The rise of virtual sensors

Earthquake sensors detect seismic waves. They come in many shapes and sizes and work in different ways. Some sensors are placed on the ground but others are attached to buildings or other structures. But what is the alternative when a physical sensor cannot be placed in the chosen position due to spatial conditions?

Photo by Alex Vog

In recent years, sensors have become an integral part of the Internet of Things (IoT). They collect data from the physical world and convert it into digital signals. In 2022, one-third of all sensors shipped were IoT sensors, and applications range from connected homes, smart cities and industrial automation. Within smart manufacturing for example, the IIoT (Industrial Internet of Things) enables factories to connect machinery to an array of sensors and monitors that gather, analyze and communicate data with other devices and systems with the aim of improving quality and consistency. This seamless connectivity throughout the plant allows manufacturers to monitor equipment and system performance. These sensors are often expensive, not only to produce, but also to design and install. That’s where virtual sensors come into play.

Virtual sensors are software-based models of physical sensors that can simulate their behaviour and generate sensor readings without the need for actual physical hardware. They can be used as digital twins to monitor or control a physical sensor, providing cost-effective and scalable solutions for certain applications.

Virtual sensors leverage developments on the artificial intelligence and machine learning front to allow for data-driven approaches to estimate key process parameters. In addition to being less expensive, virtual sensors provide an interesting alternative when a physical sensor cannot be placed in the preferred position due to spatial conditions (e.g. lack of room for a sensor) or a hostile environment (e.g. exposure to acids or extreme temperatures). Virtual sensor technology can reduce signal noise and, thus, increase confidence in the signals when a sensor’s output is confirmed by other sensors measuring the same phenomenon. Finally, virtual sensors are extremely flexible and can be redesigned as required, whereas physical sensors, once installed, often can only be repositioned by mechanical intervention.

Cost is a key advantage

Industry 4.0 is an important driver of virtual sensing technology. The information needed to digitize a factory plant is obtained from many field sensors. If only physical sensors are used for this purpose, the cost of digitizing a factory can be prohibitive for many companies. This cost can be minimized using virtual sensors.

The automotive industry heavily relies on sensing technology for many processes related to safety, entertainment, traffic control, navigation and guidance. As vehicles gain autonomy, this reliance on sensing devices will likely increase. However, physical sensors used in cars can be expensive and, in some situations, unreliable. Virtual sensors are becoming a valuable alternative for car manufacturers. They can provide a redundant safety backup to physical sensors and are fundamental in the development of more advanced driver assistance systems (ADAS) and therefore for the realization of autonomous vehicles.

How standards can help

Virtual sensors are on the way to becoming a fundamental technology for the future of society. Their use is expected to grow at a CAGR (compound annual growth rate) of over 30% over the next five years, and that trend will likely continue for many years to come.

SC 41 is a subcommittee formed by the IEC and ISO to standardize the IoT and digital twin. In a recent interview with e-tech, its Chair, François Coallier mentions “Further down the line, we may be preparing standards for virtual sensors and on the quality of data: how can we ensure the vast amount of information we collect from the various sensors out there is appropriate and meets the right criteria. ISO/IEC JTC 1/SC 42, which prepares standards for artificial intelligence, is looking at these issues, so we could be joining hands on that.”

As machine learning tools are being used to create statistical models, and AI to make automated decisions on how to apply those models in the sensor, it is easy to imagine that virtual sensors are an area of technology that can quickly become a "black box". This not only involves issues of privacy and data ownership, but also of the validity and transparency of the systems that use virtual sensors.

But these risks are, for the time being at least, outweighed by the benefits. According to the IEC White Paper on Virtual Sensors, these will be increasingly used in a world where people, machines, products, equipment and systems co-exist. Humans can easily be exposed to harm by the inappropriate or unexpected behaviour of machines. Public infrastructure monitoring systems, traditional aircraft sensing systems and current earthquake detection systems, etc., can also cause serious injury, or worse, to individuals and society if they malfunction. The use of virtual sensors can augment safety in these areas. Technical fields related to virtual sensors are diverse, with a wide range of affected industrial sectors. The fields may also be relevant to, but are yet to become an extension of, those on which the IEC is currently working.

The IEC  White Paper on Virtual Sensors provides an overview of the technology, some of its common applications and of potential gaps in standards covering this area.