“Am I the one blocking you?”
“I might never come back again”
“Every challenge you don’t try is a failure, every opportunity you hesitate is a waste.”
- From the lyrics of Golden Girls' One Last Time -

■ Can you succeed without failing?

Failure is the mother of success, so success is the child of failure. It means that you cannot succeed without failure. The opposite of success is not failure, but not trying. Challenge and experiment are actually doing it. Even advanced countries and the world's leading companies have gone through a lot of trial and error and failure behind their success.

Our country overcame the difficult circumstances of losing our country and fighting wars to become an advanced country. It is the result of many sacrifices made by uniting with the belief that “We can do it. We can do it” and the thought of “Let’s live well” and trying to think of the country rather than individuals, diligently and sincerely, following the advanced countries. It is a miracle. However, the economic growth rate has been continuously declining for the past 30 years. Everyone has different thoughts. The success formula of the past, which was to follow others as quickly as possible, is no longer valid. In the manufacturing industry, we are at an inflection point where we must lead R&D and manufacturing service transformation from a manufacturing-centered approach. If we do not lead, we may fall.

If the culture and work-processing methods of worrying about being audited in advance while only following the rules without trying, following orders from superiors without asking why, and ignoring things that are not their own work do not change, it is impossible to predict what will happen. Furthermore, the sloppy innovations that try to solve problems without defining the problems, as advanced countries do, are becoming more dangerous.

In addition, the national R&D budget, which is called 'blind money?', has been drastically cut to the point that the saying 'the first person to see it is the owner, and the one who doesn't eat it is a fool' has become common. How many R&D projects are there that claim to develop world-class technologies, and how many technologies are currently developed at the highest level? The success rate in our country's R&D field is over 95%. Because we are in a structure that does not tolerate failure, we end up only executing plans that are feasible from the start.

■ Innovation is needed

It is a good strategy to try to innovate through digital transformation. If you convert existing systems to digital, you can experiment in real life, so you can freely try and make mistakes. Innovation is very risky, expensive, and very difficult due to time and space constraints. You may fail. You may make mistakes. There may be a better way.

This is why virtual experiments are necessary.

■ Virtual experiment, “What if ...?”

Virtual experiments are a powerful tool for finding answers to the question “What if…?” Virtual experiments are about predicting and simulating what could happen under certain conditions before conducting a real experiment. This allows us to check possible outcomes in advance, minimize risks, and save time and money before conducting a real experiment.

Virtual experiments are used in various fields. For example, they are used to solve industrial problems by analyzing, predicting, optimizing, and automatically controlling to create and verify new products, processes, or systems, or to operate and maintain products, processes, or systems. They can also be used for pre-verification, analysis, prediction, optimization, and real-time command and control or decision-making before creating policies, laws, and systems to solve social problems with complexly intertwined interests.

Virtual experiments can greatly help us answer “What if…?” questions and explore new ideas. “What if ...?” questions are questions that stimulate imagination and help generate new ideas. They encourage thinking about what could happen in a given situation or hypothesis.

The purpose of experiments is to verify hypotheses, understand phenomena or explore principles, develop new products or technologies, or find causes and possible solutions when problems occur. Experiments are a core element of scientific methodology and play an important role in helping us gain new knowledge and solve problems. However, many experiments are difficult or impossible to conduct due to practical constraints such as time, space, cost, and safety. That is why virtual experiments through digital transformation are necessary. Just as experiments require a subject, virtual experiments require a digital twin of the subject. A digital twin that fits the purpose of the experiment must be created. If the digital twin does not behave exactly like the subject, virtual experiments are worse than no virtual experiments.

A test is about matching the correct answer, but an experiment is a process of finding an unknown answer. The reason for digital transformation is innovation, and for innovation, we need virtual experiments that can answer not only the 'first question' but also various What if questions. If we learn from the past, we can predict the future, but if something that has never happened in the past happens, we are helpless. What should we do if something that has never happened in the past happens? We can find the answer by conducting virtual experiments based on digital twins.

It is clear that generative AI is a useful tool for generating information and knowledge, but causal relationships in the dark are unknown, such information and knowledge are overflowing, it is difficult to distinguish between real and fake, and the speed of diffusion is instantaneous. The world could easily become a confusing place.

This is an era that requires wisdom to solve problems wisely beyond information and knowledge.

To gain wisdom, we need to go beyond learning and engage in virtual experiments that answer the question “What if…?”

