argument can still be true. Precisely because people need to understand the different physical processes in different stages of the entity and in different environments, a digital twin is obviously difficult to describe. For example, the vibration deformation, thermal deformation, and interaction between the tool and the workpiece during the processing of a machine tool...... These situations naturally require different digital twins to describe.
A digital twin that describes a physical entity from a particular perspective by different modelers should seem the same, but can actually vary greatly. A single physical entity may correspond to multiple digital twins, but the digital twin from a particular perspective may seem unique, but it is not. Differences are not only the expression of the model, but more importantly, the granularity of the twin data. For example, in the so-called intelligent machine tools, people usually obtain real-time data on processing size, cutting force, vibration, temperature of key parts and other aspects through sensors, so as to reflect the processing quality and machine tool operation status. Different modelers certainly make different trade-offs about the data. In general, fine-grained data provides a deeper understanding of physical entities and how they operate.
2) Simulation
From a technical point of view, modeling and simulation are a pair of companions: if modeling is modeling our understanding of the physical world or a problem, then simulation is about verifying and confirming the correctness and validity of that understanding. Therefore, the simulation technology of digital model is the core technology to create and operate digital twins and ensure that the digital twin and the corresponding physical entity achieve an effective closed loop.
Simulation is a technique that simulates the physical world by converting a model containing deterministic laws and complete mechanisms into software. As long as the model is correct and has complete input information and environmental data, it can basically correctly reflect the characteristics and parameters of the physical world.
Simulation emerged in the industrial field, as an indispensable and important technology, has been widely used in various fields of industry by many enterprises in the world, is the core technology to promote the rapid development of industrial technology, is one of the most important technologies in the era of Industry 3.0, and plays an indispensable role in product optimization and innovation activities. In recent years, with the rise of a new round of industrial revolution such as Industry 4.0 and intelligent manufacturing, the combination of new technologies and traditional manufacturing has given birth to a large number of new applications, and engineering simulation software has also begun to combine with these advanced technologies to play a more important role in R&D and design, manufacturing, test operation and maintenance.
With the development of simulation technology, this technology has been adopted by more and more fields, and more types of simulation technology and software have gradually developed.
For the industrial manufacturing scenarios closely related to digital twins, we summarize the simulation technologies involved as follows (Figure 6):
(1) Product simulation, such as system simulation, multi-body simulation, physical field simulation, virtual experiments, etc.;
(2) Manufacturing simulation, such as process simulation, assembly simulation, CNC machining simulation, etc.;
(3) Production simulation, such as discrete manufacturing plant simulation, process manufacturing simulation, etc.
(a) Aerodynamic simulation of aircraft
(b) Plant simulation
Figure 6 Simulation example in a manufacturing scenario
Digital twins are the new pinnacle of simulation applications. Simulation plays an indispensable role at every stage of the maturity of digital twins: the core technology of "digitalization" - modeling is always associated with simulation or a part of simulation; "interaction" is a common scene in semi-physical simulation; the core technology of "prophet" is simulation; many scholars regard industrial big data, the core technology of "vision" as a new simulation paradigm, and "common intelligence" It is necessary to couple multiple disciplines between different twins to make ideas collide and generate sparks of wisdom. Digital twins have also become the source and core of intelligence and intelligence because simulation is ubiquitous at different maturity stages.
3) Digital threads
One concept that is closely tied to the digital twin is the digital thread. The premise of digital twin application is the model and a large amount of data in each link, so how to generate, exchange and flow data similar to product design, manufacturing, operation and maintenance, how to achieve seamless flow of data between some relatively independent systems, how to connect the right information to the right place in the right way at the right time, how to trace the process of connection, and how to evaluate the effect of connection. These are exactly the problems that the digital thread is designed to solve. CIMdata recommends the definition: "The digital thread refers to a framework for information interaction, which can open up multiple silos-based business perspectives and connect the interconnected data flow and integrated view of the whole life cycle data of the device". The digital thread is supported by a robust end-to-end interconnected system model and model-based systems engineering processes, a schematic of which is shown in Figure 7.
Figure 7 Schematic diagram of the digital thread
A digital thread is a bridge between several digital twins that correspond to a physical entity or a class of physical entities that reflect different aspects of the model view of that physical entity. The relationship between the digital thread and the digital twin is illustrated in Figure 8.
Figure 8 The relationship between the digital twin and the digital thread
As can be seen from Figure 8, the mechanism or engine that can realize the data fusion of multi-view models is the core of digital thread technology. Therefore, in the conceptual model of a digital twin, the digital thread is represented as a combination of a model data fusion engine and a series of digital twins. The implementation of the digital thread in the digital twin environment has the following requirements:
(1) Be able to distinguish between types and examples;
(2) support requirements and their allocation, tracking, validation, and validation;
(3) support the actual state recording, correlation and tracking of the actual state between the model views of the system across time scales;
(4) support the correlation between the models of the system across time scales and the association of their time scale model views;
(5) record the various attributes and their changes over time and different views;
(6) Recording the process or action acting on and completed by the system;
(7) record the purpose and attributes of the enabling system;
(8) Record documents and information related to the system and its enabling systems.
Digital threads must use some "common language" throughout their lifecycle in order to interact. For example, during the conceptual design phase, it is necessary for product engineers and manufacturing engineers to work together to create dynamic digital models that can be shared. Based on this model, the visualization process, numerical control program, acceptance specification, etc. required for the production process such as processing and manufacturing and quality inspection are generated, and the products and processes are continuously optimized and updated in real time. The digital thread is an effective way to assess the current and future capabilities of a system during its lifecycle, to identify system performance defects early through simulation, to optimize product operability, manufacturability, quality control, and to apply models throughout the lifecycle for predictable maintenance before product development.
