Smart sensors become even smarter - with machine learning and creative technology development

Nationwide mobile coverage for the Internet of Things (IoT) provides brand new opportunities for more secure and more environmentally friendly solutions, and increased profitability in many business sectors.

Sist oppdatert: Feb 20, 2020
Published Nov 20, 2018
Sensor Jeremy Hoved
Machine learning for sensors, based on large amounts of big data, can shed light on phenomena and reveal knowledge that wouldn’t otherwise be possible to access manually”, says senior project developer at NORCE, Jeremy Cook. (Photo: Andreas R. Graven)

Smart sensors are closely linked to digitalisation, which is becoming increasingly important to almost every industry and business sector, and for most people.

At the same time, it is important to deliver solutions that enable Norway to become a low-emission society, to make trade and industry more competitive, and people’s everyday life easier, safer and better.

“We focus on a combination of creative technology development, machine learning and knowledge about artificial intelligence when developing smart sensors. This is only part of our expertise and is a key to our success in the field”, says senior project developer at NORCE, Jeremy Cook.

Development of a range of sensors and early testing of NB-IoT

Recently, NORCE researchers have been responsible for the development of a range of smart sensors, such as the prototype of a sensor that is placed in skis to analyse avalanche risks.

Last winter, a group of volunteer skiers with sensors in their skiscompleted test runs, and thus became a small part of the Internet of Things. A company called Think Outside will continue to work on the ski project, while Jeremy and his NORCE colleagues at Fantoft in Bergen are looking forward to tackling new projects.

Smart sensors could be used as measuring devices that alert farmers about conditions in their cowsheds, or as valve warning systems in the oil and gas sector.

For example, sensors could also be used to control the heating of your house, or alert someone that the contents of a package have been damaged during transportation”, says researcher and senior technology developer Bård Henriksen, who has worked extensively on several of the aforementioned solutions.

Smart sensor with NarrowBand-IoT (thereof SIM card holder) that has inbuilt signal processing and intelligence. Length approx. 40 mm. (Illustration: Ingvar Henne, NORCE)

Due to their long experience and expertise regarding sensor technology, NORCE researchers at Fantoft conducted early testing of NarrowBand IoT, which Telenor and Telia have now incorporated into the nationwide 4G network.

This will ensure a stable connection and completely new accessibility.

Sensors that predict events and recognise patterns

“Smart sensor support from the telephone network will provide coverage for sensors in many more hard-to-reach locations, while the battery capacity of individual sensors is significantly increased. These are two definite improvements, and the fact that sensors will be cheaper to produce in the future, will be a great incentive for us going forward”, says NORCE researcher and senior technology developer Ingvar Henne.

Jeremy Cook says that the sensors of tomorrow will be able to predict events more accurately, and based on machine learning, they will also be able to recognize critically important patterns.

“Machine learning for sensors, based on large amounts of big data, can shed light on phenomena and reveal knowledge that wouldn’t otherwise be possible to access manually”, says Cook.

“We are getting new opportunities to unearth new and critical information, both from our own data sources and in combination with other databases”, he continues.

Researcher and senior technology developer Bård Henriksen (left), has worked extensively on developing smart sensors, such as valve warning systems in the oil and gas sector. Together with senior technology developer Ingvar Henne, who also has extensive experience in the development of smart sensors. (Photo: Andreas R. Graven)

Real-time analytics in combination with historical data

Real-time analysis of data, in combination with historical data and other available information, also enables sensors to help optimise industrial manufacturing and reduce the risk of accidents.

NORCE researchers wish to continue being one step ahead in order to find the best solutions within the fields of sensor technology and measurement systems.

“We possess expertise regarding various measurement systems for industrial applications and can assist with the design of automated solutions and decision support. And these are skills that are absolutely necessary in order to develop sensors that make the right decisions and alert at the right time”, says Kari Marvik, manager of research and technology.

Now that IoT technology is incorporated in the mobile network,the potential is enormous regarding the use of smart sensors in many business sectors. But what does it actually mean, now that the mobile network is optimised for the Internet of Things?

“It means, for example, that smart sensors which provide useful functions for many of these things, receive a larger area of application due to coverage and roaming. In addition, they will be more reliable than they are today”, says Cook.

The new network will enable the sensors to communicate better with each other, not just alerting individually, but alerting each other.

Therefore, such sensors can be placed virtually anywhere, which was not possible before due to insufficient coverage.

Vibrations, noise analysis and gauging pressure - and much more

Smart wireless sensors can be installed in both existing manufacturing environments and in areas without additional infrastructure, as long as there is IoT coverage provided by the mobile network

“Efficient transfer of data signals from sensors leads to a battery life of many years, and customised data models for measurements can provide high performance when required – and still have a low energy consumption”, says Jeremy Cook.

Basically, correct information is essential in order to make good decisions, and both people and machines will increasingly need smart sensors to monitor processes and environments.

“The potential for future development is enormous. Modern sensors can be taught to recognize patterns, to communicate with each other, to process data and to send information in more advanced ways. For example, this might include vibrations, noise analysis and gauging pressure”, says Cook.