■ Virtual Experiments and Digital Twins

The limitations of AI, which machine-learns the correlation of data observed as a phenomenon in data science, are clear. Beyond learning, experiments on situations that may occur in the future are necessary. Virtual experiments are a good alternative to solve these needs and transform from a fast follower to a first mover. Is there a better alternative?

Virtual experiments are designed to design new products, processes, or systems, solve problems that cannot be solved with existing systems, or provide innovative services through experiments that cannot be performed or are difficult to perform due to realistic constraints such as time, space, cost, and safety. Virtual experiments allow trial and error or failure at will, and enable optimal decision-making through verification, analysis, prediction, and optimization.

In order to conduct a virtual experiment, the experimental subject must be virtualized, and the digital replica of the experimental subject is the digital twin (DT: Digita Twin). Digital transformation (DX: Digital Transfomation) is to digitally virtualize the entire business process. It is to build a digital environment where virtual experiments are possible by digitally virtualizing company-wide work/production technology/products, and to maximize profits through increased work/production efficiency and product added value based on virtual experiments. DT and virtual experiments are the core technologies of DX, and are 'living digital simulation models' that are linked with real systems (Physical Systems) to enable virtual experiments that are difficult or impossible to perform in reality.


As seen in the figure above, there are limits to analyzing or predicting future changes using only the present and past big data secured in the actual system, so by conducting virtual experiments on scenarios that may occur in the future based on digital twins, we can secure big data that cannot be secured in the actual system. Based on the big data secured in this way, we can find answers to the What-if questions of digital twin users. What-if questions are things that cannot or are difficult to ask in the actual system due to constraints such as time, space, cost, and safety, and digital twins are necessary to conduct virtual experiments based on this and ultimately make optimal judgments and decisions through analysis, prediction, and optimization of the actual system.

What makes digital twins different from existing simulation technologies is that they are linked to and live together with real systems throughout the entire life cycle of the system, and they must be created and maintained to ensure consistency and homogeneity with the real system, and they must learn and evolve to be maintained. If digital twins do not ensure consistency and homogeneity with the real system, the meaning of virtual experiments may be halved or worse than not doing them at all. In the defense sector, which has introduced and utilized simulation technology early, the argument that models are useless is raised because they do not match reality well. This is also why guidelines for utilizing digital twins for weapon systems were recently established to improve this.

■ Digital Twin-Based Systems Engineering Approach

A system is a collection of components that come together to satisfy functions/performance that individual components alone could not. The Earth we live on is also a system. As shown in [Figure 2], numerous systems on Earth, including the Earth, are complex systems (System of Systems) in which numerous people (People), facilities and equipment created by people, HW, SW, data, materials, and organizations, numerous products (Products), and processes (Processes) such as laws, systems, services, and operational technologies organically interact and move based on the natural environment. A digital twin is a digital model that virtualizes a system digitally, and can be composed of various combinations of digital twins of the three components of people, products, and processes depending on the purpose.


By approaching complex and difficult problems from a systems engineering perspective through virtual experiments based on these digital twins (models), it is possible to simplify complex and difficult problems, create new systems, analyze/predict and optimize future changes, and easily find the optimal solution to the problem. If you know two of the three - input, system, and output - you can figure out the other one.


In order to provide services through virtual experiments, various technologies must be integrated and utilized. Virtual experiments can be divided into services that utilize digital twin modeling, simulation, and simulation results. Using modeling simulation technology, it is possible to create a digital twin, simulate various scenarios based on the digital twin, and visualize, analyze/predict/diagnose based on simulation result data by integrating Data Analytics/AI technology, and optimize and control the experimental target system in real time by integrating optimization/CPS technology. In addition, it can be integrated with VR/AR/metaverse technology and utilized for virtual experience purposes such as entertainment, tourism, and education and training.


■ Virtual Experiment Platform WAiSER

Attempts are being made to build a virtual experiment platform to address the needs of digital twin-based virtual experiments, but the fidelity of the model is low due to a lack of modeling technology, and is biased toward the purpose of experience. For the purpose of problem solving, it is built and operated for a limited purpose in a specific field, but even then, it depends on foreign SW, and since it is not linked to the real system, it is used only for designing/analyzing the system, or the simulation time is long, making real-time analysis and optimization difficult.

The Korea Digital Twin Research Institute has inherited and developed the 40-year research achievements of Professor Tak-gon Kim of KAIST and the technology and research achievements of SMS Lab, commercializing it under the name of 'WAiSER', a digital twin-based virtual experiment platform, and launched it in 2020. Based on WAiSER, various solutions are being developed and complex and difficult social and industrial problems are being solved. WAiSER carries the meaning of going beyond the limitations of AI and solving problems more wisely (Wiser).

