Digital twin technology architecture
The digital twin reference architecture is mainly divided into two parts
: One part is the hardware mapping represented by digital twins, which is similar to the operating system core of a personal computer, that is, the interaction under the control of the driver of the hardware. Its performance index is expressed as "accuracy"; the other part is the digital twin platform and applications, which constitute the application environment and capabilities. Due to its open architecture characteristics, in addition to internal applications within the enterprise that can be developed on the digital twin platform, external certified third-party forces can also participate in digital twin application development. 
Some industry insiders believe that digital twins must be real-time control types, which actually ignores the difference between it and cyber-physical systems. The core function of a highly integrated cyber-physical system is real-time control, because it comes from embedded systems and is produced for real-time control. However, the digital twin forms a layered structure by decoupling traditional industrial systems, objectively It is said that it sacrifices certain performance in exchange for flexibility, which is one of the reasons why it has general-purpose technical characteristics.
Since it is an open architecture, the digital twin can more easily solve standardization problems through the autonomy of each module. As we all know, the standards war in the industrial field has lasted for decades and has not been effectively resolved. The fundamental reason is that everyone is unwilling to give up the barriers associated with their own non-standard solutions in order to maintain their competitive advantage. It can be seen that the use of traditional industrial automation methods is not a solution to data exchange and sharing, but is the reason for the slow progress of industrial digitalization.
Not only that, modularity is also an important feature in the platform strategy theoretical system. Any product or service can be regarded as a system composed of multiple modules or subsystems. The connection between modules or subsystems involves standards. Without an efficient standard system, it will be difficult for the system to evolve rapidly. The digital twin platform hides complex standards into capabilities, and uses personnel with professional skills to reconstruct these complex systems so that they can all be connected to the digital twin standard interface. As the platform matures and the number of stakeholders increases, new developers joining the platform ecosystem will be motivated by interests to follow existing standards.
It can be seen that open architecture, as the gene of the digital twin, constitutes its core value, which is also one of the characteristics that distinguishes it from other conceptual systems. With the help of the two-sided market economics principles contained in the digital twin platform, the network effect that everyone desires will be generated, allowing first movers and companies that dare to innovate to gain industry competitive advantages, and ultimately form a prosperous digital twin industry.
Digital twin applications
Driven by the open architecture of digital twins, digital twin applications have become the vitality of the digital twin platform ecosystem. Traditional industrial systems usually only serve the providing companies themselves. If the customer company needs to expand, the provider needs to customize it. Such a market structure helps stabilize the industry structure, but is not conducive to the rapid advancement of industry innovation, Annabeth La Gavel and others pointed out that this is a characteristic of innovation platforms.
Research on platform ecology theory shows that it benefits from the success of personal computers, and many theories are abstracted from this industry. Following the successful demonstration of personal computers, the mobile Internet pushed the theory further. Apple built an app store through a closed hardware system and "forced" developers to accept a 30% app tax through an excellent user experience, making Apple a lot of money. Boman, mobile application revenue in the third quarter of 2020 reached US$19 billion. Correspondingly, Android application revenue was US$10.3 billion in the third quarter of 2020.
Driven by the potential value of platform application ecology, many companies have invested heavily in building platforms in the hope of building scale in the future and ultimately profiting from application revenue. General Electric also adopts this approach in its strategy to promote the Predix platform ecosystem. General Electric's performance report shows that it spent US$500 million in Predix construction investment in 2015 and as high as US$700 million in 2017. Such investment naturally produced results. By December 2017, General Electric already had 22,000 US dollars. developer. At this time, it already has the necessary resources for the developer community: documentation, guides, training, APIs, development simulators and virtual machines, etc.
In order to help developers obtain necessary help, GE has also established Digital Foundry (Digital Foundry) in Paris, Munich, Shanghai, Boston, Singapore and other places. According to the Industry 4.0 Research Institute’s follow-up research, General Electric is even considering setting up a system similar to Microsoft Developer Network (MSDN) to introduce relevant certifications to enhance the value of Predix platform developers. However, due to GE's operating problems in recent years and its related budget cuts, this idea has not yet been realized.
In General Electric's Predix ecological development plan, its key performance indicator is industrial applications, as shown in Figure 5-2. In the large-scale development stage, it is required to develop more than 100 industrial applications; in the platform development stage, the number of industrial applications is required to be more than 1,000; in the end, the Predix platform will host more than 100,000 industrial applications. Data released by General Electric in 2019 shows that the number of industrial applications on the Predix platform has exceeded 10,000, which means that it has reached its predetermined goals and exceeded its development goals.
