April 20, 2022
Digital twins are exciting because they bridge the physical and digital worlds. In the future, everyone and everything could have a digital twin. This will create a whole new world of possibilities for technological advancement and societal benefits, but let’s not give digital twins all the credit. They wouldn’t be complete or detailed without data collected from sensors, insights devised from that data, and communication protocols to transmit that data to display on our screens. This is where the Internet of Things (IoT), Machine Learning, and Artificial Intelligence (AI) come in.
Digital twin technology has been advancing rapidly over the past few years, and growing in popularity since Gartner included it as one of the top ten technology trends to know in 2017. Alongside it, the technology of Internet of Things (IoT) and Artificial Intelligence (AI) has also been steadily improving, and expanding our idea of what’s possible.
In this article we’ll define digital twins, Internet of Things and AI, and then take a look at how the three of them fit together in applied use cases. There are a myriad of use cases for digital twins combined with IoT and AI, and these use cases can apply to dozens of industries. Some of these industries include healthcare, manufacturing, real-estate and retail.
To provide an example, or use case, of how Digital Twins fit in with AI and IoT technology, we’ll be doing a deep dive into how the three technologies can be used in congruence with each other in smart, commercial buildings. In other words, how building managers and developers can use these technologies to control power systems, operate machinery, proactively plan for equipment service, predict equipment failure, and control building systems, such as lighting and HVAC.
According to a technology writer for IBM, a digital twin is “a highly complex virtual model that is the exact counterpart (or twin) of a physical thing”. This means that, it’s a digital representation of a physical object. For example, a digital twin could be a virtual model of a car. It’s significant to note that the more data collected about a physical object, the more detailed, comprehensive, and (ultimately) accurate the simulation of a physical object aka the digital twin can be. The data necessary to create a digital twin is collected in real-time via sensors that can measure many different aspects of a physical space or system, such as occupancy, use, temperature and more. We have articles about a couple of the most common sensors used to measure building comfort levels: the Indoor Environmental Quality (IEQ) sensor, and the occupancy sensor.
Artificial Intelligence, or AI, is a branch of computer science concerned with enabling machines to perform tasks and make decisions that usually require human intelligence and analysis skills. In short, AI transforms machines into smart machines with its ability to analyze collected data and provide actionable insights based on such data. A practical example of AI is simply that Netflix recommends what to watch next based on data collected from watch and search history on the platform.
There are 4 types of artificial intelligence, which are:
These 4 types of AI are essentially different levels that the technology strives to reach. Each level down requires more complicated technology, and improves the likeness of machine intelligence to human intelligence.
Artificial Intelligence: Put simply, AI is a technology that enables a machine to simulate human behaviour
Machine learning: A subset of AI that gives computers the ability to automatically learn from past information, even if they haven't been programmed to take certain pieces of data into account. This is a big component that could make computers more likely to pass the Turing test in the future, which is a fascinating concept on its own.
The Turing Test: A test devised by Alan Turing where the premise is that a judge interacts with both a human and a computer without the ability to see which is which (think chatting online, or texting), and then tries to identify which of the two is the human. If the judge is unable to correctly guess over 50% of their attempts, then the computer passes the Turing test and can be considered a passable simulation of a human, and therefore, intelligent.
Where AI is a technology that allows machines to simulate human intelligence and thinking capacity, the Internet of Things (IoT) is the technology that allows machines to communicate these data points and insights with each other.
The Internet of Things itself “refers to the billions of physical devices around the world that are now connected to the internet, all collecting and sharing data”. In order to include a “dumb” object in the IoT, it’s necessary to first transform it into a smart object. This is done with sensors, which would collect data, and AI, which analyzes such data. Then, to bring this smart object into the IoT network, you would connect it to the internet, enabling it to communicate with other smart devices. One of the implications of this is that, as more and more physical objects collect data and are connected to the internet, an increasing number of security concerns arise. Cyber security expert, Bruce Schneier, puts is this way: “As everything turns into a computer, computer security becomes everything security”. A practical example of this comes from a writer for the National Law Review, who wrote about her “smart” light bulbs automatically turning on at 3:30 in the morning, and her fear that her smart home system had been hacked. Turns out that it was just a family member playing around with the light settings on their smart hub. But this is still a good reminder to make sure that you have whatever security protocols you can in place, and that you choose a password that’s difficult to guess.
With this basic understanding of what digital twins, AI, and the IoT are, it will be simple to understand how the three of them fit together to transform an object into a smart object.