WAiSER is the world's first domestically produced digital twin-based virtual experiment platform based on BAS (Big data+AI+Simulation) modeling technology that complementarily fuses simulation technology with simulation technology to overcome the limitations of AI based on big data learning. WAiSER is composed as shown in the figure below.


As shown in the figure above, WAiSER is not limited to a specific field or area, and is a powerful virtual experiment platform that can be utilized in all fields and areas that require virtual experiments, and can be linked with any system that complies with international standards. This is possible because continuous-time systems such as products or natural phenomena can be modeled using differential equations, and discrete event systems such as people and processes can be modeled using DEVS (Discrete EVent Specification) equations. Depending on the need, data models, physics/engineering models, or BAS (Big data+AI+Simulation) models can be selectively/combinedly utilized. In addition, it has been developed to be able to link with any system, simulator, or digital twin that complies with international standards.

■ WAiSER-based wisdom service

If WAiSER is built with not only virtual experiments but also the existing IoT/big data/AI platforms and GIS/BIM/CAD, as well as the configuration management platform, and WAiSER as a PoP (Platform of Platforms) concept, it will be possible to provide smart services that were not possible before. Wisdom services are possible beyond data, information, and knowledge. Information services about what, who, when, and where, knowledge services about how to do something, and wisdom services that can find the optimal solution for the purpose and solve problems wisely are possible.


We are working with companies that provide solutions or services for existing IoT/big data/AI platforms or GIS/BIM/CAD platform companies, or system integration companies, to support the provision of services at a level of wisdom that was not previously available. We look forward to working closely with professional solution companies and system integration companies, and hope that this article will serve as a catalyst for collaboration and improved service levels.

■ Virtual Experiment Procedures and Methods

In order to do well in virtual experiments, you need to know M&S (Modeling Simulation) engineering. Trying to do virtual experiments without knowing M&S engineering is like a person who majored in mechanical engineering trying to make an electronic circuit. It is urgent to train M&S experts through separate education.

The most important step for conducting a virtual experiment (simulation) is to set the purpose. This is because even if the target system is the same, the model varies depending on the purpose, and the final result also requires interpretation of the analysis index that fits the purpose. The requirements analysis and specification process must be conducted in a goal-oriented manner. The requirements specification process is a high-level model design.


After designing and implementing a model from the stated requirements, the model is considered to accurately (faithfully) represent the target system after going through the verification and validation processes. The verified/validated model is simulated according to an experiment plan appropriate to the purpose, data is collected, and the results are analyzed, and if necessary, it can be certified by an authorized institution.

Looking at related academic fields, the first is requirements engineering, which is an academic field that deals with analysis, specification, and verification of requirements. Modeling theory is the academic discipline that studies mathematical or non-mathematical frameworks for modeling target systems, as well as algorithmic modeling techniques that do not use frameworks.

In simulator implementation, we study how to efficiently execute a given model in terms of time (simulation execution) and space (program capacity). Hypothesis verification is used to validate a model by comparing the model's simulation data with data collected from an actual system based on statistical inference used in statistics. Performance engineering is an academic field that estimates statistical parameters from data collected from simulation results or measures and compares and analyzes performance indices.

M&S engineering is a multidisciplinary field of study that combines engineering, physics, computer science, statistics, and OR (Operational Research), rather than an independent academic field like electrical engineering, mechanical engineering, and computer science. Due to the interdisciplinary nature of M&S engineering, a group of experts from various fields of study participate in the M&S process. Domain experts have specialized knowledge about the target system. For example, if the modeling target is fighter maneuver, an aeronautical engineering major is a domain expert, and if the modeling target is a communications network, an electronic engineering/computer science major is a domain expert.

M&S specialists have expertise in the process of designing, implementing, and executing models, and ICT (information and communication technology) specialists involved in implementation are called platform specialists. Of course, depending on the target system, one person may play the role of domain specialist, M&S specialist, and platform specialist. Finally, OR/statistical analysis specialists design simulation experiments according to the design of experiments for analysis appropriate to the purpose and analyze the simulation results.

