Pocket-size virtual sensors

Simulation on small edge computers

Monitoring motors and other technological devices requires a constant flow of current data from inside the device. Because it isn’t always possible to install sensors inside, virtual sensors are extremely helpful. Researchers at Siemens Technology have now developed a pocket-size virtual sensor that can easily be attached to a motor, resulting in plants that operate more efficiently and are available longer.

If you want to know what’s going on in a machine, you need sensors inside it that measure temperatures, vibrations, magnetic fields, and rotational frequencies. It’s impossible to monitor motors, pumps, or machinery without sensors that convert physical and chemical variables to electrical signals.


The problem is that in the very location where sensors need to take their measurements, there’s either no room, parts are rotating, or installation takes too much time and effort. The solution is virtual sensors – smart software applications that make it possible to obtain the desired measured variables without physical measuring probes. These sensors use digital twins, meaning simulation models of a machine. The digital twins are fed operational data from which additional variables – such as the winding temperature of a motor, fan, or pump – can be determined on an ongoing basis. Until now, utilizing these sensors was anything but simple because it required knowledge about model reduction and complex mathematical algorithms.

User-friendly edge computers

A new and user-friendly development at Siemens Technology now makes it possible to pack these simulation models in small boxes in edge computers, allowing them to be attached directly to machines and integrated in control systems with very little effort. Operators can use these virtual sensors without having to understand the details of their complicated mathematical inner workings. “They now make it extremely easy to upgrade the digital infrastructure in the factory,” says Christian Wolf Pozzo, Product Manager at Siemens Digital Industries.

The reason digital twins are so versatile is because they contain a generalized motor model that can be used as the basis for rapidly generating individual motor types. It used to be that each motor was digitalized individually in a resource-intensive process, but now it’s a lot easier. Starting with the basic model, customers can simulate almost any new motor simply by selecting the design parameters. “That’s a huge simplification that speeds up development tremendously,” says Vincent Malik, who works in the “Simulation and Digital Twin” group at Siemens Technology.

Simulation in real time

In order to measure temperature in real time, the latest operational data – not just from the converter but also from the powertrain, gear units, and fans – is constantly fed into the individual digital model of the particular motor. This results in a dynamic process model that can then be used to extract physical status variables like temperature in real time, even without an on-site thermometer.


Virtual sensors make it possible, for example, to predict imminent problems, such as an unscheduled motor shutdown. And, even better, they can be used for predictive maintenance and in this way ensure greater operational safety, a longer service life for motors, and reduced repair costs.

Experts' voice

Christian Wolf-Pozzo, Mohamed Khali and Vincent Malik explain their project

First pilot project at a tire manufacturer

This isn’t just theoretical. A pilot project is currently underway in which a tire manufacturer is using pocket-size virtual sensors for an extruder that produces plastic. “We already have a virtual model of the motor of an extruder machine,” says Mohamed Khalil, who also works in the “Simulation and Digital Twin” group. The engineers exploit all the benefits of edge computers with their existing connectors and structures. “These aren’t island solutions. They’re elements that can be combined like Legos and attached to machines.”


With virtual sensors, it’s the first time that the original simulation of a motor is also transformed into an executable digital twin (xDT) that contains not only the model of a motor but also the current process sequence. All operators have to think about is whether a motor is functioning perfectly or there’s a problem. The complex model operating in the background requires no attention

Author: Hubertus Breuer, September 2021

Subscribe to our Newsletter

Stay up to date at all times: everything you need to know about electrification, automation, and digitalization.