The value of edge-hosted AI in relation to increasing the efficiency and resilience of process control and automation operations, especially at remote sites or where cloud connectivity may be compromised.

Edge AI insights
- Edge AI reduces latency and bandwidth needs, ensuring secure, real-time data processing and analysis directly at industrial production sites.
- AI in industrial plants enhances efficiency and safety, decreases human intervention, and increases uptime by spotting anomalies and optimizing processes.
Artificial intelligence (AI) and machine learning (ML) algorithms trained on cloud-hosted datasets can be executed locally on computing devices at the edge of the network to give industrial plants the benefit of secure real-time processing and analysis of data at the point of production and consumption.
There are few industries that remain untouched by the transformative power of AI as a central element in their digitalization strategies. And, from a process control and automation perspective, AI offers huge potential to support the analysis and management of vast data sets produced by thousands of connected devices, systems and processes in a modern industrial facility.
Applied in a typical industrial environment, AI can help optimize the efficiency, reliability and safety of control processes. It can help reduce the need for human intervention in laborious or routine tasks, and ultimately it can contribute to increased plant uptime while reducing operating costs. In realizing these objectives, AI also paves the way towards the longer-term goal of fully automated plant operation.
AI algorithms, fed by years’ worth of real-world operations data, can be trained using ML to spot trends and anomalies that no human engineer would normally notice. These insights can be harnessed to warn of the imminent failure of a sensor, or to suggest how a particular process can be fine-tuned to make it more energy efficient.
Effective decision-making hinges on ensuring timely access to accurate, relevant data, together with the ability to rapidly analyze and interpret that information. And in the context of a process control environment, it is this imperative to ‘do more with data’ that is turning the spotlight on the use of AI where it is most valuable – at the same location where operations data is produced and consumed.
Edge AI benefits
Edge AI describes the application of AI to execute tasks in real-time – or close to it – on a connected device. Data informing the AI engine’s decision-making process may be hosted remotely in the cloud. Equally it may reside in close proximity to the device itself, at the figurative edge of the network.
While edge computing’s origins date back to the turn of the millennium, deployment of AI models at the network’s edge is a comparatively recent phenomenon. Advances in CPU power and circuit miniaturization give compact hardware appliances number-crunching capabilities that would have been considered supercomputer-like a decade ago.
High-speed 5G connectivity also allows data to be harvested from thousands or millions of Internet of Things (IoT) devices, feeding increasingly sophisticated AI/ML models hosted in the cloud or at a remote data center. And as those models are trained with more devices and more data, they become iteratively smarter, more accurate and more reliable.
Shifting AI computing power to the network’s edge has several benefits. It significantly reduces the bandwidth requirements and associated costs of transporting huge amounts of data between field-based devices and the cloud. Executing applications locally, rather than at a distant location, it also slashes system latency — the round-trip time between data’s point of origin and where it’s processed. Whether you are sitting in a driverless car or performing robotic surgery on patients in another continent, even a hundred ms delay between system input and output could have potentially catastrophic consequences.
Co-locating computing resources where data is produced and consumed mitigates potential cybersecurity risks of connecting sites to a remote data center over the internet. While private and public clouds provide high levels of intrinsic security, the ability to keep commercially sensitive data on premises at all times, where it is not subject to unauthorized exfiltration and scrutiny, gives assurance of ultimate ownership and control over data.
Mission critical use cases
Edge computing also appeals in mission-critical use cases because it is not reliant on continuous Internet connectivity to process data. This ensures high availability of applications that might otherwise be affected by network outages, or at remote sites where connectivity is poor or sporadic.
Given these benefits, it is not surprising that edge computing is having a transformative role in many industrial process control and automation environments. Acting as an intelligent bridge between field devices and the cloud, it allows asset owners to harness increasing volumes of data gathered from IIoT devices and control systems that are used in a wide range of production processes.
Every plant operator is faced with the challenges of optimizing plant uptime, efficiency, safety, sustainability and profitability. Key to this is the timely extraction of actionable insights from the huge volume of operations, IT and engineering technology data produced by thousands of sensors, sub-systems and other sources. This can be particularly difficult in remote or inaccessible environments like mines, chemical processing sites, offshore windfarms or oil platforms, far from mobile broadband coverage. From a business perspective, AI-powered analytics — enabled by edge and cloud computing — can help unlock tangible value from an estimated 80% of production data that is currently unused by industrial plant owners.
By definition, edge and cloud paradigms are fundamentally different. However, they are widely seen as complementary technologies, combining the immediacy, security and resilience of edge computing with the effectively limitless scale and storage capacity of the cloud.
An example of this is a process optimization model for a chemical processing plant. Fed by cloud-hosted production data collected at scale from IIoT devices in the field, this model can then be executed in real time in the field using Edge AI. The model’s outputs then ensure that devices and systems can respond more quickly and accurately to the demands of their environment.
Edge AI is a powerful tool
Edge AI can also be used as a powerful tool to support other use cases for industrial plant owners, such as condition-based asset health and performance monitoring. Instrumentation data, collected from sensors, actuators and other devices at the network’s edge, can be used as a basis for condition-based monitoring. Trained AI algorithms can spot potential anomalies in the data signature of a connected device or subsystem, cross-referencing behaviors against the histories of tens of thousands of other similar devices. By sending automated notifications to plant personnel, it can give advanced warning of potential system failures that could otherwise affect plant or process performance and result in costly unscheduled downtime.
It is widely anticipated that edge computing will have a transformative effect across a wide spectrum of industrial automation applications. Combined with IIoT technology and AI-powered analytics, edge can complement cloud-hosted storage and applications to realize the full value of production data in a range of process industries.
Vikas Maurya is global product line manager, edge, cloud and IoT at ABB. This originally appeared on Control Engineering Europe. Edited by Chris Vavra, senior editor, Control Engineering, WTWH Media, [email protected].
MORE INNOVATIONS
Keywords: Edge AI, artificial intelligence
CONSIDER THIS
What ways can edge AI help your facility improve its operations?
ONLINE
Read a longer version of the article at www.controleng.com.