In order to successfully solve problems through virtual experiments, a cooperative system of three expert groups is required, as shown in the figure below. The expert group consists of domain engineers, M&S engineers, and platform engineers. In order to define and formalize the problem, collaboration between domain experts and M&S experts who know how to solve the problem is necessary to create a digital twin (model) of the experimental subject. In order to implement the model on a computer and conduct a good virtual experiment, collaboration is necessary between M&S experts and experts in programming technology to implement and simulate the designed model, software engineering to manage the model development process, and ICT technology (GIS, DB, networking, etc.) required for simulation.

Through the collaboration between domain experts and M&S engineers, the functional/operational requirements analysis and model logic of the model to be developed are specified. Through the collaboration between M&S engineers and ICT engineers, the model is designed and implemented. Through the collaboration between ICT engineers and domain experts, the operation and user interface of the implemented model are discussed. If this collaboration system is established, complex and difficult problems can be simplified and easily solved through virtual experiments.


■ Conclusion and Suggestions

The generative AI craze triggered by ChatGPT is creating a huge wind and is accelerating the popularization of information and knowledge. However, in order to break away from following others and lead, we must use it as a signal to open the 'age of wisdom' beyond the information and knowledge age. With the acceleration of digital transformation, data is exploding and information and knowledge are overflowing. In an environment where we cannot help but be exposed to unnecessary information, we are wasting time and energy due to information overload and losing focus, making it difficult to live an independent life. In addition, the digital nature makes it easy to copy and forge, making it difficult to distinguish between the real and the fake, and the speed of diffusion is so fast that it can easily become a chaotic world.

It means that it is difficult to become a leading country or a digital hegemon with only AI-centered policies such as generative AI, and that digital innovation policies cannot succeed. So what should we do? We need a wisdom service that can develop intelligence through questions and learning, clarify the purpose, and optimize through virtual experiments to solve problems wisely. Real innovation is needed. We cannot do it in the existing way of quickly following others. Wisdom is more important than knowledge. Wisdom refers to the ability to solve problems wisely. Increasing wisdom is in line with the digital hegemony that the government is pursuing. Therefore, I propose several policy directions.

First, let's build a virtual experiment platform.

Second, let's discover current issues and innovative services. Rather than an approach where the ends and means are reversed, focusing on hyper-connectivity, hyper-intelligence, and hyper-realistic technologies, we should utilize the technology necessary to achieve the ends. Once the issues and services to be solved are identified, technology is no longer a problem.

Third, let’s try to apply the identified problem (service) on a virtual experiment platform. Among the problems (services) to be solved, those with high effectiveness should be selected, their effectiveness should be verified through pilot application, and then expanded.

Fourth, creation and activation of a digital platform ecosystem.

When keeping up with technology, the priority was to do things quickly, but now, it is necessary to create a sound ecosystem where each industry can coexist, and to support this, it is necessary to create and activate a digital platform ecosystem that allows sharing, communication, and collaboration anytime, anywhere.

Fifth, it is to promote mandatory prior verification and optimization of public policies. By making it mandatory to conduct prior verification and optimization using a virtual experiment platform before implementing public policies, trial and error can be minimized and the cause of political strife can be eliminated.

Sixth, we need to train experts in problem solving based on virtual experiments. As the world becomes more complex and the pace of change accelerates, it is urgent to train personnel who can define and solve problems from a systems engineering perspective that encompasses the humanities and engineering, rather than specialists in each field.

It is very difficult to predict change. In order to achieve what we want together and be happy and successful, rather than chasing change, we should explore the unchanging truth, determine the values and vision that should not change, properly create a digital twin, optimize it through virtual experiments, and execute it to lead innovation. This is an era that requires wisdom beyond information and knowledge. In order to seek wisdom, this is an era that requires collaboration beyond competition, virtual experiments beyond learning.

※ References
1. KAIST KOOC Course, System Modeling Simulation, https://kooc.kaist.ac.kr/isms1, 2019
2. KAIST KOOC Course, Digital Twin, https://kooc.kaist.ac.kr/DigitalTwin/, 2021
3. YouTube video, Wednesday Forum (KISTEP: Korea Institute of Science and Technology Evaluation and Planning) - 2020.12.9 “A virtual world that communicates with reality, digital twin development strategy,”
https://www.youtube.com/watch?v=k-yi9wOk89U&feature=youtu.be
4. Collaborative modeling methodologies for areas requiring domain knowledge:
http://smslab.kaist.ac.kr/paper/JF/JF-58.pdf
5. Kim Tak-gon, System Modeling Simulation, Hanti Media, 2020
6. Kim Tak-gon et al., Theory of Modeling and Simulation, 2000, Academic Press