Even so, GE is unlikely to succeed on the platform overnight, and application development is a very challenging matter. In the early days, there seemed to be a large number of applications, which were mostly converted from existing businesses or obtained through the support of close partners. It was not an organic growth.
Digital twin applications follow similar rules and face two major problems.
One is the killer application. Any platform ecosystem has solutions that meet industry commonalities, and these solutions are embodied as killer applications.
The second is the “chicken and egg paradox”. Only when the number of digital twin applications on the platform reaches a certain number will developers continue to be attracted to the platform.
Let’s start by looking at the killer app puzzle. For any industrial platform, it must solve the common problems of the industry, so that it can have industry characteristics. In the past 20 years, scholars who study platform ecology have reached a consensus: only platforms that solve common problems in the industry can be called industrial platforms. The Industry 4.0 Research Institute analyzed the platform construction situation of nearly a hundred digital technology companies and found that most companies are unaware of or unable to solve common industry problems, which ultimately leads to the failure of their platforms to be transformed into industrial platforms.
Simply put, killer applications solve common industry problems and usually have rigid demand characteristics. The killer application of the digital twin platform may be asset management. According to the requirements of the five levels of digital twins, the first-level geometric model constitutes the basic characteristics of digital assets, although the data accuracy represented by this level of digital twins is not high. , but it has exceeded the object-level accuracy processed by the traditional Internet and has very high value in some fields.
When General Electric Digital introduced the application of digital twins, it also focused on asset management. In the fields of digital twin cities, digital twin buildings, digital twin aerospace and other fields, its essence is also reflected in asset management. Even in currently visible cases, more than 90% of applications only realize visual asset management. These applications are not useless. Their visual effects can simplify management difficulties and confusion, and also help promote the establishment of a single data source between various devices or subsystems - because the first level of digital twins implements geometric models.
The smart pole digital twin solution provided by Yiluo Digital demonstrates the skills of digital twin application development. First, establish a geometric model for the smart pole. This part of the work is relatively simple and helps solidify the relevant standard system. Second, provide simulation data descriptions for key components, including materials, manufacturing standards and other data. These data can be provided in reverse to Suppliers in the smart pole industry chain; third, promote the data fusion of multi-scale scenarios based on the scale requirements of digital twin cities. For example, combining GIS and BIM data to form a digital twin city solution.
Although only the first-level requirements of digital twins have been achieved so far, it has already shown its rigid value. In some application scenarios, it may be the only requirement at the moment. Therefore, asset management is the killer application of the digital twin platform. Generally speaking, when promoting the construction of digital twin applications, the killer application of asset management should be built first. Of the U.S. Navy’s $21 billion digital twin project investment, tens of billions of dollars are used to establish the digital twin infrastructure of the shipyard, the core of which is the digital twin of various equipment assets. The national digital twin strategy promoted by the United Kingdom and Singapore is directly a digital twin city application. The core is to establish digital models for various buildings, blocks, parks, etc.
In addition to killer applications, digital twin applications also have the "chicken and egg paradox". I have just explained the value of a killer application, which can attract users to use the platform. However, if you want to continue to attract users to use platform services, you need to have continuous new applications. This is how Apple and Google's app stores maintain their platform leadership. of status.
When developers choose a platform to host innovative applications, they usually pay more attention to the number of customer groups on the platform. The increase in the number of customer groups usually depends on the number of platforms. The two are cause and effect on each other, so it is called "chicken and egg". paradox". Solving this dilemma requires continuous efforts by platform operators. In the early stages of development, major platforms will find ways to attract developers. The aforementioned General Electric established digital workshops around the world for this purpose.
However, a large number of cases of successful platform operation show that the key elements to solve the "chicken and egg paradox" and establish a virtuous cycle of developer innovation ecosystem include simple development languages, rich SDKs, active promotion, etc. In order to reduce the difficulty of the development process, both Apple and Google have launched their own development languages; at the same time, in order to prevent developers from duplicating low-level development, they have provided rich SDKs; these platform operators also focus on supporting typical applications to help them Earn profits, which can attract some gold diggers to enter and enhance the activity and value of the developer innovation ecosystem.
Currently, the more successful developer innovation ecosystems are in the field of consumer Internet, and there are no successful examples of digital twin application ecosystems. The reason is that the so-called large-scale market demand and developers are just fantasies in the industrial field currently known for non-standard products, unless the problem of non-standard market definition can be solved. If the digital twin perspective can bring consistent market understanding to the industrial field, it can solve the problem of digital twin application ecological construction.