「 4. Typical application cases of digital twins in intelligent manufacturing"
1) Digital twin design material dump
There are material storage yards in power plants, steel plants, and mines. Traditionally, when designing these dumps, the design requirements were artificially planned. After the construction and operation of the storage yard, it is often found that the design at that time could not meet the needs of the site. This gap can sometimes be very large and cause huge waste.
To meet this challenge, ABB used digital twin technology when designing the new material dump. From the beginning of the design requirements, designers use the historical operation data obtained by the Internet of Things to conduct big data analysis and optimize the requirements. During the design process, ABB developed a digital twin of the material yard with the help of CAD/CAE/VR technologies (Figure 9). The digital twin reflects in real time the parameters of material transport, storage, mixing, quality, and more as the environment changes. The design for this material field was not done all at once, but was finalized after many optimizations. During the optimization phase, the physics are run virtually in the digital twin. By reflecting the dynamic changes reflected in the operation, the problems that may arise after the operation are identified in advance, and the design is automatically improved. Through multiple iterations and optimization, the final design scheme is formed.
Fig.9 ABB uses a digital twin to design a material dump
Through the operation process, it was proven that the new concept designed by digital twin can better meet the needs of the site. And, in conjunction with the Internet of Things, the digital twin in the design phase continues to be used during the operational phase to continuously optimize the operation of the material yard.
2) Digital twin machine tools
Machine tools are important equipment in the manufacturing industry. With the improvement of customers' requirements for product quality, machine tools are also facing strict requirements such as improving machining accuracy, reducing defective rate, and reducing energy consumption.
In the EU-led European Research and Innovation Programme project, researchers have developed a digital twin of the machine tool to optimize and control the machining process of the machine (Figure 10). In addition to conventional model-based simulation and evaluation, the researchers use the developed tools to monitor the machining process of the machine and to exercise direct control. Improve the performance of your manufacturing process with model-based evaluation, combined with monitoring data. Increase productivity by controlling component optimization to maintain operations, improve energy efficiency, and modify process parameters, ensuring that critical machine components are kept in good condition until the next repair.
Fig.10 Digital twin machine
When building the digital twin of the machine, CAD and CAE techniques were used to build a machine dynamics model (Fig. 11), a machining engineering simulation, an energy efficiency model, and a key component life model. These models are able to calculate material removal rates and changes in the thickness of burrs, as well as predict prop failure. In addition to optimizing the chip forces during tool machining, the stability of the tool can also be simulated, allowing the machining process to be optimized. In addition, the model predicts surface roughness and thermal errors. The machine digital twin connects these models and measurement data in real time to aid decision-making in controlling the machine's operation. The monitoring system of the machine tool is deployed in the local system, and the data is uploaded to a data management platform in the cloud, where it is managed and run.
Fig. 11 Hydraulic control system of a digital twin machine
「 5. The Future Trend of Digital Twins"
Combined with the current development status of digital twins, digital twins will develop in three directions: simulation, full life cycle and integration in the future.
1) Pseudo-realization-multi-physics modeling
The digital twin is a true reflection of the physical entity in the virtual space, and the success of the application of the digital twin in the industrial field depends on the degree of realism of the digital twin, that is, the degree of simulation. Each physical property of a product has its own specific model, including computational fluid dynamics model, structural dynamics model, thermodynamic model, stress analysis model, fatigue damage model, and material state evolution model. Correlating these models based on different physical properties is key to building a digital twin and then making full use of the digital twin for simulation, diagnosis, prediction, and control. The simulation results based on the multi-physics ensemble model can more accurately reflect and mirror the real state and behavior of physical entities in the real environment, making it possible to replace the physical prototype in the function and performance of the product in the virtual environment, and at the same time, it can also solve the problems of timing and geometric scale in predicting the health and remaining life of the product based on traditional methods. Multi-physics modeling will be an important technical means to improve the degree of digital twin simulation and give full play to the role of digital twins.
2) Full life cycle - from the product design and service stage to the product manufacturing stage
Based on the new generation of information and communication technologies such as the Internet of Things, industrial Internet, and mobile Internet, the process data generated by the production site is collected and processed in real time, and these process data are correlated and mapped and matched with the digital twin of the production line, which can realize the refined control of the product manufacturing process online.
3) Integration – integration with other technologies
As the enabling technology of digital twins, digital thread technology is used to realize the two-way interaction of models and key data at all stages of the digital twin life cycle, and is the basis for the efficient collaboration of a single product data source and all stages of the product life cycle. The U.S. Department of Defense regards digital thread technology as the most important basic technology for digital manufacturing, and the Industrial Internet Alliance also regards digital thread as a key technology that it needs to focus on. At present, there are still breakpoints between product design, process design, manufacturing, inspection, use and other links, and the continuous flow of digital quantities has not been fully realized; although the emergence of MBD technology strengthens and standardizes the description of manufacturing information based on the three-dimensional model of products, it still mainly stays in the product design stage and process design stage, and needs to extend to the stages of product manufacturing/assembly, inspection, and use; and the digital flow at this stage is one-way, and digital thread technology is required to realize two-way flow. Therefore, the convergence of digital threads and digital twins is the future trend.
Transferred from the official account: the god of PLM