Here we’ll provide a basic outline of how these three technologies would fit together in the use case of a building’s floor plan. The benefit of using these technologies in this use case is to make the life of a building operations manager easier. A commercial building includes many different “dumb” systems that can be converted into “smart” systems. Once the majority of these systems become “smart”, the building as a whole can be considered a smart building. Examples of these dumb systems include lighting and HVAC.
These sensors can be either wired or wireless, and there are benefits and disadvantages to each of these options. For example, wireless sensors reduce the cost of wiring intrinsically because they don’t require wires, and instead can transmit data via a wireless mesh network. And many simple wireless sensors can have a battery life of roughly 10 years.
The installed sensors collect different data points depending on which type of sensor they are. For example, simply installing an Indoor Environmental Quality (IEQ) sensor in a room in your building will measure various data points related to a room’s air quality (including room temperature and CO2 levels), as well as lighting information such as colour temperature of lights, and brightness. This data can be used to optimize light output automatically. For example, you could implement daylight harvesting to reduce artificial light used based on availability of natural light.
You can construct a digital representation aka a digital twin of almost any physical device. For example, even creating a virtual floor plan of a building would constitute the creation of a digital twin. Digital twins become more comprehensive, however, as they collect more data about the actual physical object.
Sensors allow for the collection of this data in real-time.
Let’s say you want to construct a digital twin for the lighting system of a commercial building. First you would create a virtual floor plan of the building, or even just a floor plan for one floor of the building, then you’d enable sensors to communicate with this virtual floor plan via an internet connection. Once this is set up, it’s a matter of representing information about the lighting system within the floor plan. For example, you could have icons that represent each of the lighting fixtures laid out on the virtual floor plan, and they could change colours depending the light's state in real-time. So the icons could be green if the lights are on, and grey if the lights are off.
As previously discussed, Artificial Intelligence is the technology involved in machine learning and thinking. It may seem like a modern concept but, believe it or not, it was initially conceived in 1950 when Alan Turing asked the question “Can machines think?”. It wasn’t until 1956, however, that the term Artificial Intelligence was coined by John McCarthy.
According to SaS.com, “AI works by combining large amounts of data with fast, iterative processing and intelligent algorithms, allowing the software to learn automatically from patterns or features in the data.” This technology works together with sensor fusion to produce actionable insights based on data collected via sensors.
If you’re unsure what sensor fusion is, that’s okay! Sensor fusion is a method used for combining information coming from several sensors. This makes sensor data more than the sum of its parts by considering information gathered from multiple sensors. In other words, sensor fusion combines sensor data, to produce intelligent, data inclusive insights.
To continue with our use case of a lighting system in commercial buildings, at this point you’d implement AI technology and algorithms to analyze sensor data. These insights could then be displayed on a smart dashboard, like an app, as graphs.
For the purposes of this article, we’ve broken down how AI, IoT and digital twin software work together into steps, but the reality is not so linear. This is especially the case when it comes to where IoT technology fits into this whole system. The purpose of IoT technology is essentially to allow intelligent objects to communicate data or information to each other via the internet or a wireless network.
In other words, while a sensor’s job is to collect data, and AI’s job is to analyze it, the purpose of IoT technology is to transmit these data points and insights to other smart objects.
To relate this back to our use case of lighting systems in commercial buildings, IoT technology would allow lighting to be controlled via a wireless hub, such as an app through an interactive digital twin. It’s with IoT technology that you can, for example, control the power output, brightness or colour temperature of your lights with your phone, or automate them to turn on or off at optimal times of the day (based on availability of natural light).
We’ve created this illustration to more easily demonstrate how the three technologies fit together:
In conclusion, the use cases and benefits of these technologies are only limited by your imagination. Almost any physical object, including even medical pills, can be transformed into digital twins or smart objects. Remember, digital twins are not inherently smart, it’s AI and IoT technology that gives them the power to analyze data about an object, and transmit that information to smart devices. Once they have these capabilities, digital twins transform a physical object into a smart object, making it user-friendly to interact with the physical object through technology, and modify its parameters in real-time. For example, to remotely adjust or automate the lighting in your building, all three technologies would work together to make this possible.
At Cence Power, our mission is to improve the energy efficiency of commercial buildings by providing Direct Current (DC) power distribution networks and transforming building systems, like lighting and HVAC, into smart systems. We create interactive digital twins of buildings so you can easily monitor, control, automate and optimize your energy use based on real-time sensor data.