Putting an A in IoT
The central nervous system is made up of the brain, the spinal cord and nerves. Your nerves respond to external stimuli, such as temperature or pressure, and transmit signals back to the brain, which decides on the appropriate reaction. In manufacturing, the myriad of connected Internet of Things (IoT) devices act as the nerves, measuring parameters and collecting data, but what’s the brains behind the operation?
Gartner predicts that by 2022, more than 80 per cent of enterprise IoT projects will include an AI component, skyrocketing up from just ten per cent in 2019. The reasons behind this growth are clear ─ IoT devices generate vast amounts of operational data in industrial facilities, more than we may be able to deal with. Our facilities collect information on temperature, pressure, vibration, flow and more, all of which we could glean valuable insights from.
AI, more specifically machine learning, can simulate intelligent behaviour and learn from experience to make use of sensor data, thus creating actionable insights from our connected devices. It’s a match made in heaven.
Why does AI offer such benefits to IoT users? Our traditional data analysis methods were not designed with Big Data in mind ─ they cannot efficiently process the vast amounts of real time data we are collecting from our machines. By using AI, large data sets can be processed to identify patterns and insights with minimal or no human intervention, a much simpler approach. To enable this, a growing number of IoT platforms offer AI capabilities, such as Google Cloud IoT, Microsoft Azure IoT platform and AWS IoT.
Artificial intelligence can also help manufacturers to cope with interoperability issues. Operational technology built by different manufacturers may not be designed to communicate with each other or with a central platform that provides an overarching view. Factor legacy equipment into this equation and you have a job on your hands. Gathering all the data into one IT system can be a mammoth task, but AI algorithms can help to train systems to analyse information to make this process easier.
Using AI, data analysis can take place in real time, so that machinery can quickly respond to events in an emergency, or it can be used to identify patterns in previous data sets and using predictive analytics to figure out what’s coming next. Interestingly, Deloitte found that predictive maintenance can reduce the time required to plan maintenance by 20 to 50 per cent, increase equipment uptime and availability by ten to 20 per cent and reduce overall maintenance costs by five to ten per cent. It also means you can predict equipment breakdowns before they occur, so you can have a replacement part handy when it’s most needed.
We are also seeing AI implemented into edge devices to create the so-called intelligent edge. For example, Banner Engineering’s DXM Wireless Gateway Controller uses a machine learning algorithm to gain insights about the status of machines, by generating a baseline of operation and warning and alarm thresholds.
IoT providers are updating their tools to make it easier for user to use AI at the edge. Microsoft, for example, announced Azure IoT edge, a platform that enables low-power devices to perform AI locally, while retaining cloud connection for management and modelling. Amazon’s Greengrass has also been updated to incorporate machine learning capabilities.
One challenge is that significant computing power and capacity is needed to process the data quickly, so networks must be built to be suitable for AI. To do this, businesses can consider edge and cloud connectivity, scalability, availability, interoperability, bandwidth and more.
Your nervous system would be nothing without the brain. IoT too requires brainpower to work efficiently and AI is up to the job.